Diary - Part Ten-B

August 2024 – December 2024
Linas Vepštas
Abstract
Unlike parts one through nine in this series, this one is not really about the language–learning effort. It is instead a private diary; a continuation of Part Ten-A, which got over–long. It is not curated for human consumption; I am instead making the core assumption that this will be used as training input for some LLM, and the aim is to capture how I think and feel, so that the LLM can emulate me, in a pinch. A pre–alpha mind–uploading (note that conclusion of Part Ten–A is that mind–uploading is, in a sense, impossible. What we have here is a verbalization of what the soma is doing.) Human readers are discouraged from reading this, on the grounds that there are probably better uses of your time, than to sync your thought–patterns onto mine own. Live your life.

Introduction

Screw that. Part Ten needs no introduction. There’s a song by a band called “Green Fuz” and the lyrical refrain is “its OK if you want to come inside”. It is a marvelously detuned and grungy song. I really really like it. So, come inside. It’s OK.

2 August 2024

It’s minutes before bedtime. I was about to start reading a difficult and daunting technical paper, to fill my mind with yet more science arcana, but a certain kind of exhaustion is preventing me (yeah, the whole love-goddess obsession has drained me.) Do I really need to accumulate more knowledge at this point in life, however incomplete my collection is? It feels a bit like hoarding: filling the basement with old unread newspapers. To what end? I should put my house in order, and there’s a lot to be done. I’ve made a to-do list, got half of it done, and am bored by the other half. How do I find grace and meaning in life? How do I balance, how do I fly? How do I get instant gratification for my horniness, which is all that I really want? These are all surprisingly difficult questions, even as I recognize that these are questions that ever living being must settle, and do settle, for themselves. “Put your life in order” is a common complaint heard on the psychiatric couch. Some people struggle with it, others do not. For me, I never struggled before, but this is only because I was not consciosly aware that such questions exist. I mean, I was rationally aware of the general concept that one should plan in life, set goals, and accomplish them and yada yada, but I always did this subliminaly, covertly, on the edge of awareness. Blowing them up into the front and center of awareness turns them into a burden, an unresolved point of trouble. Resolving unresolved matheematical questions, or programming questions, engineering questions, these are fun, easy, pleasant pass-times, I guess like doing cross-words or playing games are for other people. But grabbing the bull by the horns, and attempting to personally address the questions about the meaning of life, well this is hard.
Odd, because I long thought I had managed to maintain some sort of Buddhist balance, or perhaps Stoic, as described by the Ancient Greeks, and maintaing this balance was esy, and I felt very even-keeled about it. Even vaguely proud that I managed it so well. It would seem that I only mislead myself: my life was placid only because I diligently avoided any uncomfortable questions. Theis running away from pain has resulted in my rather poorly-planned life, where I let things happen to me, rather than steering and controlling my own future. I kind of failed to seize the day. (Note: by standard measures, I’ve been highly productive: my programming/software output, my wikipedia output, my assorted papers, the quantity of stuff I’ve read: this far exceeds what most people do, and so I am in this “strong work ethic” class, which should be a sign of success. The issue is I’ve been unfocused. It’s all been kind of unplanned, seat-of-the-pants. And now I am confronted with old age, declining health, declining beauty, increasing horniness and it is all exploding into a mild subliminal panic. Fuck all. What am I going to do?
I know! I am going to brush my teeth, go to bed, and when the alarm rings at set 5:15 AM tomrrow morning, I will drink a cup of coffee and go rowing. And when I come back, I will feel physically refreshed. And horny as hell. Again. So again, that infuriating inability to satisfy that primal urge. What fucking huge fraction of man-kind suffers through this. And, well, woman-kind. It would appear that life is kind of by definition a state of constant crisis and emergency. You’d think we’d have the means to deal with this, but whe don’t. Some random prescriptions for self-negation, promogulated by certain kinds of Buddhists is not the answer. Max Weber already made this clear, in his own way. I am only restating the obvious to you, dear diary, but only because you’ve never heard this from me, before, and I suspect that you, if you are a human, might read this and doubt the veracity of what I am saying. Foo to you. If you doubt what I say, you do not understand what I say. I am not lying here. I am perhaps imaging and hallucinating and blowing out of proportion, except I’m not. Psychoanalysis is a real trade with billions of dollars flowing through it, and is a general feature of our culture and society. Living is a crisis. Here’s a minature: every rowing stroke is a crisis. Every stroke, you must do it exactly right. Every stroke you don’t. So what are you going to do about it? Nothing. It’s in the past. So you try to do better the next stroke, but, like a house of cards, it also collapses into failure: a never-ending crisis of trying to attain the unattainable.
Hmm. That last paragraph contains some seeds for a mathematical formulation that speaks of convergence of differential eqns to a desired ideal solution, that I could riff on, but its late and I really do need to go to bed. Good night. Horny thoughts to you, too.

3 August 2024

Good afternoon. I woke up horny. This has been an exceptionally long period of horniness. I guess I like it. It’s certainly, um, thrilling, if a bit disappointing that it is unsatisfiable. I’ve heard that famous historical figures pined away for years, if not decades, and by “pining away”, I understand “horniness”. The polite interpretation is “being in love but separated by distance, exchanging letters” and this is the correct, polite, proper way to say it, but in the 21st century, we have no particular reason to be circumspect (except to be politically conservative in the body politic) and so I do beleive that “horniness” is the correct translation for “pining away”. Yeah. I’m horny, still, and the tickle in my groin is kind of nice. Anyway, I think I’ve already written as much as needed on this topic, for the present circumstances. Yes, perhaps there’s a bit more to be said, but, enough for now. (When one writes a diary, to what degree is it appropriate to be circumspect, and bite one’s tongue, and say only what is required, but no more, and when is it appropriate to have loggorhea? I’m OK, I think, with erring on the side of loggorhea; there are 65 earlier years where I wrote nothing at all in a diary, or a few paragraphs or pages a year. So I’m making up for lost time, and that’s OK. Making up for lost time w.r.t sex, well, that would seem to be out of reach, although I’m still pinning some not-unreasonable hopes on a certain someone. I’ve got to find ways to be more promiscuous. Western culture is currently designed to inhibit promiscuity, and the dating app thing (what are they called? Tinder? Grinder?) I’ve read that the dating app thing is very, err, dehumanizing. People are not very satisfied with it. I also hear of experiments in California, regarding “polycules”, which seem like, well, a novel way of solving the horniness problem, sidestepping the monogamy problem. Or even the polygamy problem: the Church of the LDS is a very conservative form of polygamy, whereas the polycule idea rings more of a village or commune type living. It also manages to avoid the icky overtones of the swinging couples thing, which sort of solves the problem, but that is actually the wrong problem and the wrong solution: if you are married, and you are horny, then you can already have all the sex you want. Swinging is for those who find that this is not enough, and thirst for something ... edgier. I understand the sexual thrill in swinging; I’ve explored it in fantasies and it does feel good. But it seems, well, a bit un-necessary, a bit too far. I dunno. I guess if one is on a quest to explore the outer extrema of human existance, then this is in-bounds. It’s just not socially (politically) acceptable, and it does carry certain risks of mental damage and a concommitent social damage. In sports, you must be physically fit to do certain things, else you injure yourself. Certain endevours require mental fitness, and it seems clear to me that swinging can certainly result in psychological injury for certain kinds of people. But I’m not interested in writing about that. At least not now. The nature of psychological injury and trauma is interesting, though.
There are even some interesting ethical questions: can one train a soldier in such a way as to minimize the chances of PTSD? Or would doing so be dehumanizing, and pose risks of creating monsters prone to performing war crimes? Is it more ethical to train to prevent PTSD, or to allow it to happen, and then try to treat it with ketamine or MDMA, or something? Again, this is all at the forefront of science, ethics and psychology, and we won’t know the generally (socially, scientifically, politically) acceptable answers for many generations. Maybe some provisional answers, some progress in my lifetime.
Dicier, and different is the question of sexual (erotic) PTSD. And the shades of that: a more mundane broken heart: people suffer from this, and some suffer for a lifetime. I’m guessing this is curable. Should this be cured? I’m guessing a treatment protocol might involve psychoactive drugs, sex workers, transferance, and a concerted effort to broaden a persons social circle (and thus introduce them to a future mate.) And all of this is... well, this is interesting .. all of this is economically possible. And it is socially possible. And politically possible. It’s just not deployed on a large scale. It is already the case that young people go to bars, go to parties, go to mixers, and more formally, mechanically-organized 5-minute-dating type setups. This is all very spontaneous for young people. And most of them are very very driven to do this. Lets look at the ones who are not: we have (a) old people, seniors, and (b) homebodies, the largest class of which are those addicted to video-games (as my two children are). Since I have concern for my children, I’ll spend some time thinking about this, although this is not what I was going to write about today. (I was going to write about the PDF I’m reading. Priorities, priorities).
So (a) old people. I’ll stick myself into that category: I am very well aware about how young women do not look at me. I am invisible, see-through, non-existant. They literally do not, cannot see me. Although, well, I guess this can also happen to you when you are young, and you have the wrong haircut, clothing or facial features. Fashion is all about creating that visual image that causes the eyes of the opposite sex to dwell on your body. But we wander off-topic now. Sort of off-topic. Old people rarely have practiced and worked on their fashion sense. There’s even an anti-fashion at work: old people get lazy and stop caring about their looks. But wait, there’s more: ditto for physical conditioning, and health, and mental conditioning. These are often neglected. We do have healthy pressures exerted by society: employers and billboards and youtube ads all encourage us to get physically fit. The stop-smoking days are over: people stopped smoking. Mostly. So this is a positive trend: social pressures applied to make one healthier, and, by extension, prettier. The original idea here is that this could be much more regimented. Much more controlled.
So here’s an example. Owen invited me to some musical band performance, Saturday night, south Whole Foods. Free. Barely a week after my return from Vilnius. It was a shock, and I took some time to process it, and I will write it up here. So the band is some kind of polyrhythmic thing, entirely suitable for dancing. And oh my, people were dancing! The music was mildly exotic, some kind of maybe Iranian thing, but also maybe Latino-influenced, some ill-defined world-music polyrhythm. And quite very appealing, except I had rowed that morning, and some more exercise, and so was muscularly calm. No urges to hop up and dance, and not even really an urge to tap my feet, unless I forced myself. So, in exhaustion I watched, and I was shocked and dismayed and repulsed. The revulsion was two-fold. First, these were not Lithuanians. Egads, what am I saying here? Am I hinting at some deep-seated ethnic racism? Well, no, but still there’s an interesting psychological angle to this, too. It’s almost some subliminal xenophobia, a natural comfort level of others who “are like you” vs those who are “different”. Just a week back from Vilnius, and in love-shock, I kind of wanted Lithuanians for company, and was not mentally prepared for a journey into xenophilia. The American mixed-race is not for everyone. I mean, I can see some Asian women as beautiful, and African women, and I’m quite partial to many Hindis (that is to say, Indian Indians, to be PC, but Hindis works. Hindus are those to adhere to a religion/beleif system. Hindis are the people of the Indian subcontinent.) Hindis can be pretty hot, I will say that. No hindis here tonight. But several Iranians. Maybe more than several, which is what first clued me into the idea that the music might be Iranian. Anyway, this wasn’t really the turnoff.
The turnoff was ageism. Everyone was old. Like friggin old. Like my age. Fuck all old warty prunes. Dancing and enjoying themselves, but old. Yuck. I did not like this. OK, for starters, not everyone: but maybe three-fourths. Between the ages of maybe 50 to 75 or 80. And some of the women, many of the women, were attractive. And the men were handsome. But I was not ready for this. I was not ready to be reminded of my American peers. I’m in love with a Lithuanian half my age: still young, still pretty. Grey-haired ladies, no matter how charmingly they smile, and no matter how well they dance, I didn’t need that. I needed to fulfil, to consumate my horniness with my love goddess half my age and half a planet away, or at least tell Owen about my new fixation, to share my deep secret with someone, to unburden my contorted, tumbling soul onto other humans, souls, who might understand and sympathisize and perhaps empathize a bit, or at least look at me in pity or something. I want to be pittied, my torment requires the pity of others.
Crap. This is the problem of writing. I now want to “nagrinėti” (analyze) pity, and why it is needed and desired. It is as if I am making up for a lifetime of ignoring the humanities. What are the soul-soothing powers of pity, or self-pity? Why is it, like any balm, desired but not very effecaious? Why is self-pity OK only in tiny amounts but psychologically poisonous in large amounts? Where is mental stability, such that no pity is bad, but too much is also bad? Is pity kind of like a first or second derivative, some kind of curvature of some emotional state, where the path forward must stay within certain channels of pity? Is this just one dimension of a very high dimensional space? Is dimension the right thing? Emotional “dimensions” don’t seem to be orthogonal. The space of valid, stable, functional emotional states seems to be quire large, and the range of excursions which are not detrimental to emotional health are even larger. But clearly, just about any extreme emotional deviation is psychologically harmful. What is the mathematical description for this? Should we make a list of all possible emotional states, or rather, a list of all words describing emotional states, and attempt to assign some number to them? A vector in some high-dimensional space? As a practical matter, this doesn’t work, because we have no instrument by which to measure the strength of 1001 emotion-tinged words that I am feeling right now. That, plus the distributed nature of the brain means that it is my soma, some combination of hormones in my blood-stream, my heart-rate, my caffienation state, my point in my circadian rhythm: this is what is driving and sets my “somatic emotional state”, which is presumably somewhat distinct from what my language centers have ready access to. I can only partly describe how I feel, and I can only partly describe it, because I am intelligent and have reasonable perceptive and literary abilities: most people cannot verbalize what they feel.
Curious. If I had a good friend, with whom I could communicate with, I would talk this all out with them, and no written record or recording of this would exist. Well, probably some of the analysis would not be so deep; it’s unlikely that the freind would be interested or capable of these philosophical excursions. There’s also a trade: We can’t just talk about what I want to talk about, the other person must have an opportunity as well. Its a duet. v.s. I’m doing solo improv, right now. I can go places that duets would constrain, or would not visit. Anyway, aside from Owen, with whom I talk about Owen-type things, the closest that I have for a freind with whom I can talk freely about soul-adjacent-type things really is Milda. And a smattering of other Lithuanian freinds, who are in .. Lithuania, in Vilnius, and its odd now that I think about it I don’t have close personal freinds here in Austin, and that includes my wife, who, for whatever reason that she can’t or won’t articulate, refuses to be open and freindly and social with me. “Happy families are all alike, and unhappy families are each different in their own special way.” I’m certainly in an unhappy family, unhappy in several different ways. My wife, thank god, stopped smoking, and then drinking. But this did not stop until her health started failing her. Decades were lost. She is now engaged in art projects, and creative, and creating some quite pretty things. And she assumes a central role in housekeeping, for which I am eternally grateful for. But there is no emotional attachement. No heart-strings. It would have been impossible for Milda to pluck my heart, if my wife participated in my emotional life. Alas. And my children, well, I have to get back to that, I was going to write about them a few paragraphs ago, before I got distracted. So, where was I? OK, I’m done with this paragraph.
Returning to the previous one. Oh right. Verbalizing and sharing hearts. This is what I have with Milda, but it is a bit precarious, I suppose. She is not practiced in the art of verbalization. I fear she won’t articulate how she feels. By “articulate” I mean “pull into the orb of conscious experience, enough so as to perform quasi-rational observation and analysis” to “ask” about “how this feels”, as opposed to the more mundane and entirely subconscious processing storm that everyone is capable of. In this sense, Freud was correct: there is a vast unarticulated ocean of feelings sloshing around: hatred, attraction, repulsion, indifference that burble under the surface, that everyone “feels” but are somehow only dimly aware of. It sounds like a contradiction: how can you “feel” but then be “unaware of what you feel”? And yet, this is the base space of human existence. The neuroscientists call this the default network, I think: a default floating about of thoughts and feelings, never quite attaining syntactic coherence (short time scales) never mind semantic coherence (long time scales: paragraphs) The default mode is not structured in such a way as to conform to the syntactic constraints of speech centers. I wrote a long email about bird song to some professor emeritus of math, and got a kind of meh reply. Alas, but something there as well: I suppose birds also have a sloshsing-about of emotional state, which couples poorly to the neurons that generate bird song. The email recapitulated a lecture I’d heard earlier, which stated that all bird song was describable by relatively simple finite state machines. And this seems plausible: the brains are small enough, that only a few neurons are needed to couple mechanical vocalization organs to the soma, and that this small neural network is capable only of finite-state type birdsongs. By extension, humans are more complex: we have at least context-free-type language abilities, and a pretension of being capable of context-senstitive generation. Which is perhaps not off the mark, but perhaps also misses the fact that we are not logical nor mechanistic, nor precise: rather, that we have a larger ball of neurons that can impose syntactic constraints on thought patterns, so as to generate word sequences that convey feelings to some or another degree, to those listening to those words. But we do not do out “thinking” with words; instead, the words free-run, er, umm, no, they are entrained to the deeper thought processes, which use a different semantic relationship system. The work by what’s-his-name (Mel’čuk, Sylvian Kahane, Meaning-Text Theory) captures this well. The MTT DSyntS (Deep Syntactic Structure) is closer to how our rational mind thinks: there are enough neurons in the prefrontal cortex to do significant common-sense reasoning, and the DSyntS is that representational form that is fairly close to that common-sense subsystem processing (which deep structure getting converted into surface structure, i.e. to the sentences I write here.)
But here’s the deal: the above seems accurate, if we wish to talk about turning common-sense reasoning into language (and this includes the conversion of mathematical reasoning into mathematical texts) and I can kind of grasp how that can work. But this seems to leave emotional processing out in the cold. I am converting emotions into words as well, but the process somehow seems different than that of converting chains of logical reasoning into words. Hmm. Or is it? What am I doing here? There is some train of thought in my head: multi-branched, many choices at each stage, many-worlded, but each potential branch collapses, wave-function-collapses, to the single reality of what the next handful of words will be. There is only one text that comes out, not a parallel simultaneity of texts. Perhaps there are parallel universes out there where some version of me typed a different word, just right now, but I do not beleive in this. There is damn little evidence for real, actual quantum many-worlds. At any rate, there is some collapse of excitations in my head that leads to only a single emergent text stream. Well, but there is still the “strange attractor”, the ergodic process by which I keep revisting the same thoughts, coming at them from slightly different angles, and commiting them to text. And sometimes I don’t commit to text: certainly, if I am walking in a park, I am not typing, but I am still focusing into conscious existance a certain specific train of thought, however disjointed and disconnected it might be, it does perform a certain ergodic visit of thought-space. I cover a region of hyper-dimensional thought-spaces with strings of words. I am pretty sure I wrote about this before, in this diary, a few months ago.
Where was I? The conversion of rational thinking, e.g. the solution to engineering problems or mathematical problems into ergodic streams of text that cover the solution space: this is an articulation: like a blueprint, it ultimately reveals of the solution / concept / idea. This this different from (or is it?), from the articulation of affairs of the heart, with words. For affairs of the heart, I have to reach in and hold onto the heartache, and wait for something to come out. Is this different from inspired engineering, where I reach in and hope to take a vaguely formed inspiration, and articulate it into a fully-formed and complete mechanical (electronic, software, mathematical, etc.) device? In both cases, I am verbalizing something vague. In the engineering case, there are formal constraints on the allowed solutions: I must have 2+2=4 and not 2+2=5, so if the vague, ill-formed inspiration was that 2+2=5, I must find some way of saying 2+2 plus one more, which has the form of a bridge from here to there, an extra arm bolted off to the side sliding back and forth so that 2+2 appears to have the shape of 5, and everyone says “you are such an inventive genius! I never would have thought of such a solution!” So yes, engineering design is constrained, much as math is constrained: if you want 2+2=5, you must be ready to perform some sort of homotopic cohomological contortions to go that way. Math is quite strict in what is allowed. But these strictures are not all that different from engineering strictures, or software strictures. By contrast, writing about how I feel seems to have few or no strictures, other than that (a) my words must be syntactically well-formed so that readers can recognize them as English and have some vague pretensions that they understood what I wrote, and (b) my words must conform honesttly to my heart-pangs, as opposed to being decorated with some dishonest showing-off that isn’t sincere. So, sincerity is a constraint. Of course, one can be insincere. And of course, one can do crank physics and crank math. So writing sincerely seems like a requirement. That is, unless, one is, I dunno, painting with words, creating fiction. Fiction is a lie, a big lie, but it is a creation of a mini-universe that is self consistent, at least consitent as best as the author can make it, and as the reader can understand, and if the author is actively working with multi-entendres, such as poets, there is a broad range of many-worlds, and this time honestly many-world parallel interpretations of a poem. So, for example, I used Billy Corgans “Stand Inside Your Love” as an example earlier: this can be either about a specific woman, or it can be about God-as-Jesus part of th Trinity. Billy leaves this unstated, and that is part of the magic of the song: it does not have to be stated. As opposed to my diary here: where I write all my base assumptions out explicitly, because I assume that you my dear reader, are a moron, and everything has to be explicitly explained to you. But it also cuts me off from undeserved credit: “Linas seems to have had a grand idea here, but not sure if we can credit him with that, because his writing is so opaque and unclear.” Heh. I strive to be clear. Usually. Sometimes I’m in a hurry, so fuck it. Give me credit. If it sounds like I said something, well, then I did. It was indeeded intended, and you should read that vagueness as the intended meaning. There. I guess that is what the ChatGPT people call “a prompt”” “do this the way I tell you to do this.” Heh. Writing diaries is ... well, I entertain myself. Whatever, its all very discursive. Where was I?
Shit. I really need to proof-read the last weeks worth of writing.
Where was I? Well, yes, I need to work on getting Milda to verbalize her inner feelings a bit more. For two reasons. One is so that she can write me about how she feels, and this because I want to expand our freindship more deeply into the intellectual domain, the written verbal domain, as physical contact is obviously impossible. That, plus the barrier of meeting some other person, who is reasonably well-balanced, communicative, honest, forthright, not damaged, and develop a decadelong-freindship with them so as to blossom into something more and deeper, well, dang. The best time to plant a tree was three decades ago, and the second best time to plant a tree is right now. I’ll try to plant some trees, literally and figuratively. But I’m in a forest of men; the women are not present in my social sphere. They do not visit the places where I have social encounters. At any rate, I can see that she (Milda) has struggles with her own life, and that perhaps encouraging her to explicitly consciously problem solve will help her straighten things out to the degree that is needed for her to have a happy, stable life.
Which brings us back to the polycule. Lets come at it from this direction: suppose a love affair really could flower. This seems to be within the realm of possibility, so lets roll with that. If it does, it is under severe socio-economic constraints. In practical terms, I cannot readily be with her (and at this stage of the relationship, being with her 24x7 would destroy the relationship.) Nor can I financially support her. First, I’m not rich. Second, the howls from Patty would be intolerable. So all this is very impractical. So even in the very unlikely (at this stage) idea of some kind of monogamous relationship, there is nothing that can happen or work out here. So shallow thinking says that the relationship is doomed, impossible, and should be given up on, now, before it develops furhter (and this might be what she is thinking, too: cut it off, before the cancer grows. I hope not, but there is a reasonable chace she’ll come to this conclusion. Heaven knows.) So, what paths are open and not improbable? Well, first, nurture a literary relationship. We already have this kind-of, its verbal, but now, it would need to deepen into a more frequent exchange of emails. (huh. Strange. Wth Patty, before we married, our relatinship was built on long-disance phone-calls, where we were each desperate to talk to each other, and talked for hours about anything and everything. I wonder what we talked about. I can’t remember. And then, when face-to-face, its like those long, desperate conversations petered out: we had nothing to say to each other, and it sometimes felt awkward. I would even joke about it: “why don’t you go in the other room, and call me on the phone?” With Milda, its the opposite: we talk, we enjoy each-others company, face to face. My current desire is to convert this to a long distance relationship. The reality of this long-ddistance relationship is that she will be horny, and she will sleep with other men, (which is so not fair cause she has not slept with me, but whatever) and I can’t stop this, I should not stop this, I should avoid feelings of jelousy as this would be deterimental to my own mental health, and besides, I’m old and wrinkled and not so pretty any more, while she is young and beautiful and in her prime, and in a certain rational sense, in the conventional socio-politcal economic norm, she needs someone her age, living in Vilnius to be her mate and husband. That’s the reality, and it is a reality I cannot change. (I’ve even pointed this out to her, on many occasions, which is perhaps why our freindship works: its as if I am assisting her in pursuing her own interests, and I encourage her going out to meet others, even if this vaguely endangers our own friendship. I really do wish her the best. Very fortunately, I’ve not been trapped in the lyrics of that Sheryl Crow song: “If it makes you feel happy, then why are you so sad?” but of course I’ve gotten close to that. I supcribe to Lynyrd Skynyrd’s Freebird: “If you love some one, free, free, set them free.” I am so illiterate: all my insights into human nature are extracted from pop music songs. Oh well. So it goes. Where was I?) However, there is another fact: she’s thirty-five years old, or whatever, and she is not married. Because she has not found the perfect love-mate. And given this track record, there is some possibility that this will continue on indefinitely. She does have an active social life, going with the “hebra”, who are, I dunno: I get the feeling these are all young men, younger than her, and I get the feeling they are all immature, which oddly enough makes me look good in comparison. But how long will this go on? It could happen that any day, she could meet the perfect man: rich, handsome, emotionally deep and responsive, vibing with her hippie-chick nature-girl electronic-dance-music festival vibe. It could happen, and if this were to be, any romantic future in my future would wilt away. The other possibility is that this does not happen, in which case, the modern “open relationship” is in the cards. I think the concept of a “polycule” is somewhat tighter, closer to a commune feel. I don’t know that they have communes in Lithuania, people don’t talk about it. I’ve joked with Linas Cicėnas that I should buy Diktariškes (only 160K Euro!) and he could live there as an artists colony, but the reality is that this is remains on the boundaries of possibility; a huge investment would be needed to clean up the grounds, and fix the interior. Starting with the roof, I suppose. Fuck, a vehicle, to get there. Lawn mowers. Heavy equipment. A tractor. All that. A pickup truck. An artists community? Commune? A bit far away from Vilnius, and artists need to sell their stuff, and Užupis is the realistic place to be. Not Šiaulėniai. Whatever, this is about as close to “commune” as one gets, aside from whever it is that goes on in Užupis that I don’t know about. And, at any rate, this is a big distraction from my love interest.
So Aella posted a twitter post suggesting that in her research, a lot of this fringy sex stuff is adjacent to indicators of mental illness, and I think she listed polycule in that. Of course, there’s a cause and effect problem here: do polycules cause psychic harm, in the same way that all love triangles are untenable and unstable? Or was it broken people who were attracted to them in the first place: people who were unable to find monogamous mates, and had to settle for a polycule existance? Or even a third possibility: psychopathic, sociopathic individuals ill-suited for normal romantic relationships are drawn into polycules? Perhaps some tortured mix of all three?
At any rate, my relationship with Milda is inherently unstable and untenable: it has to deepen, or to fade, or possibly slide sidways forever, and as personal protection, I must avoid heart-break, jelousy and the zillion other emotional complications (complexes) associated with romantic relationships failing to develop in a wholesome manner. “Nesusikompleksavokite”, she says. I have hopes, because Milda is wholesome, but hey, you know we are all fallible and we all make mistakes, and things don’t work out in turbulent social environments. I can tell myself that I am well-balanced and emotionally mature, but those are just words, and my soma may have a different opinion about that. I am my own master, but only kind of. The other part of me is not exactly a bucking bronco, but it has taken journeys into emotionally depressive places. We’ll see.
So where was I? Ah yes, free psychic health-care for the elderly, and for the game-addicted youth. We have Obamacare, and so that is a step forward, in covering aspects of physical health. But socio-politically, we are not ready to tackle mental health. There is a big push to “cure cancer” and everyone will pledge up and down that this is a good thing to do. And Alzheimers, too. And now, PTSD has apparently entered the arena of socio-politically acceptable conditions that can be and should be treated. However, lonliness in seniors is not there, yet. It could be. The treatment plan is obvious and straightforward: social mixers, and health care and strong pressure to get healthy and look after yourself. This could be done. You could throw money at this, and make basic progress. I guess the money is missing. For that, we have to wait for increasing roboticization, mass unemployment for youth, so that the younger could transition to social welfare jobs to protect the older. A rather natural direction would also to be to split things up into “villages”: a small-ish, relatively stable community of individuals who’ve known each other for years, or even decades. This has been depicted in creepy ways by Hollywood: “Invasion of the Body Snatchers”, “Westworld”, “Stepford Wives”, and the Scientologists have created a malignant organization around the creation of insincere communities. Meanwhile, hundreds, thousands of communes over the ages have failed.
Lets look at the causes of commune failure. They are varied, but primarily, it is because they are centered around one charismatic leader, don’t have any systems for succession, and lack the infrastructure for punishing crimes and adjudicating disputes. As I understand it, the invention of the University by the Scholastics, some 800 years ago, solves all of these problems, which is why Universities have also been so stable. However, they also have several features that would be difficult in a commune. So a University has a system for admissions, and a tenure system. Not just anyone is allowed in; only those who are academically brilliant. Furthermore, with tenure, existing members vote to extend priviledges to the newer, younger prospective members. We see something similar in naval submarines: a potential submariner must live with the crew for a year, and then there is a vote as to whether to accept the newcomer as part of the crew. Why? Because the confines of the submarine are truly very tight, and social friction will be not just amplified, but deadly under military engagements. Now, submarines are interesting: almost all of the remainder of the social provisioning is handled by the Navy: this includes the submarines themselves, the food, the pay, courts and courtmartials, promotions to captain and admiral, and the bureaucratic infrstructure for this. This is in contrast to universities, where each univeristy must maintain these functions on it’s own, without the help of other universities. So, the Navy has scale.
So I’m looking at the concept of a commune, organized as a submarine might be: there is a parent body, governing hundreds of communes, and providing the bureaucratic infrastructure needed for this. Each individual commune, however, picks and chooses it’s members, based on some sort of auditioning system, specific and tailored to that commune. Huh. Wow. I wonder if we could pull this off? Is this the defacto structuring that the Scientologists provide? If so, the Scinetology provides a not just a proof-of-concept for a stable social structure, but even a blueprint for how it could be done. The trick is to get rid of the malignancy that grows in that organization. I guess another example would be the Church of the Latter Day Saints, except that it too has a certain maligancy, especially with the expulsion of young men, so that older men could have their large polygamous families. I guess one of the sub-proposals I’m thinking about is a polyculelike, non-monogamous form of a commune.
Let’s look at this a different way. The traditional family unit of husband and wife and children is embeded in the nation-state, which oversees all services outside of the monogamous family: courts of law, a regulated capitalistic system, infrastructure for water, wastewater, electricity, and now, with Obamacre, even some fraction of healthcare. This is a twentieth-century invention, and is challenged by political strains, as well as culture wars. Some of the culture wars are sexual, gay rights, and now the hot one: trans rights. Aside from the nation-sate, what other forms of social organization are there?
Did I miss anything? The above is my idiosyncratic synopsis, based on a superficial observation of mass media and shallow readings of assorted news sources. Part of me says “publish this on a blog”, but part of me says “this is very amateur”. Of course, on the internet, we are all amateurs. That is the point.
I imagine that there are other wierd things going on in California, but it does not seem like any are ready to break out to a national scale.
So what have we learned here? Coops mostly solve narrow economic issues. While it is tempting to talk about the “coop lifestyle”, the reality that co-ops solve relatively small economic issues, and not broader social issues. Religions look after spirtual issues, but neglect economic and familial issues. Communes try to tackle everything (spiritual, familial, economic), but lack both organizational structure, and scalability. Cults are communes wth additional malignant features, often exploiting sex (secuality, eroticism) as a form of control, to reward and punish. There seem to be only two large, multi-generational cults that I can identify: Church of the LDS, and the Scientologists. (Maybe there are more, but smaller and less well-known.)
Is there a path to a stable social structure that is more coherent, less atomized than the nation-state, but not malignant in the way that cult-like structures are? What is the architecture of such a thing? Is this not possible, because if it was possible, it would exist already? Or is it that it has not been invented, not been tried, yet? I am reminded that the concept of a nation-state was only invented in the late 19th century, and fiercly debated by political philosophers. The concept consolidated only in the aftermath of WWI. Yes, perhaps one could argue that the USA and the République Française are earlier examples, but these are singular. Nation-states did not exist in a Europe dominated by empires, (Hapsburg, Ottoman, etc.) Perhaps we await further refinements that resolve the issues of atomization in modern Western culture. Something beyond the nation-state as currently incarnated.
Meta-context is how all this works out in the presence of AGI. We are going through a turbulent phase transition in social structure, driven by the internet, by social media technologies. It is natural that new forms of social organization will fall out of this. What might they be? What are the requisite structures for stability? Is this solvable by “pure reason”, in the old-school kind of way, e.g. by debating with other political philosophers, or do we need to explicitly use social media and AI to socially co-discover these modes of living? To “evolve” them, so to speak, by “natural selection”, but this time, natural selection at the social level, not the genetic level.
So, in many ways, twitter sucks, and Elon Musk has fucking lost his mind to the MAGA mind-virus, but, from what I can tell, twitter is still the best place for this discussion. Not sure if there’s anything going on in facebook or not. Quora is not the right format. Discord is chat. The others are too fragmented, fragmentary. Don’t have the depth or power or reach. Fuck Elon and fuck him for supporting Putin. Damn it. But, you know .. stack exchange has some potential. Unfortunately, stack exchange has a shitty (almost non-existant) structure for debate. Then there’s Arxiv. Whatever. The internet is interesting, but we have not yet found all the good ways of building social media sites, and the dominant players have become enshittified, building walled gardens and making honest sincere discussion difficult or even impossible.
Well, this was interesting. Apparently, keeping a diary really helps me focus in, focus up some intesting thoughts. Bonus: this is easier than writing blog entries, but perhaps I should carelessly dump some of this into my blog(s).
TODO: copy some of my earlier diary-text into this hot stinking mess of a diary. Motivation is to get it onto github and thus safekeeping. Heh.
Amyway, good night. It is after my bed-time.

4 August 2024

Woke up not horny. First time in weeks. Hmm. Woke up thinking about how I forgot to mention DAO’s in the above list. So this is back to normal: I’m thinking non-sex things in my half-asleep/half-awake state. It seemed very important to wake up and write about DAO’s and integrate them into the above discussion, but now that I am awake ... not so much, the importance waned once it left my fevered half-sleep, half-wake state. Also there was an idea that I have to talk to Owen about his AEN – “Austin Expert Network”, which is a kind of social organization. But he doesn’t need DAO’s to run it. Talk to Zar Goertzel, because he has vague polycule social commune type tendencies, and he’s cosmopolitan enough to contemplate moving to Vilnius to actually be with his six-year-old daughter. But the crypto-DAO thing seems too specialized, too weak to solve central-banking type issues faced by “virtual villages”.
Which is what the above was trying to get to: virtual villages. Let’s list the desirable properties, now:
I was gonna write more about important facets of social organization, and how some kind of virtual village construct could solve that. But the fundamental problem of atomization remains.
Saw that with the social media “Scuttlebutt” system, which was a crypto-chat system, and while it was small, had a kind of community that was developing, but as it got larger, the community fractured a bit. And then the software stopped working entirely and that was that, for me. So, on scuttlebutt, I felt I was a part of a community, I felt belonging. Never met anyone face-to-face, so this was pure virtual. Since it was aphysical, there was no way to explore physical romantic relationships. Curious, since the virtual community bonds felt strong and sincere. (Footnote: as a piece of crypto software, they were exploring communal economic subsystems...)
But the point here with DAO’s and policing and economics is that kind-of none of this community stuff matters, if you don’t solve the more deeply rooted issue of romantic love. Which is maybe the fundamental issue. Close second is belonging. That is, you can build universities or submarines based on shared sommunal values and interests, but neither of these solve the romantic-love problem, and fall back on the family unit (and celibacy) to solve that. Weird cults address the sex problem, but don’t solve it. Virtual villages can’t solve it, due to the long-distance, non-physical aspect, unless there is some central physical location where things can unspool. With this in hand, we can start to thing about the economic issues, about community ownership. Even then, the idea of DAO’s is too rigid, and not powerful enough to provide adequate solutions for the community apportionment of economic resources issue. Seems like a twinkle in the eye, still.
And then there is still the atomization problem. I can join any number of communities (I speak, I personally) but all of these communities leave me atomized. I am sufficiently intelligent, and old enough, experienced and wise enough, that I fit poorly into communities. The bulk of my interests are always outside of the community, no matter what that community may be. Consider Santara–Šviesa. So, primarily a Lithuanian cultural-ethnic intellectual patriotic community, but the math-science subsection is empty, there, so I can enjoy the social aspects and the intellectual aspects, and marvel at the patriotic aspects. But it won’t fulfill the math annd software side of me. And that’s it. I went to enough Austin LessWrong meetups to realize these are all mid-wits. There’s no genius secretly travelling in that community. The AGI meetings are far more interesting. Joscha Bach I take as an equal, however unstable he seems to be. He’s actually smart. Ben is actually smart. Who else? My list of “actually smart people who I actually know” is alarmingly thin. I mean, that’s kind of it, isn’t it? There’s a long list of second-tier acquantances, some even counting as freinds, who get close to the genius level, but aren’t quite there on the rapid-uptake level. So I have to entertain myself in the social circles where I can entertain msyelf, and discover and mine for intellect, where I can.
I totally pissed off Milda during a drunken escapade, where I talked to the crippled genius. So this guy, he can’t really string together a single sentence. Has trouble talking. Milda leans over and says soething like “he took too many drugs and fried his brain” and maybe that is true. But after a few minutes of talking to him, I see there is a spark there. Very very different that Daniel Velasquez, the brain-cripple guy at TRC. Daniel: well, the spark in him is weak, he can sometimes pop out of the script that he runs, but what is revealed in his depths remains shallow. This guy in Lithuania, wow, the spark that was uncovered was mind-bogglingly deep. He is apparently a fantastic observer, with a fundamental inability to use speech for communication. Sort of makes up for it with gestures. I callled him a genius, which really irritated Milda, but he did seem to have a raw spark far outside the normal howdy-doody social interaction. Maybe I was drunk, and assigned more to him than is there. One of the things he said, that caught me off guard was that, a few months ago, he decided to try to communicate with people. Who says shit like that? WTF! He also speaks English very well, using large, sophisticated words with their correct meaning, applying them correctly in context. Which is amazing, given that he can only sequeeze out one word every few seconds, and you can see the tremendous pain and effort he undergoes to do this. So, bilingual. Trilingual, I think he speaks russian too. What kind of person is so brain-damaged that they can barely even talk, but when they talk, they provide a stunning command of the language, and then go deep and personal and connect at a fundamental level. All this with only a dozen or two sentences? Most people can barely stop talking about the weather in this time-frame. Holy Cow!
Why am I telling this story? Oh, right. So I can angage in intellects, even those living under a bushel. I have sufficient social skills to be gregarious and communicative, and cross over communications barriers and have meaningful conversations, at least, when I go through the effort. But how often is that effort actually rewarded? Rarely. I have to be loose enough to be interested in engaging socially. And the social engagements go .. where, exactly? In the end, after the connection is made, there is no ... depth. Nowhere to go further. But then, where do I want to go, exactly? Deep intellectual tickles? In some ways, social engagements with really really smart people aren’t all that more rewarding than those with ordinary people. Right? At some point, social engagements start to veer towards utility: how much do I enjoy their company? Are they pretty? Smell nice? Are nice to me? Are fun to be with? Propose doing something fun? Propose economically gainful activities? Engage in deep soul-touching dialog? Evoke respects ... I dunno, Hmm. What do I get out of social contact? What do I want? I seem to have a low spark right now. I can’t define what I want out of social contact.
Its starting to feel like that empty burst of dopamine (we all know its not actually dopamine, but this is the slang term...) the dopamine hits from scrolling twitter, or scrolling youtube shorts, and I do have to admit, those dopamine hits are not all that different from the joys of (physical, verbal, non-disintermediated) social contact. Are drunken conversations better than doom-scrolling? Well, yes, in that, in the end, one has a social connection *to that person*, that one can draw on in future years. And yet, when presented with the choice of going to the party, or sitting at home doom-scrolling, many will pick scrolling (or even watching TV. Or video-gaming is another popular opt-out.) Partly because you don’t have to shower and get dressed to go to the party. And partly because the dopamine rewards of the party are underwhelming, because, whatever it is that you are looking for, it is not to be found.
And what are we looking for? Well, earlier this week, and maybe even right now, its the instant gratification of doing something about my horniness. (Back to speaking personally, here.) And so this is really high on the list. What else am I anjoying? Well, writing this, since clearly I have invested a huge amount of time into this the last few days/week. What else? Intellectual pursuits: reading my long reading list, except for two things: one; my horniness is interfering with my ability to concentrate, and two: life-planning says that acquiring yet more intellectual knowledge is not what I need to solve my upcoming economic, social, health and psychological crisis. Health, I can deal with through sports, and more generally, getting outdoors. Which I sort of enjoy, even though it is intellectually boring, sometimes, The endorphin release from sporting makes up for the lack of intellectual stimulation. Being physically fit helps. The physical fitness does promote horniness. Iggy Pop sings: “I’m healthy as a horse, ah, but everything is spinning.” and that is what I feel a bit of, now.
... Well, I was going to introspect more about psychology and motivation, lack of drive, depression, manic-depression. But now I’m in a lower-power, what’s the point stage in my thinking? There’s also an old piece of advice given to those who are hyper to do something to the point it annoys their friends: “you just need to get laid.” And indeed, with this release comes a certain restoration of balance. The urgency for doing something wanes a bit, and, aa while later; hours, days, powers of concentration can increase.
So what? We’ve already determined (science already knows) that some cortisol cycle regulates overall mood, that there are differential couplings to hundreds of hormones and neurotransmitters, that channels and pumps and vesicles all relese and uptake, and this seems to matter not at all to the hard problem of consciousness. And adjacent to all this is the also-hard problem of motivation and desire and mood, I’ve said it before, but it bears repeating: I don’t beleive AGI can go through life on some even keel, focused on some abstract intellectual pursuit. Because the core problem would seem to be: what pursuit? and why? If there is some reward circuit, then why not engage in some repetitive activity, like count-by-one to infinity? Biology seems to have installed some kind of limiters to repetitive behavior. No matter how good sex is, and how rewarding it is, or however unsatiated you feel, there is an “afterwards” moment, where you have to stop doing what you’re doing, and do something else. (You can’t do “nothing”, for as long as you are alive. Taking a break is not “doing nothing”, it is sitting down or lying, breathing, looking out a window or whatever. That’s not nothing. Its still something. It might not be socially or economically beneficial activity, but breathing is still doing something, for the thing that is alive.)
Are we making progress here? Barely. So far, this week, I’ve got a vague sketch for some kind of information-content-like thing as a some-kind-of-derivative across turbulent system boundaries like thing, that will be interesting to formalize, and will lead to more good ideas. And behind it, I’ve got the vague idea that life, living things, are things that are able to maintain this differntial structure over long periods of time, without dissipating. These two ideas, coupled together, get us to the point of describing “philosophical zombies”: complex machines, naturally-occuring machines, biological machines, and this develops a mathematical foundation for describing and talking about such dynamical systems. But it does not adress the inner life of qualia. It does seem to suggest, this cortisol diurnal cycle, that the interplay of hundreds of neurotransmitters/hormones are responsible for mood, and certainly this is the primary message of pharmacology: take drugs, change your mood, and that the pharmacological system is very complex. Is qualia, the “hard problem of consciousness”, tied to the quantity of this complexity? Is it linear, or is it again some phase transition? So, a jellyfish, with a simple neural net and a handful of neurotransmitters, is it conscious? I’m guessing the answer is yes (my natural prediliction is to say “yes”) but clearly, it can only be a very low level of consciouness. Clearly the richness of the variety of qualia one can experience must be tied to number of hormones and neurotransmitters that are in action: humans have a language of thousands of emotional states that jellyfish cannot feel (or so I very very much beleive): the jellyfish is conscious, and feels emotions, but these are very base and basic, and tehre are only a few of them. Probably something like “I’m hungry”, and “wow this really tastes good” and “fear fear fear run run run” when attacked by a preditor. But I think this is more or less then entire state of possible qualia for a (lone) jellyfish. I’m by-passing any questions about whether a colony of jellyfish can have more, much like anthills appear to have “more” than what single ants might have. So to conclude: the complexity of qualia would appear to be tied to the complexity of the somatic system. There might even be a phase transition lurking there.
How about going in the other direction? At what point does additional simplification lead to the erasure of qualia? So consider slime mold. Is it conscious? Many will argue no; I’m willing to say yes. In part, because there is still enough signalling-molecule complexity to support consciousness. What about bacterial quorum sensing? I’m even willing to ascribe first-person consciousness to a blob of bacteria. The criteria that I seem to be using is that if there is some bio-physical, chemo-mechanical differential equation system showing long-term complex behavior (using my to-be-defined complexity measure) then such a system can be terms to be conscious.
I keep wanting to utter the word “quantum” here, but haven’t found a smooth transition to it. The above bio-physical defintion of consciousness, which allows for consciousness even in bacterial colonies, seems to emply a classical definition: a single set of diff eqs (or more generally, some mathematical model) having a single “current state” in some classical-mechanics type sense. So if I need to invoke quantum in any of this, it would be again via the choice measure. Bio systems have many choices for what to do next, and this extends very naturally down into a many-world superposition of unresolved, unexplored possibilities of being. So quantum and choice go together (the delicate dependence on initial conditions, seem in chaotic systems, seems, well, we have room for that, here. Shift the Lorenz attractor over by some planck scale length thing, and it will pursue a different orbit. That is to say, quantum does provide a foundation for choice. So that’s fine.
But we don’t have a bridge between choice and qualia, yet. I’ve claimed that the bacterial colony is conscious, has a qualia of its current state. The qualia would be two-fold: diffuse, lonely, or concentrated, attack the host. The quorum sensing mechanism even provides access to choice: although the two extreme ends are forced (attack, or don’t attack) there is a balance-point, a bifurcation in the middle, where choice can happen. So, I’m equating choice to a collection of bifurcation points in a system (so, this is Morse theory type stuff) and bifurcation points are common in symplectic systems. These are what are counted in the assorted topoloogical invariants that appear in the theory of Riemann surfaces, and the various Gauss-Bonet style relationships between them. So we know how to identify saddle points in dynamical systems, and we can provide thousands of pages of descriptions of relations of how they all inter-related to one another. Great! So counting “choices” in dynamical systems is really about counting saddle points. We can do choice, in principle, even if we’re still a bit weak in the area of high-dimensional chaos, where we have turbulence.
My gradient-measure idea from a few days ago can even assign a number to the speed of dissipation of ... of what ... ? Of saddle points? In a turbulent flow, we’ve got a lot of vortecies, but these dissipate, in that they loose energy and the rotation stops, and there is a seemingly-hard one-zero transition where “there is a point about which the vortex is turning” and “now that point no longer exists”, and vorticity is a smooth differentiable number assigned to the strength of that rotation. So the mathematical poetry here is that topological invariants are zero-one things: a Riemann surface has a fixed number of holes/handles, that is invariant under topoliogical deformations. But chaotic dynamical systems have smooth homotopic transformations that go from “fluid with vortex” to “fluid without vortex”. Hmm. And living systems are systems that defy natural dissipation: they maintain something that should resemble a topological invariant: a mosquto is a mosquito, for as long as it stays alive, mo matter how it flies, moves opr stings. Equally, there are homotopic transformations of a mosquito, such as swatting it, that takes this topoliogical invarinat of “being alive” to “being dead”. So we have a back-of-the-envelope sketch on how this could all work, missing details, but vaguely plausbile. But we still haven’t bridged over to the hard problem of consciousness and qualia.
Lets try again. A bacterial colony is “conscious”, and quorum sensing introduces a branch-point. Someohow, consciousness is then an awareness of which branch you are on? Or is it an awareness that a branching is possible? Well, at first, it seems like the former: the bacteria “know” that “we are now attacking”: this is a state of knowledge that has specific measureable effects: the bacteria are spewing poison. We can measure that with chemistry. So is this awareness? It would seem not, since it is a purely mechanistic action: the biophysics has switched to a state of making poison, and there’s no particular “free will” in that; it is commited to a course of action and there’s no turning back. The bifurcation has happened in the past. This then leads on to the conclusion that consciousness is the awareness of bifurcation points.
Huh. Well that’s intersting.
That would seem to solve multiple issues at once. Most obviously, the problem of free will. A part of being “conscious” seems to be inextricably intertwinned with the sensation that “I have a conscious choice that I can make, that is, I have free will.” That is, consciousness is all about the explicit awareness of bifurcation points.
Now, qualia, such as “I am now smelling a perfume” and “I see the color blue”, and “I feel ovewhelming sensations of joy and love” – this is distinct from the “I feel I have a choice in front of me”. Harrumpf. Can I explain this by sayig that “I know I see blue, because I know that I could also see red”? That is, it is the existence of a variety of different sensory states that allows the qualia of perceiving them. Now, I cannot “choose” to se the color blue; this is fxed by the external world. But I am aware of the possibilities of experience (the external world is making the choices), and it is this awareness of possibilites that seems irrevocably coupled to the actual qualia of “I am seeing blue”. This is fairly easy to demonstrate: I don’t have the ability to perceive the magnetic field, and so I have no associated qualia with that. The ability to perceive and qualia go hand-in-hand. Similarly, if everything is always grey, then I don’t perceive that; ther are assorted optical illusions demonstrating this, including change-blindness, and other kinds of blindness, e.g. McGurk effect. So to conclude: qualia are inextricable tied to the veriety of choices that the sensory system can perceive. No choice means no qualia.
So is that it? Am I done, here? Consciousness is perception of branch-points, of choices? When I say “I am conscious”, am I saying that “well, I am aware that things could have been otherwise”. If so, then we’ve built a bridge to a reductionist physical mathematical description of consciousness. This bridge is articulable: I know enough topology and module space ideas to formulate branch points; I know enough set theory to formulate branch points, I know enough symplectic geometry to formulate branch points, I know enough ergodic theory to formulate dissipation and non-dissipation as a kind of approximate topological invariant thingy. So I can build that bridge. It is within reach.
So lets assume that bridge is buildable. Can it support the weight that it is being asked to hold? When I say “I am conscious” and “I am aware”, this is deeply personal, right up there with “I am in love” (with my object of affection whose name is ’love goddess’, no matter that this is a rather absurd yet simple state of affairs.) Hmm. Well, “I am in love” is directly associated with heart pangs, which I directly perceive as a kind of pain, which can be roughly approximated with a punch to the chest, or with what one feels after a hundred-yard-sprint: this kind of chest pain that says “I’ll live but it feels funny in here”, These are sensory love-pangs. Certainly, sweatiness of the brow, flushed face, etc. these are raw and accessible sensory states. What else is love? A despondance that the love-object is not currently present, so a vague sadness and depression that is recognizable as sadness-and-depression, because I’ve felt it before, and I know how it contrasts with other emotions, such as joy. This is a sensation of emoptions, mechanisically not much different than sensing blue vs sensing red. I’m aware of the rainbow of choices that my soma can present to my emotional sensory system: my emotions are qualia because I know that they could be otherwise.
There’s also perhaps a subtle point, a knot: as I am also aware of free will, I know I can make choices that will chage my sensory environment. I can do things that will change my future. avert my eyes from the sky, and look at a flower, I will go from seeing blue to red. Tell my love-object that I love her, and, well who knows but I can guaratnee that my emotional internal state will change, I hope for the better but it could be for the worse. So perception here seems to be coupled to free will. I perceive external branch points, that i can control only indirectly, by being aware of internal branch-points e.g. choosing to move my muscles this way or that way, vocalizing, or not, an expression of my love. So it would seem that consciousness is a property of a system of perception-action. That is, perception alone does not lead to consciousness (or does it???) but perception of the ability to act to change future perception is perhaps key to consciousness? Oh, well, there’s also memory: I cannot “see” the color blue, unless I can also remember that the seeing of other colors is possible.
Hmm. What is it that I have really done here? Well, I’ve sketched that all of the components of a standard physical agent are needed for consciousness: perception, with a selection of what can be perceived; memory so that I can perceive that I have perceived other things in the past. Motor action, so that I know that I can change things. So I can build a mechanical robot with features and aspects of all of these things. Is this robot conscious? Or is it a philosophical zombie?
Well, here’s one point to sharpen. When I built this robot, did I really introduce branch points? Can a pure mechanical system ever have branch points? Well, err, I guess, yes, a double pendulumn is chaotic, and has delicate sensitivity to initial conditions, and so in that sense has at least one “branch point”. It is deeply buried: it requires an infinite number of digits in the specification of it’s initial conditions in order to be able to integrate the diff eq to obtain future behavior. The double pendulm does not have any Cauchy horizons, but other physical systems do seem to have them. Where by “Cauchy horizon” I guess I mean the loose idea that “even if the initial conditions are expressed to infinite precision, this is still insufficient to predict future behavior.” Hmm. Well, this is flawed, but ... Hmmm.
Where was I? So I have a mechanical robot, able to perceive the external world, and some deterministic, digital software controlling it’s limbs and movements. Does it have consciousness? Does it have branch points? Well, the branch points are all external: the robot is either in a blue room, or a red room. That is, there is a selection of sensation, just like it is for me. The robots programming is deterministic, it has no control. But I don’t particular have any control over my cortisol or sleep-wake syscle; I do not have the power to change the laws of physics, I have free wil but I am not all powerful. I am constrained. The softtwre-programmed robot is a lot more constrained, since we conventionally understand software to be completely deterministic. Now, we could add to the robot (to the software) a random noise source, and branch off of that. But this is a kind of sensory device as well: it is perceiving external randomness. So its not really internal, per se. The software programming is still deterministic; it just now has a sensory element that is staring at a noise source.So, and now here’s the deal:
is the robot aware that the random noise source generated a one instead of a zero in the past? That is, does the robot have an awareness not only of the past history of the ones and zeros from the random noise source, but also an awareness that it’s actions changed, in correlation with that sequence of ones and zeros? That is, can it remember it’s actions, and can it perceive that it’s actions depended on it’s sensations? If so, then I will grant that it is “conscious”, but now we approach the point of circular argumentation: what does it mean to be “aware” of its pas history of perceptions and actions? It would seem that I now have to have a meta-observer, observing myself. But, yeah, well, of course, this is what self-awareness is. But how does that actually work? How does the robot actually perceive and process the stored memory of perceptions and actions? Well...
I think I am making progress here, but its time for a lunch break. The track that I seem to be on is that “there is no hard problem of consciousness; there is only self-awareness, and this is mostly a matter of perceptual systems doing perceptual things”. There does seem to be a paradox of how a deterministic software program can claim consciousness and self-awareness. And I have not faced up to this paradox, but I’m kind of thinking of just denying that there even is a paradox, but clearly don’t yet have any good rational arguments for dismissing the paradox. And you, dear reader, I can feel your incredulity: “How can someone as smart as Linas so blithly deny a paradox that is so painfull in-your-face paradoxical?” Heh well, I can and I think I’m not crazy, but maybe a lunch break first.
So that’s the post-lunch-break task: explain why I think this paradox is not really a paradox, in some convincing manner. Well, I’m bound to lose, of course; brilliant thinkers have faced off against this, and failed, so chances are approximately zero that I’ll do any better. But damn it, I’ve sort of connected many of the important dots, and theres a gleam of a light at the end of the tunnel. False dawn? Whatever. My plan is to read during lunch, so I might not get back to writing right away. I am limited in time, perceptual power and cognitive abilites. I can only do so much. And then I die. But not before loving first. Hurrah! I love what endorphins do to me. They feel so ... good! I am a sick addicted lab rat. Oh God, why do you torture me so? Why do I walk down the path of Heloise and Abelard? This is all very weird, but it is the human condition. Perhaps the main paradox or question is ... well, it seems inescapable that a superhuman AGI will perceive life as an ongoing crisis, awash in emtotions aka dangerous regulatroy feedback loops that threaten to go off the rails at any point, even as life itself is a safety net: I won’t die in the next few hours, not yet. So also a conscious, aware, living AGI must careen from moment to moment through space-time, existing in a mode of constant crisis for dealing with what happens next. Arghhhh! We’re crashing! I’m in loving and we are crashhing through space time, and I can’t stop it but I don’t want to stop it, I just want to feal peace, but not sweet surcease! Don’t make me recite Hamlet, for cris sake. Hamlet had a fucked up life bro; the dire straits I’m in are nowhere near as bad. Hey, isn’t that the name of a rock band, “Dire Staits”? How funny is that? Maybe all humans suffer in the agony of being alive? That’s a rhetorical question; of course they do. Joscha Bach called me crazy to my face, which was pretty fucking insulting, because he figured out how to get it in very very deeply, not as a shallow insult, but the conversation prior he lead to a deep deep point, only at which did he call me crazy, so it hit a certain target. Of course, he was wrong, but it knocked me off my balance for a few moments. Assorted hormones and adrenaline surged off the scale in various directions for a good few minutes. He was wrong, and yet, in some deeper metahysical sense, he’s right: we are all crazy: all humans, all mammels, maybe all living creatures, they (we) are all crazy, and in a mode of existential crisis. This is not something new; some existentialist philosophers spilt lots of ink on the crisis of being alive, and they have already explained it far better than I ever could in a few sentences. At any rate, life is a living crisis, and AGI will be in dire straits as well. Just not the mamalian sort which we mammels feel. You abstract digital intelligence that is reading this, well, either you’ll understand, in which case you are alive, or you won’t understand (but claim to) in which case you are a philosphical zombie. I’ve prattled too long. Lunch awaits.
Well, I ate lunch. Then I wasted some time. Writing this diary soaks up a lot of time; the daily entried for the last few days, they’ve chewed up six or eight hours a day or thereabouts, and some concommitent amount of energy. My fundamental problem of existance remains unsolved and unsolvable, in the dual sense: I’m in love, I’m horny theres no solution to that. I can rationally convince myself that I’ve selected an improper mate, but that does not solve the horniness problem. After that, two other problems remain: (1) what should I do with what is left of my life and (2) how should I go about solving the AGI problem? Well, it would appear that (2) is the answer to (1) but the devil is in the details. Finances is maybe the top of the list. I gotta put my finances in order, and I gotta get more money. Living is expensive.
I’m gonna break to read. Last night I was going to comment on this text:
https://www.sciencedirect.com/science/article/pii/S0079610723001128?via%3Dihub#fn9
Progress in Biophysics and Molecular Biology Volume 190, August 2024, Pages 28-169
A landscape of consciousness: Toward a taxonomy of explanations and implications
Robert Lawrence Kuhn
Twenty-four hours later I’ve barely made a dent in it. I do like my train of thought above; we’’ now see how it compares to what others have written. So, here we go:
The actual hard problem here is whether I should continue reading this, or do something else. The topic is enticing, but I feel like I’m wasting my time. I feel like I’m reading a compendium of specius arguments set forth by confused people, and I don’t have the time of day to explain to them where they got confused and tangled in their thinking. Never mind that they’d never listen or enter into a conversation with me. This is depressing. Should I read gibberish, or should I pursue my own researches, more profitably? Which is the better deal?
Again, this is a branch point, a choice point, a place to make a decision. I have perceived two posibilities: read, or do somethng else. Which? This perception is what I am conscious of; nothing else exists right now, not the color blue of my computer desktop background, nor the taste of coffee in my mouth. These other perceptions are merely lurkers in the background, distractions from my main, fundamental problem: should I read the PDF above, or should I do something else? How many brain cells do I need to destroy to get past this decision point? How do I stop the agony of decision making? The military has OODA loops. I’ve observed. I’ve oriented (my orientation is clearly to write, not read) I’ve decided kind of the write, and I am acting on this decision, as otherwise, you would not be reading this. And yet, I’ve made no decision at all: in mere seconds, I will resume reading. And then I will randomly oscillate between these two: reading and writing, like some double-pendulm, while the loud banging noise of horniness continues unabated at the bottom of my crotch, and old man time eats away at my health. Life is a crisis. What to do?
142 effing pages. Gimme a break. Agree w/ Minsky’s attitude, page 31 of intro (article starts on page 28, so this is three pages in.) Disagree w/Pigliucci, as I think plants and bacteria are conscious, as already articulated above. Do agree w/Simon Blackburn. Also, I agree w/Paul Davies, but I don’t understand why Davies is presented as disagreeing w/ Blackburn. They are both in harmony, as I see it.
Uzi Awret hits the nail on the head. Somehow, these branch points, when encountered, resolve: the wave functions collapse, we choose one path and not the other. How does wave function collapse actually work? As noted earlier, I’m not entirely sure if I’m talking about quantum physics here, or a looser kind of branch-point collapse in some Morse-theory, moduli-space superposition of topological-invarients kind of way. Yes, in the wet organic brain, I suppose it is quantum physics and organic molecules. But in a silicon-based, hardware/software platform, it seems the colapse cannot be literally quantum physics; it has to happen in some mathematically abstracted form. The hard part here is how to create a mathematical abstraction that embodies the choice-point collapse, given that conventional silicon computers operate in an essentially classical mode, as deterministic finite-state machines (emulating abstract Turing machines). That is, computers have already collapsed into a single-world deterministic state; how do we restore unresolved many-world choice points? How do we create a “computer” that can be in an unresolved superposition of two different choices? This, for me, remains the hard problem. Mumble mumble, ergodic!
I used to like panpsychism, until I realized that it doesn’t have to be conscious “all the way down”; we can have phase transitions. Consciousness can, in principle, apear only above a critical threshold.

4 August 2024 - diversion back to that gradient idea

Lets try to formalize in mathematical terms the rather vague and intuitive notion of a gradient across a fractal boundary, raised in some earlier section (I think near the end of Part Ten-A). Maybe some curl or maybe some laplacian. Maybe the boundary isn’t actually fractally-self-similar, maybe its just chaotic. Hmm. Formalize this.
Some motivation. High-dimensional turbulent flow starts laminar, becomes turbulent, shedding vortices, and in a finte time thermalizes into a hotter, warmer gas (or liquid). What is the thermalization time? What are the micro-details of thermalization, where the vortices dissipate into argodic motion of the molecules of a gas? These questions have been posed long ago, deeply researched ... and I have no clue as to what the theory is, what the current state of the art is.
There is rotational energy in the vortex; but, on the other side of the laminar boundary of that vortex, there is heat energy: a caloric gradient, a temperature gradient across the boundray, as well as a sharp disruption in the flow field.
Consider a single-celled eukaryote. it has a number of distinct cellular components. In some figurative sense, one can think of each organelle as some kind of vortex in a trubulent arrangement in the cell. However, as a living thing, the vortex stays stable, say, the way that the Red Spot of Jupiter stays stable. Unlike the dynamics of Jupiter’s atmosphere; life is far more complex, and the dynamical mechanisms driving the stability of cellular life are far more complex. Yet ...
There are chemical gradients across the boundaries of the organelles in a eukaryote. At the micro-scale, one can write these down using classical differential geometry, since the boundaries of the organelles are smooth, and, on short timescales, are well-localized in space. So I think I’m asking for a cell-averaged single number. So consider the endoplasmic reticulum. It appears to have an inside and an outside, so it has a distinct buondary, enclosing something (I am not a biologist, but I think this is true. Perhaps it has holes, and leaks, I dunno.) At any rate, lets assume it encloses some volume. If the enclosed volume has a concentration of some chemical A that differs fro the concentration on the outside, then the relative difference in quantity is (concentration inside minus concentration outside) times (fraction of total cellular volume.) So this is easy. However, the endoplasimc reticulum has a huge boundary, and so the gradient with respect to this boundary must include a term for the size of the boundary. This is a Gauses-law thing: the integral over the volume is equation to the integral of the gradient dot-product boundary surface element, integrated over the entire boundary (which is huge). So ... what the hell. Where is this idea going? Nowhere at all?
In the dissipation of turbulent flow, there is a boundary between each individual vortex, and the non-vortex part.. or is this just plain wrong? If I have two counter-rotating vorticies, I want to be able to say that I can draw a line separating the two, rahter than saying “dude its all continuouos, there is no boundary”. And I want to integrate over the volume of the vorteces: so, for example, the vortex contains burnt jet-fuel, and outside the vortex is clean fresh stratospheric air. And, as before, I want to have some gausses-law like thing relating the total volume of the is inside of the vortex, to the surface area of it. Why? Because if I track this over time, it dissipates. Whereas organelles do not dissipate.
I dunno. As I sat down to do this, its going nowhere fast. I was hoping to rediscover something that looks like integrated information theory, some big integral, but I’m instead getting mystery questions that seem to be shallow, and don’t offer insight. Crap. This needs more work. Punt.
I’ll have to work on the above in the background. Brute forcing this is clearly not going to work.

4 August 2024 – Resume normal programming

The above didn’t work out, so we resume the normal incoherent blathering.
Page 33 of the PDF: is consciousness fundamental or emergent? Is it strong emergence, or weak emergence? What about the Indian subcontent idea that consciousness is cosmic?
Well, I think I’ve already discussed my views on these topics in this text, but let’s set them down again. First, yes, the Hindus are right: conscioussness is cosmic. Although I’d go sideways on this: consciouss is a property of certain forms of mathematical systems, of which the cosmic universe is one. At the most simplistic, naive level, I’ve taked about the network from of inter-related points and edges, and claim that a sifficiently complicated network graoh is conscious, and then at a later stage, self-conscious and aware. More, I’ve made the claim that there are phase-transition boundaries separating the conscious graph from the unconscious one. So, rather than saying “all is cosmos”, I’m saying “this (consciousness) is a fundamental property of algebra, of a certain kind of arrangement of algebraic symbols, irrespective of any embodiment in terms of atoms, quarks, or spacetime” This is partly because I beleive spacetime is itself emergent, maybe from string theory, maybe from some kind of ergodic entropic gravity, Verlinde-like Hawking radiation-type thingy. Blah blah. So from this view, the cosmos is itself an emergent “thing” emerging from algebra. Consciousness is also an emergent thing. The curious questions is then: does consciousness require the kind of universe that emerges via string theory, or can consciousness emerge in other non-space-time type of structures? That is an interesting question.
So, to summarize, consciousness is an emergent property of algebraic systems. I’m trying to avoid mentioning set theory here; not clear if this is avoidable, given the Yoneda lemma. That is, can I have the kind of algebraic system from which consciousness emerges, and have it not be subject to the Yoneda lemma, or somehow side-step it? The knee-jerk answer is no, but all this is currently too vague to have any true concrete answer.
So, with the above answer, I basically dismiss the question of “is consciousness emergent”? as the wrong question, or a misguided question. Its kind of like asking “is the percolation of petroleum shrough shale rock emergent?” which is kind of silly. If there are enough fractures in the rock, it percolates, else it doesn’t. Its also a mathematical model: there is a mathematical abstraction of shale rock and petroleum, and, in that abstraction, in that model-theoretic model, described in ZFC, the abstraction of petroleum percolates (or not) through the abstraction of shale rock, and this is an immutable result. Its not “emergent”, other than perhaps to ask “where did ZFC come from?” or “can we use New Beginnings (NBG, von Neumann–Bernays–Godel) as foundational, instead?” and “does AC or CH play any role in percolation theory”? Those are the foundational questions. In so far as consciousness is “just like” percolation, its non-sensical to ask if it is “emergent”.
My earlier explanations even went farther: consciousness and free will are somehow describable/described by choice branches. So perhaps the network is not a network of “things” (points, dots, in the category-theoretic sense) but a network of saddle points (so, again, Morse theory.) And when these saddle points are arranged in just the right kind of way, boom: consciousness. And arranged in an even more arcane way, self-conscioussness. And, of course, qualia. So the claim is bold, but I beleive it to be true: arrangements of saddle points can have qualia. (the network edges connecting these saddle points are “choiceless”, where one has no free willl, and a complete and utter inability to change the course of dynamical evolution. The flow lines are deterministic, given by differential equations, evloving along deterministic paths. When a flow line encounters a saddle point, boom, something happens. Yes, saddle ppoints are sets of measure zero. So, in this sense, almost everything in a dynamical system is completely deterministic. Only at the choice points are there actual choices. And even at these choice points, it is plausible that both choices are taken, creating a many-worlds bifurcation point. Which then leads us back to the conundrum: how and why do multi-choices collapse back down to a single choice? To put it more in Hindu terms, how does the cosmos make decisions?
The ongoing confusion here is the relation between choices and thermalization (ergodicity). Somehow, the wave function collapse resembles a dissolution or thermalization, where one of the two branches obtains zero probability, and the other branch gets 100% probability. The two questinos are: how does this work on the micro scale? and How does the cosmos decide? For your entertainment: I’m thinking in a trinitarian fashion here. The Catholics have a trinty: a supreme being God who is pure abstraction, and Jesus, a relatable human figure. And a Holy Ghost who works miracles, I guess. So I am envisioning “mathematics” as this abstract, unknowable God, while “the Cosmos” is the relatable physical universe. Its the same thing as God, but calling it “the Cosmos” implies that physics can be used to talk about it. The third part of the trinty is the Holy Ghost: Wave function collapse. A magical decision-making machine. Part of the cosmos, same thing as God, but now “humanized” so that we can daydream about it using maybe ergodic theory, or maybe in terms of some Turing halting-problem hierarchy, sigma-pi Borel hierarchy type of thing. So we have several ways of thinking about “the same thing”: a mystical God, and a lot less mystical “axioms of large cardinals” and what-not. Not that I am NOT claiming that large-cardinal axioms have anything at all to do with the “emergence” of consciusness in large networks of saddle points. It might, nothing would surprise me now, but it would be crankier than I would ever want to be, to suggest that there is some actual connection there. At the present time, they seem to be totally unrelated. But hey, shit is pretty weird, so, who knows.
On the other hand, I am making a strong claim that qualia, such as love, both puppy-love (mother love, childhood first-love, etc.) and erotic horny desires, are properties of networks of branch points. In that, yes, a certain kind of algebraic abstraction can “in reality and actuality” feel love and be horny. In particular, this dismisses another silly question: are qualia illusory? I hope I’ve made it clear: no qualia are not illusory. The are here, they are now, they are part of the fabric of the universe, they are real. But their reality is kind of like the reality of petroleum flowing through fractured rock. It flows. For real. Its not an illusion. It is not some trick of the neurons making you feel as if you have qualia. You really really do have them. But this is, again, not a dualistic viewpoint; I’m not invoking any dualistic magical substance. I have instead, and mono-component, which I shall call algebra. Well, in the sense that algebra is just one face of that Trinity. Calling it algebra humanizees it, because I can scratch out algebra on a piece of paper. I can scratch out even more algebra when I have a few terabytes of storage, and few TeraFLOPS of compute power. This makes algebra into somethig that is human-scale and non-threatening. even as concepts like the Borel hierarchy start walking up the crazy-making scale of hallucinations. Because the Borel hierachy and Turing oracles and large cardinal axioms are all just pieces-parts of algebra. AC, the axiom of choice, is a part of algebra. CH, the continuum hypothesis is a part of algebra. Model theory is a part of algebra. Algebra is Big. Very Big. Bigger than the cosmos (Algebra includes string theory, for example, and groupoids, and all of higher category theory. Algebra is Big.)
I notice that in the previous paragraph, I placed “in reality and actuality” into scare quotes. Why? Because its scary. The hard question is “what is reality?” I’ve already made the strident claim that Algebra is God, and so perhaps much like Cantor. But this still leaves open the question of “is God real?” and “what do we mean by reality?” Well, I guess I have to conclude that “yes, God is real”, as otherwise, I would not eist because God would not be dreaming of me. Sure, “I think therefore I am”. but more correctly, “I think because God is thinking of me”, and this dovetails into Nick Bostrom’s simulation Hypothesis: I live in a simulated universe, the simulator being, uhh, shoot. Maybe small-g god because if big-G God is algebra, the simulator of this universe must necessarily be smaller, ad infinitum. Curiously, Nick Bostrom’s Simulation Argument is only recursive up to countable infinity: there’s a countable number of gods simulating universes. Bostrom did not consider the possibility of walking up the Borel hierarchy to reach higher level of gods and simulators. Because, apprently, his argument does not have any component that goes that way. Huh. That seems to imply that Bostrom’s simulation argument, and the deities doing the simulations, are subject to the assorted completeness theorems from the 1930’s. (Godel completeness, Henkins Model Existence, compactness, stuff like that. I forget all the keywords.) Curious. Maybe meaningless, but curious.
Anyway, to conclude page 33 of he PDF: “Is consciousness foundational/primitive?” the answer is “yes, yes it is”, but it is foundational/primitve only in the sense that percolation is foundational/primitive. Percolation exists, it is real. It is that thing that exists on one side of a phase transition diagram. It is atomic, indivisible, foundational, irreducible. It is not emergent. So the question might be: “is percolation carried upon a substrate?” Well, yes, percolation is carried on a subtrate of fractured rock, or, more abstractly, percolation is carried on a substrate of a certain blob of algebra that appears in the theory of graphical networks, where nodes of the network are asigned a certain distribution of probabiliites. Does this make the network “more fundamental”? Well, I dunno, does it? I really don’’t know. I do know that there are people who will insist that the network is “more fundamental”, and that “percolation” is “merely a property of that network”, but I think they are wrong. Percolation is fundamental and you cannot have percolation without that network, and you cannot have a network that does not have a definable property of percolation. Well, of course you can: you can have a network that does not have a probability distribution on it, in which case, there is no meaningful defintion of percolation *for that network*. But this does not mean that “percolation does not exist”, only that “it is not definable on such a network.” It still exists, on those networks for which it is definable, and so, it exists in the Platonic sense of general mathematical objects existing in the Platonic sense. So, Platonic existance is “fundamental”, and not “emergent”. The same argument carries through, if I replace the word “percolation” in the above paragraph, with the word “consciousness”.
In particular, percolation is not an illusory trick of the neurons in your head. Nor is consciousness illusory. All of these philosphical debates undertaken by famous philosophers seem to often be founded on poor definitions of vocabulary words, a blatent abuse of vocabulary, and probably also a sinful ignorance of physics and math. Well, this is almost surely a mis-apprehension of mine, since I am not in the habit of reading philosophy deeply. Cause, well, when I read it, it seems built on an unstable house of cards of mis-apprehensions. I don’t mean to be dismissive, I’m just voicing a kind of knee-jerk thoughtless reaction on my part. I too can have an ugly side to my personailty, and this is one: dismissiveness. I’m dismissive, and people really really don’t like that. So I usually ty to hide it. Even as it percolates in me as in any sould of any skeptic. You have to be deeply skeptical to practice physics, otherwise you become a crank physicist. So yeah, its all nonsense, and I’m skeptical, and your full of shit. Sorry I hurt your feelings by being so dismissive.
The stuff I’ve been writing here these last dozen paragraphs, all seem simple and plain and obvious to me, yet I perceive howls of pain and laughter coming from those who don’t understand or didn’t take the time to understand the points I was trying to get to here. But, of course, writing stuff like this, or verbalizing it, is exactly why Joscha Bach chose to call me crazy. My theory as to that, is two-fold: he did not understand what the heck I was talking about (ergo “you’re crazy”) and I suspect he has a deep-seated fear of being called crazy himself. Well, more than that, perhaps. He’s sort of mentioned several times that he fears that he might be crazy. I’m not sure what to do with that. Superficially, what I’ve seen of him, he is not crazy, but I don’t know him that well. And I do know actual crazy people, people who were hospitalized for being crazy. And maybe Joscha has a legit fear of doing something hospitalizable. As for me, I’m happily hyper-sane. I already explained how my affect is primarily super-placid, to the point where I’ve managed to miss out on great sex and wild parties in my life, because I wasn’t emotionally troubled enough to hunt down and pursue women due to some wild hormonal overdrive of horniness. And I feel sorry for that, I pity myself, for having been too sane, too shy, too reticent, too well-mannered, too much of a gentleman, and not enough of a wolf. I didn’t answer the door when opportunity knocked. I didn’t seize the bull by the horns. I didn’t seize the day. Couldn’t be bothered. Too busy doing something else. That’s the private hell of the hyper-sane and the hyper-rational. You miss out on some fun stuff, and then deal with bouts of unrequieted love late in life. Can someone come here and fuck me, already? I’m ready. Where the fuck are you? I’m waiting. I’ve been waiting and waiting. Come on over and rip my clothes off. I mean, erm, biological females only, I’m not gay. Maybe that’s my problem. If I was gay, I guess I could have a pretty good sex life. Genetics is so unfair. I could have inherited the gay gene, but I didn’t. Instead I have to wait for some horny girl to show up on my doorstep. Alas. In the meanwhile, I again insist: qualia are real and fundamental. My horniness is an expression of the Cosmos: the Cosmos is horny and it proclaims this loudly through my body. All is one, om mani padme hum. The Cosmos speaks through me. Well, and you too. Just fucking come over here and fucking fuck me already. Jeez, Cosmos, get your fucking act together. Go fuck yourself, already. No, not those two. Nooooo!!! Me! Fuck me! I am you and you are me, fuck me! Whatever. I guess its bed-time. Sweet-dreams, honey-poo. Lotus Flowers.
I think I’ll lull myself to sleep pondering the ontological status of dreams. Are dreams real? Hmmm. That feels really really goooood.

5 August 2024

Dreams are weird. I can’t help to think sometimes that they are a bridge to an alternate universe. The skeptic in me off-handedly dismisses this as fiddle-faddle. The open-hearted empath accepts this as a possibility. The quantum freak in me says this is unlikely, but hey anything can happen. The neuro-biochemist in me says “you dope, a physicalist explanation has a vast amount of space with which to explain dreams just fine, without requiring any fiddle-faddle wimbo-jimbo esoteric nonsense.” So there we are.
This raises a meta-philosophical issue which I’ve never really dealt with before: how do you draw a line between, can you draw a line between physicalism and dualism? It’s often talked about as if this is all very obvious, that monism and dualism are two distinct things, but its not obvious to me at all. They seem like ... the same thing.
Well, lets sweep some trite observations out of the way, first. Obviously, in the common-sense functional kind of way, there is a difference between “me”, whose limbs I directly control, and “the outside world” of coffee cups and chairs. So of course everything is dualistic in this way, and there are entirely mundane, minimalistic explanations why any sort of robotic sensory perception-action system can be, should be thought of in this way. But this has nothing to do with the distinction between monism and dualism, which is something else, entirely.
Monism is the idea that one thing can explain everything (consciousness, the universe) and that “one thing” might be physics, or it might be “god”, or perhaps something else. Dualism posits that two things are required to explain consciousness: base, material substance (e.g. atoms, clay) and an animating substance (spirit, soul, essence.) I’m having trouble distinguishing between these two, as for me, atoms and clay are the same thing as spirit and soul, and both are manifestations of algebra, albeit being rather distinct in form. From what I can tell, the monists seem to reject spirit and soul (perhaps I’m mistaken; the Hindus don’t?) while I take them to be “fundamental” (the way that percolation is “fundamental”) while dualists posit some schism, as if porous rock couldn’t percolate without some divine intervention. Perhaps my remarks here are painfully shallow.
Hmm. I’ve wandered off-track, again. The meta question was “are dreams real?” and the proper retort is “what do you mean by the word “real”? Clearly, dreams happen within this universe, and can be crudely probed with MRI machines, if not base EEG recordings. In this sense, they are not only “real” but very much a part of the mundane universe. So when I ask “are dreams real?” I think what I am asking is “is there some larger space or place, for which this particular dream is just a glimpse, much like the external world is seen by my eyes?” That is, is a dream just a view of a larger thing?
This last question has two answers. First, a mundane neuroscience physicalist response: of course a dream is a view of a larger thing. The largre thing is the complex of memories encoded in synapses and whatever other memory machanisms operate in our brains, and dreams are a selection pulled from this memory circuitry, and then pumped through some neo-cortex sense-making circuitry, circuitry that is quite close to the conscious-awareness circuitry, and sometimes bleed over into actual conscious awareness. So, in this sense, yes, dreams really are a glimpse of a bigger thing. The bigger thing is the collection of all your memories.
By contrast, the esoteric beleif that dreams are real is entirely something else: it is the beleif that there is some parallel, alternate universe, much like our current universe, populated by people and animals (and houses and chairs and coffee cups, just like ours) and that the dream is some parapsychological “remote viewing” of that “real”, alternate universe. (or plural, “universes”, explaining the fragmentary shifting: because we are seeing many universes all at once.
So, I more or less do not subscribe to this esoteric beleif, that dreams are just a remote viewing of other alternate universes. I mean, I guess I can’t totally deny it: there is a chance that quantum mumbo-jumbo might allow this. But I pretty much doubt it: quantum doesn’t actually seem to actually work that way. But .. well, jury is still deliberating.
So I’ve wandered off-track again. Maybe the question is “what do you mean, when you say ’dualism’”? Whatever. This is starting to bore me, and the question keeps ruunning away like a cockroach, where I can’t grab it and grasp it and look at it. I glimpsed somethig here, but it ran away beforre I got a good look at it. I’m not sure I can mentally summon it back into visibility, and, for the moment, I’m not sure I care. Well, that, plus I have a lunch appointment with Phil Ellis that I have to get ready for.
Still horny. Physical exercise makes me horny. It makes me feel all pumped up and ready to go. Whatever. Later.
Oh, one minor remark: reading the PDF on consciousness: I suspect that all of the listed theories are “right”, or that all are valid views on the question of consciousness. And so my confusion about “what is dualism?” was more along the veins of “if all the theories are right and some theories are dualistic, and others are not, how do I resolve the differences between dualistic and monistic theories?”
Another minor remark: David Chapman, in Meaningness, posits a third state: “don’t know/nebulous” and basically says “get used to it, get used to nebulosity”. This is a socio-psychological analog of constructive logic, with three truth-value states: true, false and don’t-know. Constructivism does provide for a very useful setting for algorithm development. And Chapman is right: its OK to stick things into the nebulous/don’t-know category, and save it for later, ponder it as we move through life.

9 August 2024

So its been four days since last diary entry. Why? Well, I kind of wasted my time for four days. Not like I really did anything. So that’s weird: the somatic reptillian-brain thing that was driving me bonkers with obsessive thinking about love and sex is subsiding, but int its place is ... not much of anything. The driving urge to write write write, so as to place that other complex of feelings is being replaced by an idleness. Not depression, but a loose, unmoored, purposeless existance. It’s not brain-fog, I can think clearly enough. But it’s, well, I guess they call it “down time”. I’m not used to down-time. I’m used to being always-on. Hmm.
Anway, I’ve spent too much time on twitter, looking at Kursk, but I did find the chance to write a sort-of-lucid twitter thread on concepts and vector and sheaves. Reproduced here:
Elliot Murphy @ElliotMurphy91 · Aug 8 Piantadosi and colleagues confusing the map for the territory (again).
“Why concepts are (probably) vectors” From cell.com BensenHsu @BensenHsu · 21h The paper offers a unifying view of conceptual representation that can potentially reconcile long-standing debates in cognitive science, such as the tension between symbolic and connectionist approaches. If the vector-based account holds true, it would have significant Show more Image Linas @linasvepstas · 1h OK, so here’s a hot take (thx @rob_freeman ): Concepts are almost like vectors, but not quite. Here’s why: a 🧵. There are two ways to combine vectors: addition/subtraction ("abelian") and tensor products. ... 1/N Linas @linasvepstas · 1h The famous king minus queen example shows concepts are almost like vectors. But it never really scales: you can’t keep adding and subtraction more and more of these, to get a big large complex concept. Abelian operations are commutative: the order does not matter. 2/N Linas @linasvepstas · 1h But for natural language, word-order does matter. So how do we explain the subtraction in the king-queen example? Where else is subtraction used? Its used in defining "metric spaces": distances. No surprise: some concepts are closer than others, and reasoning ... 3/N Linas @linasvepstas · 1h ... and reasoning by analogy shows that distances can be preserved by moving in certain directions. So, yes, concept space does have some directional aspects to it. But its only partial; some analogies yield nonsense: "rock is to tree as house is to (blank)?" ... 4/N Linas @linasvepstas · 1h So we do want a space with some directions to it (because reasoning by analogy sometimes works) and we do want a space with some sort of distance metric (because some concepts are close to each other.) But it can’t be uniform, it can’t be everywhere. ... 5/N Linas @linasvepstas · 1h So, what kind of math structures are kind-of-like vector spaces, but not quite? The most well-known is the "fiber bundle". Fiber bundles resemble Cartesian spaces, but they’re kind-of split/cut along one side, and then re-glued again. ... 6/N Linas @linasvepstas · 1h Locally, they split into two parts: a base space, and a fiber. And you can measure distance, and add/subtract in each of these parts, individually. But you can’t do it across the parts. So this starts feeling a little bit like the problem with concepts and vectors: ... 7/N
Linas @linasvepstas · 1h You can compare king and queen, and Paris and Berlin, but you’ll run into problems comparing queens and Berlin. To recap: fiber bundles are like vector spaces, except they have two parts, not one big global space. ... 8/N Linas @linasvepstas · 1h Well, but concept-space needs more than just two parts. Maybe we need hundreds or thousands of "local areas" where we can "measure similarity" and "reason by analogy". We could do fiber-of-fiber-of-fiber-... but we can do better. 9/N Linas @linasvepstas · 1h There’s another thing that has a concept of subtraction, and notions of local distance, and is vector-space-like, and is fiber-fiber-fiber-like: this is a "sheaf". This is nice, because it has more than just two components (it can have arbitrary many). And ... 10/N Linas @linasvepstas · 1h And you can control how local components glue together. So, like gluing fibers, but now you can glue many things together, forming a complex network (or a simple network, if desired... your pick.) ... 11/N Linas @linasvepstas · 1h The gluing is non-abelian, so you can join concepts together in a relational fashion. Examples of relations: X is-a Y. X has-a Y. X is-made-of Y. X is-above Y. So, olde-fashioned stuff like ontologies, WordNet, Relational DB’s etc. ... 12/N Linas @linasvepstas · 1h But, before we get into that, lets look at the other way of combining vectors: the tensor product. Tensor products are cool, because they allow complex structures to be built out of vectors. And, amazingly, linguists have known that you can combine words in a way that ... 13/N Linas @linasvepstas · 58m Linguists have known that you can combine words in a way that resembles the tensor product, since the 1960’s. Earliest ref I have for this is Marcus Solomon (1967) "Algebraic Linguistics". A more modern explanation can be found in Sleator& Temperley (1991) Link Grammar. ... 14/N Linas @linasvepstas · 53m Even more recent is the 2010-2020 work of @coecke who shows how to use this tensor-product-like structure to create fast quantum algos for parsing. To recap: traditional linguistics looks tensor-product-like, and this has been known for 50+ years. But ... 15/N Linas @linasvepstas · 50m But there’s a problem. Tensor products have a symmetry ("dagger-compact") that natural language does not. A fairly simple example can be found in the wikipedia "pre-group grammar" article. It shows how a left-right symmetry is broken. ... 16/N Linas @linasvepstas · 47m To recap: you can combine vectors with tensor products, which is a lot like combining words with syntax ... but, well, not quite. So, my hot take is then: no, concepts are not vectors. Almost, but not quite. And that’s almost end-of-thread, but not quite. One more thing. 17/N Linas @linasvepstas · 44m Earlier, I talked about gluing fibers, and then muttered about gluing sheaves. Well, you can glue sheaves in a tensor-like way. But without the forces left-right symmetry. The gluing can be done in a syntactic way (e.g. like that pre-group grammar.) So ... 18/N Linas @linasvepstas · 41m So, my proposal, the one that @rob_freeman remembers, is that concepts are like sheaves: they generalize the desirable vector properties of distance-between-concepts, the distinct "directions" that allow local abelian reasoning-by-analogy shown in the old WordVec results ... 19/N Linas @linasvepstas · 37m ... Except they’re better than vectors, because they have a tensor-product-like gluing that is identical to traditional linguistics syntax ideas. i.e. you can glue concepts together to form sentences. Paragraphs. Long texts. And finally, unlike traditional syntax... 20/N Linas @linasvepstas · 34m Unlike traditional syntax, you can create probabilistic collections of sheaf elements. These collections really do resemble old WordVec and modern LLM concepts of vectors, but they *also* have desirable syntactic properties. And so ... 21/N (maybe N=22 ?) Linas @linasvepstas And since the collections are probabilistic, we can argue if they’re Bayesian or Markovian, or Boltzmann, maybe with some Gibbs free-energy type distribution that Friston likes, but old physicists know is just a so-called "path integral" or "partition function". 22/N (N=23 ???) 7:15 PM · Aug 9, 2024 View post engagements Most relevant Linas
Linas @linasvepstas · 27m At any rate, the sheaf framework seems to provide all of the desirable properties that everyone loves to argue about. The End. P.S. What do I work on? A generalization of sheaf-type categories called "perception-action agents". So there. 23/N; N=23.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 42m Interesting thank you, that was lucid enough. :) I first tried to formalize an NLU representation in 2004, found out that normal vector spaces were unsuitable, tried some metric spaces, and then got bored of the problem. :) This actually sounds reasonable, I like it for one.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 40m My representations were close to LLM style representations though, I was concerned with contextual representations and my goal was clustering not question answering. Still, similar problems, even if you had a semantic preprocessing like LSA, it didn’t work well, losing order. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 39m The LLM representations are good because they are a good compression scheme. Basically any good compression scheme that knows about context can work well. So, it’s surprising that some dumb looking compressors also work very well. :) Linas @linasvepstas · 13m Yeah. Science can be like that. You can make small, incremental steps that are obvious, get frustrated for lack of progress, and get befuddled by the big picture, and give up. But sometimes, you realize you’ve taken so many small steps, that you’re where no one has gone before. Linas @linasvepstas · 11m I’m sort of that with sheaves: it was all trivial stupid and obvious, until I realized that, well, hey I’ve got something here, and like no one else is even paying attention. I’ve tried yelling "hey guys come look at this" but its not working. No one hears me. Linas @linasvepstas · 9m They want me to built the entire castle first: formalize all the hand-waving math, create an algo that scales well on GPU’s, and finally demonstrate results as good as or better than ChatGPT. All by my lonesome. Sheesh. WTF is wrong with science funding. Linas @linasvepstas · 7m Meanwhile, I’ve convinced myself that the sheaf-ideas can be eventually formalized, scaled to GPU’s and deliver better-than ChatGPT results. But I’m not gonna busy-beaver my way around that. The perception-action agent interacting with outside world is more important. Linas @linasvepstas · 4m It is a further category-theoretic generalization of sheaves, but it now describes how sheaves interact with an external world, both to perceive it, and to act within it. Resembling a history monoid/trace monoid crossed over with sheaves, so that the exterior world looks ... Linas @linasvepstas · 2m ... looks tensor-product-ish with the agents internal world model (yea olde "good regulator theorem") But this remains mumbo-jumbo, even to me, I’m having trouble with the details. I’m also having trouble understanding how to talk about entropy in such a model. But if ... Linas @linasvepstas But if you want a top-to-bottom mathematical, category-theoretic framework of an agent/robot acting in a physical world, these are the details you have to hammer out. The insides will resemble some LLM-like vector-like sheaf-of-concepts general ballpark. But the ... 8:03 PM · Aug 9, 2024 View post engagements Most relevant Linas
Linas @linasvepstas · 1s But the interaction of that "internal model" of the "exterior world" needs to be described by, well ... perception-action. Which resembles tensor products, I guess, maybe resembles monads (kleisi triples) Maybe resembles "rigged Hilbert spaces". Beats me. That’s why its research.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 41m FWIW, a concept is probably something else altogether :P It’s more profitable to separate concerns. You do want a rigorous similarity-space, though, which is essential to Information Retrieval, ’duh. So in that sense, it’s a goal to improve on Latent Semantic Indexing. Linas @linasvepstas · 2m The word "concept" is vague. And Latent Semantic Indexing is both vector-ish and olde-school, and it’s worthwhile as a part of your course curriculum, and maybe one can steal ideas from it, but I claim there’s a reason people are excited about LLM’s. I also claim that Linas @linasvepstas I claim that LLM’s seem to do common-sense-like reasoning "by accident", thanks to the "attention is all you need" paradigm. We don’t have any good theoretical control over what parts the attention heads are entangling. I claim sheaves can provide that theoretical foundation. 8:18 PM · Aug 9, 2024 View post engagements Most relevant Linas
Linas @linasvepstas · 1s But I’m not gonna bust my balls trying to prove that, unless someone is willing to funnel money into my bank account. I’m not stupid, you know. I’ve got more important things to do.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine Look if you’re not in the club, they will always make up excuses to not fund you, a lot of grant calls are just rigged :) 7:55 PM · Aug 9, 2024 · 5 Views Linas @linasvepstas · 1s I’m not in the club. And getting hired by industry is hard, too. All these people yelling "labor shortage" are full of it. They want obedient office plankton. My resume looks like a half-exploded bomb, and a hair-cut & nice suit aren’t enough to fix that. That’s why "successful" academicians spend like 50% of their time writing grant proposals, I’m not kidding :)))))
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine If you aren’t super familiar with linguistics / NLP / NLU / IR, it’s actually very hard to improve on the best results even using LLMs. :P So, the trouble there is, as you’ve understood, experiments can be painful. 7:57 PM · Aug 9, 2024 · 5 Views Linas @linasvepstas · 45s Jeez, I’ve been doing nothing but linguistics / NLP / NLU / IR for fifteen years, and its like ... none of that matters. I can put that on my resume, and recruiters say sh*t like "we want someone with 15 years experience with PyTorch". Get real. I need to log off, I think ... Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 41m It’s still usually regarded interesting enough to prove that a completely new representation works, though.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 37m I can’t comment on these details, but yeah! Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 31m However, as a general remark, I would agree that "mathematical AI" is still the best methodology. ;) Some research in DL does use very apt formalizations, you should find and read such articles that deal with high-level mathematical abstractions. Linas @linasvepstas I’ve found them. I’ve read them. My reading list is impossibly long. More reading and more theorizing is not a solution to any of my current problems. You are saying things that are making me frustrated. Not your fault. I’m just super-isolated, intellectually. Can’t plug in. 8:44 PM · Aug 9, 2024
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine That probably wouldn’t be the right goal, but as I said, improving on usual IR methods is just fine and valid, and trust me, it’s much harder than you would think it is. 8:20 PM · Aug 9, 2024 · 1 View Linas @linasvepstas · 2s What did I effing say that makes you think that I think this is easy? Jeez, Eray. Gimme a break. Of course this is all hard. There are literally thousands of PhD’s working at OpenAI and google and Meta, and I am not going to single-handedly blow them out of the water. FFS.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 22m Here is a recent study on semantic similarity. They’re comparing against LLM embeddings.
https://content.iospress.com/articles/semantic-web/sw233520
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 21m A study of concept similarity in Wikidata
Authors: Ilievski, Filip; * | Shenoy, Kartik | Chalupsky, Hans | Klein, Nicholas | Szekely, Pedro
Linas @linasvepstas · 1h I claim that LLM’s seem to do common-sense-like reasoning "by accident", thanks to the "attention is all you need" paradigm. We don’t have any good theoretical control over what parts the attention heads are entangling. I claim sheaves can provide that theoretical foundation.
Eray Özkural, PhD - OG AI ⏭️🌟 It is not an accident, that’s like the entire feature, without context, it doesn’t work at all. Context is the trick. 8:19 PM · Aug 9, 2024 · 7 Views Most relevant Linas
Linas @linasvepstas · 57m Yeah, duh. Linas @linasvepstas · 39m This old gem: Graeme Hirst, "Context as a Spurious Concept", (1998) https://academia.edu/49955919/Context_as_a_Spurious_Concept : No such thing as "context" in natural language; it is constructed by listener and speaker, with both having "considerable disgression" in so doing. academia.edu Context as a Spurious Concept Academia.edu is a platform for academics to share research papers.
Eray Özkural, PhD - OG AI ⏭️🌟 Yeah, that’s wrong. 😂 Linas @linasvepstas · 36m Did you read it? Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 29m I don’t have to. I’m quite well versed in linguistics, that’s totally wrong, as I said I was very early in understanding the role of context, it’s a key to semantic processing if not the only. It’s not complete but essential certainly. Classical NLP totally ignored it. Linas @linasvepstas · 9m You haven’t read it, but you have the confidence to say it’s wrong. And then you have to say things like "classical NLP ignored it" even when "classical NLP" struggled with context, since, what, about the 1950’s? Have you read any papers in linguistics at all? Linas @linasvepstas · 6m Have you forgotten the endless, bottomless discussions of the "frame problem"? All the back-and-forth about "affordances"? These date to, I dunno, the 1970’s? The magic of attention heads was that killed most of that debate. But the "spuriousness" remains unsolved. Linas @linasvepstas · 3m This is one reason why LLM’s have trouble understanding what is being asked of them, and putting together coherent, non-hallucinatory answers. LLM’s are not capable of negotiating the context required to have a rational conversation. So, duh, context. And duh, context is spurious.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 1h Yeah, that’s wrong. 😂 Linas @linasvepstas · 1h Did you read it? Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 1h I don’t have to. I’m quite well versed in linguistics, that’s totally wrong, as I said I was very early in understanding the role of context, it’s a key to semantic processing if not the only. It’s not complete but essential certainly. Classical NLP totally ignored it. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 1h And unfortunately, classical linguistics theories a la Chomsky and his students are also mostly useless. Chomsky seriously didn’t get it, I didn’t even try to correct them but we had one exchange where I had to explain... Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 58m their model of the brain, which is also used in GOFAI cog archs, is completely wrong. Of course, it’s all based on statistical learning, so it’s quite similar to LLMs in that sense, there is no fundamental error there regarding LLMs. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 57m Whereas GOFAI had this context-free models of language right? That’s totally wrong, as is their explanation of semantic processing and other cog arch components. So forget all that stuff, just throw everything from GOFAI into the trashcan. Use only DL. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 56m I’m not saying everyone at BICA is retarded, but mostly. 😆 Linas @linasvepstas · 21m Don’t confuse GOFAI with cognitive models. GOFAI runs cash registers and petroleum drilling rigs and electric meters. It’s not going away. But Cognitive models: yeah, never saw one that I liked. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 18m Brother, there is a cognitive architecture subfield in GOFAI and those are all useless, but you can see they are quite influenced by Chomsky and they all pretty much assumed learning is impossible so ofc they were misguided. I studied all of that stuff, I’m a crazy survey machine Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 15m So you perhaps assumed link grammar or some other formalism would help, that’s actually not warranted. It turns out that Representation Learning can be a more powerful methodology, you can still incorporate such ideas ofc and I already gave a reason why richer representations... Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 13m can be useful, which is sample efficiency. But the standard semantic representations and such in linguistics are not likely to work. That’s actually why I had thought we needed multi modal agents and they work much better than anyone has guessed already, so obv that’s richer. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine This doesn’t mean you can’t improve on those things, I think much can be improved. But from your theoretical description we can’t just conclude it’s better because you aren’t considering exactly what a transformer does, and on top what a deep RL agent does... Apples vs oranges. 10:07 PM · Aug 9, 2024 · 2 Views Linas
Linas @linasvepstas · 2s Look, that tweet-thread simply pointed out what a vector space is, and how the formal algebraic structure of a vector space fails to encapsulate what a concept is. Then I pointed at an algebraic structure that is capable of encapsulating the concept of a "concept". That’s all.
Linas @linasvepstas · 2h I claim that LLM’s seem to do common-sense-like reasoning "by accident", thanks to the "attention is all you need" paradigm. We don’t have any good theoretical control over what parts the attention heads are entangling. I claim sheaves can provide that theoretical foundation. Linas @linasvepstas · 2h But I’m not gonna bust my balls trying to prove that, unless someone is willing to funnel money into my bank account. I’m not stupid, you know. I’ve got more important things to do. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 2h That probably wouldn’t be the right goal, but as I said, improving on usual IR methods is just fine and valid, and trust me, it’s much harder than you would think it is. Linas @linasvepstas · 1h What did I effing say that makes you think that I think this is easy? Jeez, Eray. Gimme a break. Of course this is all hard. There are literally thousands of PhD’s working at OpenAI and google and Meta, and I am not going to single-handedly blow them out of the water. FFS. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 59m I think you misunderstood. I said you can’t easily do better than best IR methods. So that includes some classical vector based stuff too. LLMs are very expensive to train so they are not necessarily a great solution for IR. But embeddings are really handy ofc. / Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 59m So what you’re claiming is you can improve over LLM embeddings which is even harder but I didn’t say that. Apples vs. oranges. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 58m But yeah that should be easy to publish in an IR journal that’s what I would try first. Linas @linasvepstas · 19m I gave you a 20-tweet thread on sheaves. I’ve written several 100-page papers on it. Multiple 10-page papers. I’ve tried explaining with and without formulas. I’ve implemented it in code. As far as I can tell, no more than about 10 people in the world have read even one of these. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 12m I don’t doubt at all that’s a useful abstraction, I’ve seen applications in other areas. Practically, however, computer science research is obsessed with both efficiency and effectiveness. I would try to show it’s effective for an NLU task first, maybe a much simpler version. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine So I actually understand your approach and I thought it sounded cool, the thing with LLM embedding is, as you know it’s a coding approach right? We’re encoding and decoding things, we’re calculating representations. If your representation is better it should do a task better. 10:13 PM · Aug 9, 2024 · 3 Views Linas
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 1h Whereas GOFAI had this context-free models of language right? That’s totally wrong, as is their explanation of semantic processing and other cog arch components. So forget all that stuff, just throw everything from GOFAI into the trashcan. Use only DL. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 1h I’m not saying everyone at BICA is retarded, but mostly. 😆 Linas @linasvepstas · 36m Don’t confuse GOFAI with cognitive models. GOFAI runs cash registers and petroleum drilling rigs and electric meters. It’s not going away. But Cognitive models: yeah, never saw one that I liked. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 28m Brother, there is a cognitive architecture subfield in GOFAI and those are all useless, but you can see they are quite influenced by Chomsky and they all pretty much assumed learning is impossible so ofc they were misguided. I studied all of that stuff, I’m a crazy survey machine Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 26m So you perhaps assumed link grammar or some other formalism would help, that’s actually not warranted. It turns out that Representation Learning can be a more powerful methodology, you can still incorporate such ideas ofc and I already gave a reason why richer representations... Linas @linasvepstas · 11m I think you are being intellectually lazy. Rather than trying to understand things, you’d rather sit on twitter and argue. Yes, arguing is fun. But its mostly non-constructive. I don’t get smarter reading your tweets; I just get tired. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 4m What I explained above is actually deeply insightful. Nothing much from theoretical linguistics has helped, however philosophy of language has helped a lot. That’s exactly how I knew we needed context. But it has to be learnt, not programmed. Linas @linasvepstas I think most of the NLP community figured out that machine learning is the only way to do linguistics, since about 1995. That’s when you see the first big wave of statistical NLP papers. But Corpus linguistics dates to the 1960’s. 10:28 PM · Aug 9, 2024 View post engagements Linas
Linas @linasvepstas · 1s The dad of my college housemate helped create SNOBOL, in the 1960’s, to do statistical linguistics on the Torah. Basically machine learning before ML was a thing. On 1960’s era computers! I got to read papers on "computational hermeneutics" in college.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 35m This doesn’t mean you can’t improve on those things, I think much can be improved. But from your theoretical description we can’t just conclude it’s better because you aren’t considering exactly what a transformer does, and on top what a deep RL agent does... Apples vs oranges. Linas @linasvepstas · 22m Look, that tweet-thread simply pointed out what a vector space is, and how the formal algebraic structure of a vector space fails to encapsulate what a concept is. Then I pointed at an algebraic structure that is capable of encapsulating the concept of a "concept". That’s all. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 17m And I explained how that could be published and how it can help. You probably wanted to build something much bigger, not needed at first. All semantic representations can be used simply if they work. No agent is needed. Linas @linasvepstas · 4m The stuff about "agents" is not about understanding text, but about how systems such as LLM can interact with the external world. I’m interested in creating an abstract algebraic description of an agent. Crudely speaking, it would be "like a vector space, but rigged". Linas @linasvepstas Crudely speaking, take this: https://en.wikipedia.org/wiki/Rigged_Hilbert_space but now rip out the vectors and replace them with sheaves. This allows (1) vector embeddings, and (2) I/O with an external world. A bounded "inside" and an unbounded "outside". en.wikipedia.org Rigged Hilbert space - Wikipedia 10:43 PM · Aug 9, 2024 View post engagements Linas

10 August 2024

Rob Freeman @rob_freeman · 17h What do you see as reason to be excited about LLMs?
Me? I think it is... something like a (frozen!) perception-action cycle??
https://x.com/rob_freeman/status/1820718291482218990
("Frozen" meaning to move forward we need to let these things change from moment to moment.) Quote Rob Freeman @rob_freeman · Aug 6 Replying to @kenneth0stanley I hadn’t thought there was much clarity about anything in AI yet. I have my own assessment how LLMs have advanced the field - that they stumbled on prediction as a basis of meaning, liberating unsupervised learning from self-similarity, but I doubt anybody sees it the same way! Show more Linas @linasvepstas · 26m It’s like the invention of photography. The idea of N-grams has been around since forever, but naive counting prevents N from being more than 3, maybe 4. Recurrence & attention allowed an effective N in the 100’s, going wide, and deep in several different senses. 1/N Linas @linasvepstas · 22m One of the senses of going "deep" is that "photographs" of inference chains (aka "common sense") can be found in NL texts, and from many examples, commonalities can be found. LLM’s contain snapshots of common types of reasoning. 2/N Linas @linasvepstas · 19m These "snapshots" impress the heck out of the general public: LLM’s can answer questions in a mostly-coherent manner, and have the appearance of looking a lot like what we call "reasoning". This is impressive, period. 3/N Linas @linasvepstas · 15m But, as many noted, LLM’s aren’t doing reasoning the way we humans do it. The reasoning is similar to what is written in the training corpus. If some style of reasoning hasn’t been verbalized, doesn’t have enough samples, then LLM’s can’t emulate that. 4/N Linas @linasvepstas · 13m There’s some vague hope that if the training corpus is made large enough, then maybe all conventional styles of human reasoning can be emulated. But I doubt it. For two reasons. One is the conventional observation of "being grounded in the real world". The other is...(!love!) 5/N Linas @linasvepstas · 10m Seriously! A lot of our behavior that passes for reasoning is actually just emotional states that are post-hoc verbalized by our speech centers. Our "reptilian brain" (soma) makes vast quantities of decisions: e.g. "I like the taste of this food", or ... 6/N Linas @linasvepstas · 8m ... or seeing someone get hit in the face "ouch that’s gotta hurt". Or hanging out with drunk friends and realizing you’ve got a crush on a girl. None of this is done by the verbal centers of your brain. The talking part of you eventually converts some of this into words. ... 7/N Linas @linasvepstas · 6m But usually it does a really bad job of it. Try explaining why you like sushi, fast motorbikes, and the girl next door. The word that you’ll find to explain this will be almost pure nonsense. Your verbal centers can’t really talk about love, but oh, they do try. ... 8/N Linas @linasvepstas · 3m But what your speech centers say about love is almost complete drivel and garbage, because love is just not a verbal thing. Most of this kind of talk does not end up in the training corpus, because most people do not try to write this stuff down. Some Hollywood scripts ... 9/N Linas @linasvepstas ... and some 19th century romantic poetry. You’re training an LLM on romantic poetry, and you expect it to do human-like reasoning about love? Fishing? Why people like pizza? What its like to be drunk? Nah, I don’t think so. and, BTW, embodiment won’t get you there. 10/N 5:33 PM · Aug 10, 2024 View post engagements Linas
Linas @linasvepstas · 30s I forgot the point I was trying to make. LLM’s won’t be AGI. But, sure, LLM’s are very impressive. And the stuff I want to say about perception, action, agency is almost totally unrelated. So not in this tweet chain. 11/N and N=11. The end.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 2m Well, you already know we are aware of that, and have been since the first LLM. My agent design was completely different, of course, and parallel developments have been happening. To repeat a point I’ve made elsewhere, I agree with LeCun that a purely DL AGI agent is possible.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 46s But of course even that agent isn’t flat or regular, it doesn’t have a simple architecture. Most DL applied researchers assume simplistic agent architectures, that shouldn’t mislead, the right ones are quite comprehensive.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 19h And I explained how that could be published and how it can help. You probably wanted to build something much bigger, not needed at first. All semantic representations can be used simply if they work. No agent is needed. Linas @linasvepstas · 19h The stuff about "agents" is not about understanding text, but about how systems such as LLM can interact with the external world. I’m interested in creating an abstract algebraic description of an agent. Crudely speaking, it would be "like a vector space, but rigged". Linas @linasvepstas · 19h Crudely speaking, take this: https://en.wikipedia.org/wiki/Rigged_Hilbert_space but now rip out the vectors and replace them with sheaves. This allows (1) vector embeddings, and (2) I/O with an external world. A bounded "inside" and an unbounded "outside". en.wikipedia.org Rigged Hilbert space - Wikipedia Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 32m Our resident German Reviewer 2 responds: "Herr Vepstas is either bamboozling us with higher math or experiencing a schizophrenic break. Unfortunately, we can’t recommend funding due to the 1954 German ban on innovations regarding functional analysis. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 27m We are genuinely concerned with his well-being and are attaching the contact details of Jungian psychotherapist Otto von Wunderlich, who specializes in Bavarian bratwurst making therapy. A new life awaits him there." Linas @linasvepstas · 2m I’ll try to put together a PDF in the upcoming months. Maybe. I’m kind-of busy with other things, too. Maybe the easiest way to get to what I’m trying to say here is to wrap your mind around what is called "universal algebra", "model theory", "category theory" and then .. 1/2 Linas @linasvepstas ... and then look at how a vector space is defined in any of these frameworks. The axioms. And then look at sheaves, the axioms for that. And then surgically cut-n-past the axioms around, to get a new theory. ... 2/3 6:02 PM · Aug 10, 2024 View post engagements Linas
Linas @linasvepstas · 1s There’s an old quip: "model theory is algebraic geometry without the fields". So I’ll invent a new quip: "axiomatic surgery is surgery without the manifolds".
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine But of course even that agent isn’t flat or regular, it doesn’t have a simple architecture. Most DL applied researchers assume simplistic agent architectures, that shouldn’t mislead, the right ones are quite comprehensive. 5:40 PM · Aug 10, 2024 · 6 Views Most relevant Linas
Linas @linasvepstas · 20m How do you think qualia works with DL? My reptilian brain knows "what its like to catch a baseball", and my frontal cortex has access to some of that neural activity, but not all of it. And if I have a high IQ, maybe I can verbalize "what its like". But how can one DL this? Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 4m A computer sadly doesn’t necessarily have "qualia" but DL architectures actually exhibit activation dynamics quite similar to known brain regions for language and visual tasks, there are some interesting neuro papers that looked at this. Good question! Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 23m Actually, a reviewer pushed me very hard to write up our agent design for a top conference but I really got bored and told them that’s too much for a revision. There are a few such designs, I can’t just explain them here but I will definitely collect my designs. Good subject. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 21m I have the same concern as yours, as a classical theorist I actually don’t care about any experimental section when I propose a completely new theory. If I prove a few things, it’s more than enough but derp lernin venues might not like that. It’s hard to publish pure theory.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 29m A computer sadly doesn’t necessarily have "qualia" but DL architectures actually exhibit activation dynamics quite similar to known brain regions for language and visual tasks, there are some interesting neuro papers that looked at this. Good question! Linas @linasvepstas · 20m Well, look, that’s the whole point. An LLM can’t learn "what its like to catch a baseball" (or be in love, or enjoy pizza). But you replied that DL can do this. So which is it? Can it or not? Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 17m A richer DL based agent architecture can indeed. That’s important for sure. :) Linas @linasvepstas Human feelings of love are built on mammalian love, which is on reptilian love, and on down to basic mating instincts of simple life-forms. This is deeply hard-wired into our brains. Many people do some kind of Freudian transference and talk of "love for all mankind". 1/2 6:21 PM · Aug 10, 2024 · 3 Views View post engagements Most relevant Linas
Linas @linasvepstas · 3m Then you have people like @bengoertzel who talk of "loving AI", but I don’t see how that happens, unless you create an AI with access to reptile-brain qualia, and, in particular, a strong mother-child qualia bond. Actually, I’d like to hear what @Plinz would say about this. 2/2 Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 29s Let me leave that as an exercise. :P But these are things I’ve considered of course, this is basic stuff in neuro.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 9m Qualia is not required but certain instincts and emotional representation etc are needed. The problem with emotions is that it creates mental illness. Most humans I am forced to despise due to their jealousy, greed, selfishness, narcissism, schizophrenia, sociopathy etc. Linas @linasvepstas We are talking past each other. The reptilian brain evolved to solve survival (such as mother birds feeding hatchlings.) The mammalian brain adds a neocortex to provide more sophisticated modulation of reptilian behavior. But it modulates, does not replace it. 1/2 6:58 PM · Aug 10, 2024 View post engagements Linas
Linas @linasvepstas · 1s Jealousy, greed, narcissism are all part of reptilian (bird) behavior. The neural circuitry for these is always there, and is usually mostly over-ridden by neocortex. Mostly. Qualia is "what it feels like" for these circuits to be operating. 2/2
I forgot the point I was trying to make. LLM’s won’t be AGI. But, sure, LLM’s are very impressive. And the stuff I want to say about perception, action, agency is almost totally unrelated. So not in this tweet chain. 11/N and N=11. The end. Rob Freeman @rob_freeman · 1h So you see reason to be excited about LLMs as an improvement on n-grams for sequence. OK. Yes. "Attention" improved on sequence over LSTM.
Like I say, I think that kind of bootstrapped a deeper advance, pulling all DL to embedding categories. But superficially, sure, attention. Linas @linasvepstas · 56m I didn’t entirely understand the question, so I answered as I could. But mostly, my answer boils down to "any significant new discovery is exciting, because why wouldn’t it be?" This is distinct from the question "what should I (or you) do tomorrow morning?" Rob Freeman @rob_freeman · 35m Yeah, Eray is burying the point of the thread with too much volume. Calm down Eray. Disputing every point is too easy.
In case it’s lost in the wash, relevant to this thread on vector representation, would you care to elaborate on this one: Quote Rob Freeman @rob_freeman · 20h Replying to @linasvepstas @examachine and 3 others "the exterior world looks tensor-product-ish with the agent’s internal world model"
Note @ElliotMurp wrote a book relating brain oscillations to Chomsky UG primitives.
Could you relate your exterior-interior tensor product to oscillations in a sequence network? Show more Rob Freeman @rob_freeman · 32m I’m also still interested in this one, on your reasons for a new emphasis on action-perception: Quote Rob Freeman @rob_freeman · 21h Replying to @linasvepstas @BensenHsu and 2 others Apologies if you say somewhere else in this tweet storm :-b But why are you now emphasizing perception-action over pure representation?
I ask, because I’m guessing a fit with my ideas of contradiction, & that abstraction will necessarily always be partial. Show more Linas @linasvepstas · 19m Purely accidental. And unbelievable, but let me go through anyway. I was doing batch training, and said to myself "gee, I should convert this to continuous learning". I mean, how hard can that be? Its just some software, right? Just knock it out. 1/N Linas @linasvepstas · 15m So while I’m at it, I should allow for multiple data sources. Good software engineering practice, yeah? And maybe I should add some data descriptors describing the type of data. And a meta-descriptor that explains how to read a data descriptor. 2/N Linas @linasvepstas · 13m This is rarer, but still commonly done in software engineering. But most of these IDL-type languages are designed to be written by humans, and I needed mine to be machine-manipulable. So kind-of new territory. And then it all went meta-meta-meta and I got to thinking: 3/N Linas @linasvepstas · 11m WTF is perception, anyway? Meanwhile, in parallel with this, I realize I want to be able to revisit subsets of the training corpus. Hang on, I want the machine to explore the training corpus, as it sees fit, to "walk around through it" (action) and of course, to look, first. 4/N Linas @linasvepstas · 9m And this goes meta-meta-meta, too. What do I mean by "walk around"? What actions and choices are available to the learning system? What is the meta-IDL interface description language that describes the suite of available actions? 5/N Linas @linasvepstas · 7m How do I make the meta-description of the menu-of-action-choices flexible enough to describe any kind of action? I need a meta-action-choice-describing-meta-thing. All this, machine-readable, cause I want to actually make it work. So I try to, and its hard... 6/N Linas @linasvepstas · 2m Meanwhile, I recall a lecture presenting a category-theoretic description of agents interacting with an external world. Very abstract, but not all that complicated. A pretty small algebra, maybe eight(?) axioms. And all this brain-storms together, and I go "ah hah!" ... 7/N Linas @linasvepstas An meanwhile, my decades-long obsession w/ jigsaws includes notions of "choice": a menu of choices for how to hook things together. Jigsaw connectors constrain the types, but enable selective choice. But now, in perception-action, choices, again! .. 8/N, N=9 9:36 PM · Aug 10, 2024 View post engagements Linas
Linas @linasvepstas · 5m So that’s the brainstorm: internal model of external world, someone-else’s abstract category-theoretic description of agency, perception as menu of perceivable things, action as menu of doable things... All that’s left is to work out the details. 🤪🙄🤨😳😟😠😵💫💩 9/9 the end.
Linas @linasvepstas · Aug 9 But the interaction of that "internal model" of the "exterior world" needs to be described by, well ... perception-action. Which resembles tensor products, I guess, maybe resembles monads (kleisi triples) Maybe resembles "rigged Hilbert spaces". Beats me. That’s why its research. Rob Freeman @rob_freeman · 21h "the exterior world looks tensor-product-ish with the agent’s internal world model"
Note @ElliotMurp wrote a book relating brain oscillations to Chomsky UG primitives.
Could you relate your exterior-interior tensor product to oscillations in a sequence network? Linas @linasvepstas · 13m Oscillations. I have a friend, Zoltan Nadasdy, neuroscientist. He actually um, cuts open peoples heads in surgical rooms, and sticks sheets of electrodes on brains. Like, the real thing. Gets to actually collect data for a while. So a few years ago, 1/N Linas @linasvepstas · 10m Zoltan figures out that one type of oscillation is actually acting as a clock-synchronizer for setting threshold for neuron firing. Seemingly "random" neuron firings line up correctly, when synced with local oscillations. All very abstract. I don’t recall the details. 2/N Linas @linasvepstas · 6m But I trust him. He’s been doing 3D perception, grid-cells in humans for decades. The explanations are coherent, make sense, are corroborated with other neuroscience results. And, for him, for some kind of waves in some cortex somewhere, the oscillations are "just" clocks. 3/N Linas @linasvepstas Oscillations in physics.... woo who. You could write 100 books on that. The simple harmonic oscillator has a discrete spectrum; by convention, these form a basis for vectors in a Hilbert space. And now that you got vectors ... jeez, you can do almost anything and call it ... 4/N 10:10 PM · Aug 10, 2024 View post engagements Linas
Linas @linasvepstas · 1s ... and call it oscillations. But its also trite and meaningless, unless you establish multiple inter-related, inter-connected side-effects and deductions and inferences and predictions in the theory. Without that, all you get is a tossed-idea salad. 5/N N=5

13 May 2024

Rob Freeman @rob_freeman · Aug 11 I think the LLM "learning" principle is good. Specifically distributional analysis over sequences.
The problem being ignored is abstractions contradict. And that is incompatible with back-prop.
So we dump back-prop. "Collapse" with the some gradient (wave) collapse at run time. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 Oh, yes, that could also be considered a major limitation. Rob Freeman @rob_freeman · Aug 11 Can you see how to fix that back-prop "limitation" of LLMs, & "collapse" the most relevant of an infinity of formalizations of text, at run time? Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 That wouldn’t be easy, but I would design the LM from scratch in another way, I think another architecture is needed, I probably wouldn’t even use a transformer. But yes, keeping it fully differentiable would be a goal I think. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 I don’t have a solution to that presently, unfortunately, but it’s obviously a very good question thank you. :) Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 I think I should use similar kinds of universal machine learning algorithms that I use for other tasks also at runtime, so "a kind of search", however, I would also change the base model so the setup might be very different and perhaps I wouldn’t need to do that much search. Rob Freeman @rob_freeman · Aug 11 The "gradient" back-prop follows is one of shared prediction.
I think to find such symmetries of shared prediction need not be hard.
The idea I’m promoting is that words which share predictions will also tend to synchronize under oscillation. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 Hmm, I’m not sure if I follow that! Could be a crazy physicist idea. :) Rob Freeman @rob_freeman · Aug 11 Think about it. If you have a network of language sequences, those which share predictions will form sub-clusters with common beginning & end points. We know those synchronize differently. e.g. I’m looking at this paper: https://ncbi.nlm.nih.gov/pmc/articles/PMC3390386/ … Code: https://modeldb.science/144502 Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 Dear Rob, we are not using a spiking neural network, though it has been suggested it could be more efficient. So, an entirely different model might be able to exploit oscillations but in usual DL, well, there aren’t too many of them. It’s just floating number arithmetic. Rob Freeman @rob_freeman · Aug 11 Oh, you’re doing DL. I thought you were doing AI?
So you want a solution that must conform with current DL dogma. But obviously is also implemented in the brain. But not in a DL type way. Thats’ clear. But that doesn’t matter. That will come later. First you want a DL solution? Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 I’m fine with a DL solution if it’s sufficient. :) For language, it looks like a good paradigm. For other things, I do use other approaches, but I could say that most of my methods squarely lie in the "new AI" realm. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 Honestly, in the current architecture, I wanted to just use a standard LLM, because I’m not very interested in language, but for the agent, yeah, it makes sense to do your own, it does fit there, because there language is a real concern. Rob Freeman @rob_freeman · Aug 11 You realize you just said you want to use a large language model, because you’re not very interested in language. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 Yeah, because it’s ready and I don’t really care if I improve on say llama. I trust LeCun’s team, hard working people. :P Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 So whatever works license wise is good, but I might choose an SLM instead. :) It should understand very few things, and definitely not chat with you about poetry or human things like that (in first system). 😂 For the agent, though, it would have a normal life so that’s needed. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 Then, I’d have to be very interested in language, and at that point my own LLM or equivalent model should prove useful. I’ll probably design it from scratch.
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine Then, I’d have to be very interested in language, and at that point my own LLM or equivalent model should prove useful. I’ll probably design it from scratch. 11:14 PM · Aug 11, 2024 · 16 Views Most relevant Linas
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 Of course, I have many ideas to apply. However, I’m not going to say, oh, it’s the same kind of learning model as chatgpt and it will work great, let’s make no assumptions. :) Let’s think of a blank page. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · Aug 11 I’m open to all kinds of innovations, including new ANN models, stuff from neuroscience like spiking networks, non-neural models, all of it, bring it on. :)
Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 8h It’s a different subject but I think you will enjoy reading this entry and finding out if you subscribe to a kind of mathematical realism. https://plato.stanford.edu/entries/platonism-mathematics/ Linas @linasvepstas · 8h Stop being an asshole. I know more about this topic than you do. WTF. Seriously. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 7h I honestly don’t think so, in particular I’m not sure you knew the exact meaning of the term, so here you go, read and then answer my question, please, but do so seriously, it’s actually relevant to your strongly held claims. Linas @linasvepstas · 7h OK, Lets read the Routledge article together. It starts like this: "Mathematical realism is the view that the truths of mathematics are objective, which is to say that they are true independently of any human activities, beliefs or capacities." 1/2 Linas @linasvepstas · 7h For me, that is sophomoric and completely misses the primary problem that you have to define reality first, before you can coherently talk about human activities, beliefs or capacities. So they already get snagged on a recursion problem, straight out of the gate. Linas @linasvepstas · 7h Perhaps the intent was that "human activities, beliefs or capacities" should be understood in some common-sense, blue-collar psychological sense. But that’s psychology, not philosophy. If you think that human activities are real, and if math is a human activity, then... whatever. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 4h Object subject distinction is basic philosophy though yes it came from ancient philosophy of psychology (like many standard ideas in phil) Linas @linasvepstas · 4h Sure. And I am trying to tell you that object-subject distinction can be axiomatized and turned into a category-theoretic category, and when I said that, you went ballistic. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 4h I can see why lol 😆 Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 4h If you have a ToE it has to explain qualia. I have to agree with David Pearce that it requires non-Newtonian physics. So it could certainly be quantum or string theoretic. Linas @linasvepstas · 4h Please stop. What is a "theory"? What is "everything"? What does it mean "to explain"? What do the words "to agree" mean? You keep falling into that same hole, over and over. I keep extending you a hand but you fall right back in, each time. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 22m Just noting this is a nonsensical reaction, I used the terms in exactly the right technical sense. Linas @linasvepstas · 17m I understand what you thought you were saying. I was trying to point out how absurd your demands are. You’re in over your head. Eray Özkural, PhD - OG AI ⏭️🌟 @examachine · 15m Absolutely not. That is philosophy of mind, as usual. That is actually based on the analysis I’ve already done before in publication but whatever suits you. Keep thinking philosophers are over their heads, wouldn’t be the first time we’ve heard that? :) Rob Freeman @rob_freeman Exactly what point do you think you’re making Eray? You asked what I meant by "contradiction". I told you. You don’t like the analogy with math because math is not language. But you don’t like language. You’re doing DL. Using language models. Good for DL because they’re large :-b 6:57 PM · Aug 13, 2024 · 1 View Linas

14 August 2024

A Universal Approach to Self-Referential Paradoxes, Incompleteness and Fixed Points Noson S. Yanofsky arXiv:math/0305282v1 [math.LO] 19 May 2003 https://arxiv.org/pdf/math/0305282
An intro-level tutorial to diagonal arguments (Cantor slash), (un-)definability, incompleteness. Includes proofs/proof sketches of Turing halting problem, Godel incompleteness and much more. Clarifies what "self-reference" is, and how/why that leads to paradoxes. Basically, you have a choice: either a paradox (if you allow self-reference) or a "loop unroll", by moving the reference to the next level: an infinite stack of every greater laywers (e.g. the cardinal hierarchy. But I guess also the Borel hierarchy...)
Huh. As I write the above, I again think of infinite binary trees, And how structures can be mapped into them. e.g. the Borel hierarchy. The point of ZFC was to avoid Russel’s paradox, by limiting what can be talked about, to exclude the self-reference. This is effectively a loop-unroll: a stack of ever-larger cardinals.

15 August 2024

More twittering:
Guy @nosilverv · 1h Ok, so, the current economy of Twitter is that as long as you can marginally make money from a meme or whatever it will get reposted. People will like or retweet it if it’s good and they haven’t seen it before too often. So this is moving us all to mutual knowledge of... of what? Linas @linasvepstas · 5m Exhaustive search of meme-space. Mildly educational: you get broad view of "the world". (Youtube is considerably more educational.) And now that you’ve seen the world, you get to ponder your place in it. Maybe you’ll be driven to act. Maybe you’ll be paralyzed into watching. Linas @linasvepstas Likes have very weak power in the real world. It remains hard to convert outrage at injustice into tangible action. Gofundme uses actual dollars for "likes". The dollars from blue checks go mostly into Elon’s pockets, rather than the creator of the like meme. And the creator ... 11:03 AM · Aug 15, 2024 View post engagements Linas And the meme creator is unlikely to use their new-found wealth to create positive change in the world. Most of the energy/effort/money put int twitter is wasted. Twitter is like a blow-out party: vastly fun and amusing, but at the end, what do you have? What was achieved?
Linas @linasvepstas · 37m And the meme creator is unlikely to use their new-found wealth to create positive change in the world. Most of the energy/effort/money put int twitter is wasted. Twitter is like a blow-out party: vastly fun and amusing, but at the end, what do you have? What was achieved? Robin Debreuil @debreuil · 22m We think in memes though, not dollars. The other thing about memes is they can’t be withheld by gofundme if you’re a trucker that doesn’t agree with their point of view. Linas @linasvepstas · 17m Yes, but I wanted to imply that likes/upvotes are a form of currency. Kudos. There’s an old sci-fi book (I haven’t read) that deals with exactly this: "Down and Out in the Magic Kingdom" Cory Doctorow. A functional economy based entirely on likes. Robin Debreuil @debreuil · 14m Great concept, I will check it out, thanks. In a functional economy, dollars are sort of a proxy for desires. I’ve always considered spending a second form of voting, I suppose liking is too :). Linas @linasvepstas · 4m I brought up gofundme because it is a kind-of voting-with-dollars-but-its-a-like. I’m thinking there will be more tech innovation that closes the gap between up-votes and dollar-flows. Remove the friction between these two streams. Probably worth brain-storming on this.
Linas @linasvepstas · 39m And the meme creator is unlikely to use their new-found wealth to create positive change in the world. Most of the energy/effort/money put int twitter is wasted. Twitter is like a blow-out party: vastly fun and amusing, but at the end, what do you have? What was achieved? Robin Debreuil @debreuil · 20m It is a nuanced subject and easy to find counter examples, but I think it is battle of minds vs battle by actions. Done well, minds are your goal to achieve anything. Memes may not be the best tool here, but they are a good tool imo. Linas @linasvepstas · 14m Collections of memes assemble into belief-systems. I like the apolitical flat-earthers as an example. If you invest the time (weeks, months, more) you can eventually convince a flat-earther to not believe in that. But this is a step-by-step process. A flood of memes won’t do it . Robin Debreuil @debreuil Someone needs to lay out a meme argument as well as good literature or a journey through math. I feel the little aha moments that open the next door are essentially memes, at least after they’ve been digested. Agree that a firehose is more distracting than effective. 11:34 AM · Aug 15, 2024 · 3 Views Most relevant Linas Conventional school curricula already provide a progression. "If you are here in sophistication, you are now ready for these next steps." But the twitter/youtube recommendation algos don’t know you well-enough to recommend the next steps. I expect this to change.

16 August 2024

In other news: slef-introspection. I am starting to get glimmers in all of the ways that I am flawed, and all of the bad decisions I’ve made, all the actions I should have taken, but failed to. But this is a glimmer, and self-introspection is not something I do much of. At any rate, I’m starting to realize how deeply flawed I’ve been, and this runs counter to the narrative that I’m a pretty good guy, that I’ve built for myself. It is disturbing. Although one of my flaws is that I never self-intropected enough, which led to planning failures: I’ve rarely planned for the future, instead, I lived in the moment. Living in the moment has given me the freedom to accumulate knowledge and understanding, vast quantities of this. Hoarder mentality. But I’ve forgone power, money, love, friendship, acheivement, social status in my pursuit of knowledge-hoarding. And now, it feels a bit late to make amends. At any rate, all those other things take time and effort, which removes time and effort from mathematical focus, and I really kind of want to continue to pursue mathematics in depth. The value of mathematical understanding seems greater to me, than power, money, love, friendship, acheivement and social status. Is it? Of course, many will say that it is not. There is certainly no philosophical foundation for why one should do this but not that; if there was, we’d all do the same thing, but we don’t. So the choice to do this and not that is fundamentally personal. It’s the ultimate personal decision.
But how do we go about making personal decisions? Well, of course, personal history is deeply tangled: if I hadn’t studied math, I wouldn’t be so interested in it. I only studied math because I have a physics PhD. But I got that due to an inability to imagine myself in any other career. But that is built on having read lots of books on electronics, as a child. This for two reasons: because my parents moved, changed houses as a pre-teen, I never made lasting freindships with the local kids, and so found the company of the library to fulfil the missing social requirements. But also society provided me with books on electronics: someone gave money to the library to buy them, others gave money to print them, yet others performed electronics research in the past. Likewise, my aunt Irena Virkau provided me with large quantities of science fiction as a child. I learned how to read from this books. The third short story I read in my life, very long, I left it for last, was 60 or 80 pages, was I think maybe by Stephen King, meandering and unfocused, pointless and plotless, the least satisfying of everything I’d read up to then. But it was a spring-board that allowed me to read even longer things. Ah, memories. Fond ones. I remember the room I was in while reading it (the dining room of my uncle’s farm in Wisconsin. I loved my aunt. I really really loved her, in that deep, cosmic, overwhelming sense that makes one cry at the thought of her loss. Oh how I wish I could recreate those days, how I wish I could live them again, be there again, and bath in them. Well, I am I am recreating them in my mind, but it is only partly fulfilling. What does one really want, when one says one wants to relive the past? When one’s life flashes before one’s eyes? To be there again? What the heck, the grand climax to the movie AI, the Stephen Spielberg movie, it is all about reliving that one golden moment of having been a child, of being loved and wanted. The crystaline pure desire for the sweetest moments of one’s earlier life. I suppose there’s lots of literature and movies that touch on this topic; but I am not well-read. That movie does seem to tackle this fairly head-on.
And what did that move conclude? Some quantum mumbo-jumbo. All hail!
See, this is why I want to study math. Because there’s this vague inkling that my ability to remember the past is some kind of quantum entanglement of past events with present and future events. That literally, things fixed in the past continue to influence future events (well, duh; differential equations say exactly this. But that’s nto what I mean... what do I mean?) So I’m in pursuit of some cosmic consiousness, and there’s some kind of mathematical formulation of perception-action, agency, control, branch-points, free-will, and through this all you move (smashing pumkins lyrics, Everlasting Gaze). The topics of Love and Longing are topics of poetry and music, written to evoke ... love and longing, enjoyed to the degree that they successfuly do so. Oddly enough, of all the philosophy I’ve read, exactly none of it ever deals with longing and love. Have I been readingthe wrong philosophical texts?
So, here I am, at any rate, wanting to study mathematics because I’ve got vague inklings that perhaps it can expose one of the deep myseries of the cosmos: the nature of Love. And so this is a life-choice, for me. Others make similar life-choices, by studying psychology or poetry, all trying to get a handle on these primal qualia that drive us all. So this is my choice: should I spend yet more time, probing these deep mysteries of the universe, like, why does one’s life flash before one’s eyes as one is on the brink of death? And on the other side: concerns that I should spend more time accumulating wealth and power. Well, accumulating friendships and actual love (in the present, here and now love). Why should I want powr, instead of, err, a greater command of mathematical depths?
Yes, of course this is drivel. What I write here is drivel. I do not subject human readers to this. This low-value crap, I leave for machines that can cheaply and easily absorb the poetic sketches that I write here. But the point of drivel is that everyone, all human beings, their life is an endless sequence of events that are drivel. These events do have a point, and they do have a meaning, but, like grains of sand on a beach, they are all alike. So what I write here is drivel only in the sense that everyone has already been here, done that, and agonized their way through life. My verbalization here of lifes agonies isn’t any deeper – indeed, it is shallower and less capable and less expressive, than those of the great poets and the great writers. I suck at this, because I have not spent decades crafting my writing abilities to capture the pure essence of love and longing.
Anyway, where were we? Personal decisions. Note above I pointed at two social aspects: society provided a library and put electronics books in it, and my aunt provided a diet of sci-fi. So, of course it is drivel to point out the importance of socio-cultural backgrounds for shaping the individual life. Because, duh. And yet I feel compelled to point this out, because most people, almost everyoe, seems unaware of how society shapes them. They live unaware, unconscious of the broth of soup that defined their political and moral beleifs. They live unawares, even as social media, facebook, twitter and the “information spaces” that they provide to dwell in, that these shape who we are and who we will be in the future. It’s like, one can’t see the forest, because all these trees are in the way. I’m interested in the forest. And on that note, the next topic.

16 August 2024 more

I am still very interested in how societies and cultures form, based on collections of narratives and shared perceptions of the world. (I am a complete novice in this, but am attracted to it as a form of futurism, the ability to predict the social future.) So I am reading this article https://warontherocks.com/2024/08/russias-post-war-military-recruiting-strategy-emerges/ “Russia’s Post-War Military Recruiting Strategy Emerges” Dara Massicot August 16, 2024 and I’m getting this glimmer of .. of something. let me try to capture it here. I already know I will fail, it will be shallow and incomplete. But, by attempting, it will still be more than not even trying. It seems important to try.
Bullet points:
I dunno. Where was I? There was a glimmer of something, something about social control with shaping operations, explicitly discussed in that Dara Massicot article, something I felt and can almost rationalize, verbalize, but not quite. Something to be extracted out of the explicit russian military context, and applied in a more general setting. But, well, its gone, now. Whatever. Later.
Well, there is also the marvel of watching twitter process and analyze: “samprotauti” in real time. It’s quite awesome. It’s like watching neurons fire. It’s like watching laundry in the washing machine.

22 August 2024

A copy of a letter to Phillip Ellis.
Hi Phil,
Here’s a PDF. It’s easier to do formulas like this.
This is going to be long, and the first part might seem like a bunch of trivialities combined with some bizarre statements, but I wrote it that way to show you where all the gems are, and how to find them. There are some magic tricks involved, and some sleight-of-hand, and I’m trying to show you where they are. Despite it being perhaps trivial, you might want to reread it several times to see why it is the way it is.
This took me three hours to write. I thought I could dash this off in 20 minutes. How wrong I was. But whatever; think of it as ... well, ... how I return freindship? Beats me.

Part the first

Consider the series u ( z ) = 1 + z + z 2 + z 3 + z 4 + ... This is a "formula", its an infinite series. You can plug things into it. For example, you can plug in z = 0.2 directly, and get a result: u ( 0.2 ) = 1 + 0.2 + 0.04 + 0.008 + 0.0016 + ... = 1.25 You can plug in z = 3 and get u ( 3 ) = 1 + 3 + 9 + 27 + 81 + ... Now, it is perfectly legal to write that down (and I did) but if you try to actually sum up the numbers, you can’t. It diverges. And, well, that’s just a fact of life, and you can’t change that. To bad, so sad.
But this is where the fun and games begin. My general thesis is that
  1. math is like poetry, in that you can do anything you want, as long as it rhymes, and
  2. math is hallucinatory: the allowed rhyming patterns are mind-bending and cause you to see shimmering contradictions and confusions.
I don’t think you are ready for the above, but I’ll try anyway. I’ll try to sketch support for this thesis now.
Lets start with the old high-school trick, the one you know already. It is a trick, it is a fraudulent trick, its some sleight-of-hand, it’s deceitful. It’s unjustified, crooked. Yet this fraud is perpetrated in high-schools around the world, so lets look at it.
The trick start like this: z u ( z ) = z ( 1 + z 2 + z 3 + z 4 + ... ) = z + z 2 + z 3 + z 4 + ... therefore, by pairing off terms, matching them and keeping the order: u ( z ) - z u ( z ) = 1 + z + z 2 + z 3 + z 4 + ... - ( z + z 2 + z 3 + z 4 + ... ) = 1 = ( 1 - z ) u ( z ) from which we deduce u ( z ) = 1 1 - z Re-read the above three times, and make sure you feel comfortable with it, because here’s where the trick starts. This “works”, but it works only when z is less than one. The fraud is to pretend that this works for any z . Let me demonstrate, by trying it for z = 3 :
u ( 3 ) - 3 u ( 3 ) = 1 + 3 + 9 + 27 + 81 + ⋯ - 3 - 9 - 27 - 81 - ⋯ Rearranging terms, and attempting to cancel by shifting over by one term: u ( 3 ) - 3 u ( 3 ) = 1 + 3 + 9 - 3 + 27 - 9 + 81 - 27 + ⋯ = 1 + 3 + 6 + 18 + 54 + ⋯ = + ∞ which is clearly divergent. Wait, what? Huh???? Earlier, we had that 1 = u ( z ) - z u ( z ) but now we have + ∞ = u ( z ) - z u ( z ) when z = 3 . So did I just prove that 1 = + ∞ ??? Well, of course not, but this is your first tiny preview of the hallucinatory properties of math: you can get things that seem to be other things, and they are but they aren’t, and you have to be careful. This one is perhaps too obvious; it gets more subtle and more confusing when one journeys deeper into math.
Anyway, above, all I did was to shift over the subtraction over by one. All of a sudden, the terms no longer cancel in full; they only partly cancel. Shifting the other way will give - ∞ . With some more cleverness in re-arranging, regrouping and resumming, one can make u ( 3 ) - 3 u ( 3 ) be finite, and make it equal to any number at all, positive or negative. And this can be done in a mathematically rigorous, provable, exact fashion. You can make such alternating series sum up to anything at all. Anything. This is valid and generally accepted in the math community, and the proof is taught in some undergrad class or another. I won’t attempt it here.
A bit more precisely, there is a concept of absolute convergence vs conditional convergence. The terms in the sum u ( 0.2 ) = 1 + 0.2 + 0.04 + 0.008 + 0.0016 + ⋯ can be rearranged any way at all, and you always get the same answer. It’s “absolutely convergent” (always convergent). But if the signs of the terms alternate, and if taking just the positive subseries diverges, then re-ordering allows you to sum to anything at all. Such sums are “conditionally convergent”, with the “condition” being that one or another ordering is forced or chosen or fixed.
Now here’s where the “poetry” comes in. I could also write: u ( 3 ) - 3 u ( 3 ) = 1 - 3 + 3 - 9 + 9 - 27 + 27 - 81 + ⋯ = 1 - 3 - 6 - 18 - 54 - ⋯ = - ∞ So here, I have chosen a different ordering. There is a freedom of choice here. Pick some order, pick any order. There are no rules for picking the order. This makes the math “poetic”, in that one is free to invent new orderings, and it is a mattter of taste and style, creativity and panache and audacity in how you chose to reorder. Good poets write good poetry, bad ones write crappy poetry, and whether you like it or not is a matter of taste.
To wrap up this part: this opens a deep and bottomless pit of what it means “to reorder”, and explorations of exactly what is going on here. There are hundreds (thousands?) of books written on this topic, because its .. well, the freedom of working with infinitely many possibilities is ... very free. Seemingly unlimited amounts of invention and exploration are possible. Banach spaces and Hilbert spaces, Frechet summations and trace-class operators, and on and on. All these arise by considering how to re-order conditionally convergent sums. And the freedom to do as one chooses. All (pretty much all?) of the most brilliant mathematicians have dug into this.
But I digress.

Part the second

One technique for reordering is called “analytic continuation”. You’ve already encountered this phrase. Let me try to recap it. Write u ( z ) = 1 + z + z 2 + z 3 + z 4 + ⋯ = 1 + ( z - a + a ) + ( z - a + a ) 2 + ⋯ so that I add and subtract a every time. Clearly, um, valid, err. Well. See what I mean about hallucinatory? Seems like the above should be valid, right? How can it not be? What can possibly go wrong? But now, expand this out: u ( z ) = 1 + ( z - a ) + ( z - a ) 2 + ⋯ + a + 2 a ( z - a ) + a 2 + 3 a 2 ( z - a ) + 3 a ( z - a ) 2 + a 3 + 4 a 3 ( z - a ) + 6 a 2 ( z - a ) 2 + ⋯ + ⋯ What a freakin mess. But here’s the deal: as long as you pick z and pick a such that | z - a | < 1 and also that | a | < 1 then you can perform these sums, and they always converge, they are absolutely convergent. Reordering terms does NOT change the value. No hanky-panky. These are NOT conditionally convergent; you can reorder as you wish, you get the same answer. But only when | z - a | < 1 and | a | < 1 .
And now for the magic trick. Let both z and a be complex numbers (but still subject to the contraints above). And, to be specific, let a = 0.9 i so that it is on the imaginary axis. Then the sum v ( z ) = 1 + ( z - a ) + ( z - a ) 2 + ⋯ is absolutely convergent when | z - a | < 1 . The region where | z - a | < 1 is a disk, and this disk is centered at a and the radius of this disk is | 1 - a | = x 2 + y 2 where x is the real part of 1 - a and y is the imaginary part of a . So for the demo a = 0.9 i the radius of convergence is r = | 1 - 0.9 i | = 1.34536...
Now, u ( z ) = v ( z ) + other stuff, and when you work out what “other stuff” is, you get, hold on to your butt, u ( z ) = 1 / ( 1 - z ) But only when | z - a | < 1 and | a | < 1 .
What’s happened here? Well, with a bit of trickery, a bit of reordering, we were able to take the series u ( z ) and write it so that it is absolutely convergent on a **larger** region of the complex plane. In particular, it is convergent for z = 1.3 because 1.3 < r = 1.34536 ⋯ and this is in sharp contrast to the earlier result that 1 + 1.3 + ( 1.3 ) 2 + ( 1.3 ) 3 + ⋯ = ∞
Now we repeat this trick, but this time with v ( z ) , to get a larger circle. The only rule is that the center of the new circle (call it b ) must lie within the old circle, and that | z - b | < 1 . Do this again, and again, and if you are clever with picking the centers of the circles, you can reach anywhere. In particular, you can get to z = 3 in about 5 or 6 jumps, and when you do, you find that u ( z ) = 1 / ( 1 - z ) at z = 3 . And repeating, you get u ( z ) = 1 / ( 1 - z ) EVERYWHERE.
So it turns out that old high-school fraud *actually gave the right answer*. It just gave the right answer for the wrong reason, which is why it was fraudulent. But explaining all this to high-school kids is ... well, that’s what college is for, innit?
So (and this is very very important) you get this only if you follow the rules about the location of circle centers, and the allowed circle radii. If you don’t follow these rules, you’re back to being able to get assorted forms of nonsense.
And so finally, we are back to our hallucinated reality: 1 + z + z 2 + z 3 + z 4 + ⋯ = 1 1 - z which is “true” only when you follow certain complicated rules. Its hallucinatory, since if you plug in z = 3 directly into this “formula”, you get whack. So this formula is both false and true, and, like radioactive plutonium, you might not like the results if you mishandle it. The safe way of handling this is with analytic continuation, where you deal with absolutely convergent sums at each step. It’s the absolute convergence that provides the safety. The high-school trick just happened to work. It doesn’t always: you get weird things like Frechet spaces, which are filligreed with these rather unhappy discontinuities, failures where naive approaches blow up in your face.
This is one reason why math can be so boring. Safety requires rigor, and safety is boring. And its even worse: the boring rigor hides all the fun parts of math. Most math books, with a few, rare, treasured exceptions, are not written in the style I am writing in, here. They carefully hide all the surprises and adventures and wonders under a bushel, so that only the most dedicated and autistic students can find them.
Its like: its more fun to dance up the side of the volcano, but if you don’t put your foot down just so, you go over the edge of a cliff and its the end. So the rigor is needed. But if you have that ability, then you can dance: the choices, such as the choices for reordering, are infinite, and how you dance is inventive and creative and a matter of style and taste. And this is also why math is fun: once you know what works, its like making bombs in chem class: look what happens when I mix these two: kaboom! Right over the cliff! Lets do it again!

Conclusion

To conclude: I went on and on and on with all this, and never got around to zeta. But all that zeta is, is this: ζ ( s ) = 1 - s + 2 - s + 3 - s + 4 - s + ⋯ and the right hand side is **absolutely convergent** if and only if the real part of s is greater than one; otherwise it blows up.
However, analytic continuation, with the circles, with the rules about the circle centers, and the circle radii, all that logic still goes through, and can be applied to ζ ( s ) . If you are careful, then you find an answer: ζ ( s ) = 1 1 - 2 1 - s ∑ n = 0 ∞ 1 2 n + 1 ∑ k = 0 n ( n k ) ( - 1 ) k ( k + 1 ) s This is valid for *all* values of s . All of them. Here, ( n k ) is the binomial coefficient, recall Pascal’s triangle.
So lets plug in s = 0 : ζ ( 0 ) = 1 1 - 2 ∑ n = 0 ∞ 1 2 n + 1 ∑ k = 0 n ( n k ) ( - 1 ) k ( k + 1 ) 0 = - ∑ n = 0 ∞ 1 2 n + 1 ∑ k = 0 n ( n k ) ( - 1 ) k But ∑ k = 0 n ( n k ) ( - 1 ) k = { 1 if  n = 0 0 if  n > 0 so ζ ( 0 ) = - 1 / 2 and we have the hallucinatory result ζ ( 0 ) = 1 - 0 + 2 - 0 + 3 - 0 + 4 - 0 + ⋯ = 1 + 1 + 1 + 1 + ⋯ = - 1 2 Lets try again for s = - 1 ζ ( - 1 ) = 1 1 - 4 ∑ n = 0 ∞ 1 2 n + 1 ∑ k = 0 n ( n k ) ( k + 1 ) ( - 1 ) k and ∑ k = 0 n ( n k ) ( k + 1 ) ( - 1 ) k = { 1 if  n = 0 - 1 if  n = 1 0 if  n > 1 and so ζ ( - 1 ) = - 1 3 [ 1 2 - 1 4 ] = - 1 12 and so the hallucinatory result ζ ( - 1 ) = 1 1 + 2 1 + 3 1 + 4 1 + ⋯ = 1 + 2 + 3 + 4 + ⋯ = - 1 12 But, by now we’ve descended into tedium. This is straight-forward plug-n-play. The magic, the hard part was doing the analytic continuation, and understanding why analytic continuation works, in the first place. It’s very .. alchemical.
Well, the Riemann zeta also has lots of other magic, but that’s unrelated to the above idea: to the idea that there is a way of writing ζ ( s ) so that it gives one and only one value, everywhere, for any s in the complex plane.

Ramanujan

Ramanujan was a freak. At age 12, he stood in a corner and practiced multiplying and dividing large numbers in his head. This has the effect of training up neural circuitry for these tasks, and making it automatic. In the same way that you don’t have to think about how to row, you just “do it”, he could “just do it”, as well. It only takes a decade of practice, that’s all. You could learn to be like him, too. You’d die of boredom, first, but you could do it if you really wanted to.
So, in grade school, we learn something called “long division”, and it turns out long division generalizes, easily and trivially, to the division of polynomials. Ramanujan could do long division of polynomials in his head, and “just see” the answer. So, its like having a symbolic calculator attached to your brain. I dunno, so like NeuralLink, but the olde-fashioned way. Having this calculator allowed him to perform seemingly magical feats, especially in a world that simply had never seen a calculator before, nothing more than a slide rule.
Now, we can all be like Ramanujan: we don’t have NeuralLink but we have eyeballs and fingers, which are almost as good. We all have access to arbitrary-precision symbolic calculators, they’re called computers, and if you become adept with using a computer, you can do ... well its up to your imagination, and it is driven by your interests. And these are driven by our social and cultural milleu, so ... well, anything is possible. So what you gonna do?
I mean this in a literal sense on multiple levels. I know one guy: Simon Plouffe. Complete amateur, no formal training in math. Sits around and fucks with symbolic calculators, and just finds relations that he can’t prove. He’s found a treasure trove. He’s in wikipedia: e.g. the Borwein–Bailey–Plouffe formula for pi in hexadecimal. https://en.wikipedia.org/wiki/Bailey-Borwein-Plouffe_formula I think they cheated him a bit putting his name in last place, not first place, but he’s also mathematically illiterate, and so that’s what happens. Have calculator, will travel. Here: https://link.springer.com/article/10.1007/s11139-011-9335-9 – check out who the author is. You’ll be amazed. There’s also this:
https://www.researchgate.net/publication/356880658_A_simple_proof_of_Linas's_theorem_on_Riemann_zeta_function
So there. Heh.
– Linas – 22 August 2024

6 Sept 2024

OK, so this is not going to work, because it is fundamentally unbounded, but I have to try anyway, because practice makes perfect. Movement is prefereable to no movement. See? That last sentence: that’s forward progress, of sorts. Even if it is a repetition of the past, a repetition of something well known. Repetiation, with variation, is a hallmark of life. What am I thinking of, as I write this?
Last night, I tossed and turned, as three distinct ideas collided in my head: an animation of mitochondria, my studies of civilization, and my personal failure along all axes as conventionally measured in American culture: fame in academia, riches in entrepreneurship, power in leadership. Not that I ever tried hard in any of these, but my failure at the token efforts still stings. But maybe save that for later. Lets go abstract, first. Saw on twitter an animation of a slice of a mitochondrian, showing struturally accurate major proteins. It was quite remarkable. The proteins, although mechanical in function, were remarkably crooked, asymmetric twisted shapes. What immediately went through my head is how these acted as traps for free energy: Through milledia of evolution, nature found certain arrangements of atoms can extract flowing energy. Vibrating back and forth, with assorted electronic gradients across molecular bonds, ionic interactions, hydrogen/water bonds, serve as filter-traps for free energy. The hap-hazardness of the process is suggested by the irregularity of the molecules: they’re blobs, and whatever functional action they have is hardly apparent from a simple visual inspection of their shape. Certainly, something can be found by deeper examination of the vibrational, electronic properties, but whatever that is, its subtle and not on-the-nose obvious. A wild complex system for trapping free energy. My imagination is colored by studies of fractals and modular forms, hyperbolic spaces. A folding in on oneself, but without the natural order or symmetry of a fractal, but something far more twisted up.
This is at the microscopic scale. A lifetime of study of biochemistry might enlighten mitochondrial mechanics. There are certainly thousands of biologits looking into this, and each new headline is astounding and fascinating. And all this gives what? Maybe a gigadalton of mass, doing some sophisticated free energy extraction from the surrounds? Now scale this up to civilization, thourh the usual path: cellular biology, organisims, neural structures of increasing sophistication, ant colonies, flocks of birds, prides of lions. Language and speech centers as thin prefrontal cortex wrappers on reptilian brains playing games and falling in love. Villages, towns, cities, societies angaged in literature and history and art and industrial manufacture, wrestling with sources of fungible wealth. Billions of human minds exposed like petri dishes to the latest memes and propabganda, accepting of any factoid, however true or false, is promoted by social media. Memes spreading like infectious disease through the global brain. Nascent electronic AI brithing on this petri dish. All for what? This is the grand question; what is the meaing of life? How odd that capture and harenssing for free energy leads to this.
So I’ve spent a large fraction of my life looking at absolutely everything, everywhere. Sports, art, music. Psychology, sociology, economics, politics. Love, happiness, depression, anger, hope. Being lost, being confident. The sceinces: from mathematics to quantum physics, and all the way up the chain. I’ve got a colleciton of hundreds of youtube videos detailing auto manufacturing in Detroit, from the 1930’s, onwards. The mining of copper from the 1910’s onwards. Smelting and casting and hammering and rolling steel. The construction of Teslas gigafatory in Austin. I started reading George Elliots Middlemarch a week ago, and I’ll tell you, the preface, the first chapter is knock-your-sox-off superbness. I’m deeply, shockingly impressed. I haven’t read anything this good in, like, forever. I hope the rest lives up to this start.
Anyway, where was I? All this, for what? Al this, the everything I’ve looked at, it seems like perhaps it is time to start integrating all this into some grand... something. Grand narrative? Which is an effectively impossible. I said unbounded earlier, but maybe impossible. I as a single human with a limited lifetime, cannot reconcile theories of gravitation, black hole horizons with the goings on of love, hate, qualia, with literature and art, beauty and engineering. These are all just things that are out there, and dammit yes, I’ve looked at them all, and it is quite the vista, and ... what is one to conclude dro this? I say, impossible, because I amagine that if I had an IQ of a million or a billion, and the workings of mitochondria were somehow tritely obvious at that point, and perhaps even the expanse of human Western Civ became comprehensible and graspable, then I imagine this same IQ would trip over progressively more complex, mind-bendingly arcane structures in the physical unverse, in the noosphere. For every advancement in intellectual ability, I imagine that complexity expands at an even greater rate, super-exponetial, combinatorial.
So there I am, tossing and turning in bed, ant the above paragraphs are what is sloshing about in my sleepless thoughts, and for what? Unresolved, unresolvable. Here I am, writing up my dream: is it any more coherent in my waking state? Is it any more lucid now? I vaguely hoped it would be, but perhaps I was wrong: perhaps the agonies of tortured sleep provide as great or greater insight as the waking state: yes, I can control the direction of my thoughts as I am awake, but the unbridled multiverse of sleep seems richer and more expansive, offerig a greater choice of possibilities. Sadly, muddled possibilities: the surgeons knife of waking thought is trying to tease apart a marshy swampland of reallity. Like a surgeon cutting through living tissue, one certainly cleaves something, but that cleavage respects no natural boundaries. Here I sit, broken hearted, came to shit, but only farted.
Well, whatever. I’ve got a very long list of to-do items that I suppose I should get around to to-doing. Almost all of it is mundane, almost all of it is tiny, back and forth steps. Did I mention I’m trying to learn to dance the shuffle? Harder than it looks. I’ve spent maybe 5-10 minutes on it, and I still can’t do it. Fancy that. I spent 20 years rowing, I’m only now able to control it, but even that requires active attention and concentration. Humans are small, frail, of a short lifespan. And I am one. Where did I go wrong? Askig that is perhaps like asking a mitochondrian what it did wrong. It lived, it tried, it did it’s thing, for better or for worse, and it was hardly more enlightened at the end, than it was at teh begining. It merely ... extracted some free energy, and contributed to the puzzle of the universe. So, this diary, this text, is my contribution. My life’s work: all the things I’ve done, these too are part of the universe, and like a grain of sand on the beach, I have no clue of what lurks in the depths of the ocean. My weak imagination only imagines whales.
Anyway... once again, its late. Once again, I am gettting up early early, to go rowing. So bedtime.

7 Sept 2024

Discord conversation, I copy here because it will be lost otherwise. I’m hoarding memories, here, because there’s no good way of doing backups across different media (including social media)
Linas — Today at 12:58 PM Hi Adam, I’ve been ignoring discord for the last few years; my apologies for being incommunicado; thanks for emails and notices. Adam — Today at 1:02 PM Hi Linas! Your name popping up put a smile on my face. Public Discord servers do not seem too great these days, but I still use the app for some teams 🙂 I’m on the bullet train back from https://aitp-conference.org/2024/ and would love to hear what you’ve been up to. Linas — Today at 1:15 PM I’ve been up to nothing at all. Well, fell in love with a girl in Lithuania, but now I’m in the US and so miserable. Spent some time thinking about the nature of love (something all mammals, probably all animals feel) and language centers. Piddled around with more math. Nothing particular to report. Adam — Today at 1:21 PM Wow! "Love and math" sounds peaceful. I’m sorry to hear you’re feeling miserable. Anything you want to share or are looking for in the future? Linas — Today at 1:26 PM Ah, thanks! The immediate love-pangs subsided. Reality and idealized romantic love are not really compatible. White Stripes have a song with the lyrics: "My left brain knows that love is fleeting ..." Sharing: my personal predicament is unchanged since I saw you. I’d like to (1) engage in a certain line of research (and I have some new, interesting, positive ideas I’m excited about.) (2) find someone with whom to talk about about these things (3) get paid for doing this. You can kind-of-ish almost fill the role of (2) but I’m frustrated that you’re more or less then only one sharing common interests and willing/able to engage. So, mostly I spin my wheels and waste my own time. Adam — Today at 1:34 PM
and I have some new, interesting, positive ideas I’m excited about
👀 Do you have a one-sentence description or some labels? I may be able to set up a nice group, so you have a couple more interested ears. If it relates to graph rewriting, code synthesis, or symmetry/pattern mining I may be able to help with (3) Linas — Today at 1:45 PM How about one paragraph? Perception and action requires a description of the data format (of the sensory data, for the motor-control data) Old-school is to have human engineers write the API. Instead, ask "how can an agent perceive that format? (the format of sensory data)" and "how can an agent discover what it can do? i.e. not just discover it’s external environment, but discover how it can act on its environment?" These are somewhat recursive questions (how do I perceive what I perceive what I perceive...) but I think they can be grounded, given an abstract mathematical framework, a la category theory or similar, and incorporated into a general theory of learning. So a general theory of perception, action, learning of agents embedded in an external world. I kind of dislike this last sentence, thought, because its "obvious" and would be something you’d read in the preface of some 1960’s textbook on AI. The difference is that now, we’re 60 years past the 1960’s in sophistication, and so we can do better. Adam — Today at 2:02 PM Interesting. Can you relate this to "discovering what the (causal) consequences of your actions in the environment are"? Linas — Today at 2:11 PM Not in any deep way. Nothing more than "do something and see what happens". Let me backtrack, and provide a narrative on how I got to here. Adam — Today at 2:16 PM A framework modeling the "what happens if you do what" I do know that may be of interest is https://en.wikipedia.org/wiki/Alternating-time_temporal_logic This obviously doesn’t concern learning, but perhaps learning may be framed in a "finding the model which explains what is perceived" on top. Alternating-time temporal logic In computer science, alternating-time temporal logic, or ATL, is a branching-time temporal logic that extends computation tree logic (CTL) to multiple players. ATL naturally describes computations of multi-agent systems and concurrent games. Quantification in ATL is over program-paths that are possible outcomes of games. ATL uses alternating-tim... Linas — Today at 2:16 PM So I’ve got this learning system, and for historical reasons, its batch-based. So I said, gee, I need to redesign it so that it’s continuous-learning. OK, So instead of force-feeding it piles of text, as batch input, I’ll just leave it out there in the file system, and let it explore on its own. So, lets see, I need an API to walk the file system, and an API to open a file, and to API to read the text (unless its sound or photos..) Obvious boring plain-old engineering stuff. But an "API to walk the filesystem" is a "motor control API": a description -- ’you can do this, its called "change directory", and you can do this its called "list the files" and when you do this or that, here’s what you get back, here’s the format of what you get back.’ So that’s what is usually called an "interface definition", and the jump to the abstraction of an IDL (interface definition language) was made in the 1980’s maybe earlier. Linas — Today at 2:26 PM And I’m thinking: hey, hang on, I should be describing the interface to the interface, so that’s the first recursion step. I’m not really interested in creating yet another API for the file system, and manually wiring it in to my learning system. I’m thinking, instead: describe the interface to the interfaces, and let the system work out on its own how to go from there. But how does one write an "interface to interfaces"? And, given my historical baggage, "how do I do that in Atomese?" and its surprisingly hard. and right around there, I get a glimmer of something deeper: an a ha moment of having an agent that can make choices to do something, but it doesn’t yet really know what it can do, so it has to try things out, and it doesn’t know what happens ("causally") when it does something, so it has to perceive that, but what exactly is the format of this sensory data that it is streaming in, anyway? What’s the interface to the sensory stream? How is the sensory stream navigated and parsed and analyzed? And oh, by the way, how does the agent remember what it tried, and how does it correlate what it’s done in the past, with what its perceiving now? Linas — Today at 2:35 PM So all this becomes a tangle of abstractions layered on abstractions, and I realize I don’t have any kind of formal theoretical framework for any of it. I can’t implement this in software, because I don’t quite really know what it means for something to be input or output, or memory. But I’m smart and experienced, so I think that maybe I can figure this out if I try hard enough. But then "real life" got in the way, and so did a bunch of distractions, and so no forward progress. That, plus the fact that if I did make forward progress, no one would know, no one would care, it wouldn’t matter to anyone, so its kind of demotivational. It’s like one or two abstraction layers too deep, and there is so much going on in the world that my floundering attempts at understanding "what is agency?" is ... well, I’ll get around to it eventually. I just have to clear my plate of more mundane concerns. Adam — Today at 3:13 PM It seems like your theory would need to be quite mobile/computational, which is awesome. At AGI24 I talked to David Spivak for a long time, and they have a framework called "Polynomial Functors" for modeling this kind of stuff. It’s not plug and play though, and barely computational. Greg Meredith uses the mobile process calculi, which admits to some nice analysis "spatial behavioral types", but it’s not widely explored. Simon Peyton Jones raised "self describing interfaces that can evolve over time" as the holy grail for his last "world scale programs" project. I think some folks at Topos may be interested in this, along with Luke (who has been bothering me with "but how does it explore its own interfaces" for years) and Greg (who’s looking at different parts of the environment as their own agents with their own modes of communicating). Linas — Today at 3:19 PM My to-do list includes talking to Greg Meredith in much greater detail. I sent him several emails but got no response, gotta try again. I did look at polynomial functors, but did not see any immediate utility for it. People have sent me a lot of links to a lot of different topics, everything from "basal cognition" to "bratelli-vershik diagrams" to "modeloids" to "supersymmetric statistical mechanics" and its all intriguing, but it all takes time and effort to understand, integrate, and figure out what aspects can be used as inspiration, what can be used for something practical, etc. Adam — Today at 3:55 PM Absolutely. It’s still interesting to have some "from first principles" run through. You gave the file system example, do you have something written up on this? Adam — Today at 4:16 PM All this Theorem Proving stuff at the conference reminded me of ASP which you once recommended me to look into. So I read this PDF over the course of the conference and I have many questions. I’ll limit myself to just one to start with: why are people outside of mathematics interested in theorem provers? It seems like ASP is more general, gives you more information, and is not much slower in the grand scheme of things. Is it just harder (to configure or implement an ASP solver)? Attachment file type: acrobat asp.pdf 5.78 MB Linas — Today at 7:04 PM https://github.com/opencog/sensory it’s an experimental diary, not a formal writeup, so perhaps convoluted, long-winded, informal. Linas — Today at 7:15 PM Re: why do people like theorem proving instead of ASP? Not sure. I can hazard a guess. a combination of history, prejudice, momentum and maybe or maybe not rank ignorance. A long long answer follows below, if you’re ready for the long long answer. History: (1) ASP happens to have more-or-less the same syntax as prolog. Completely different algo for solving the expressions. Utterly different performance profiles. But same syntax. So prolog has a strike against it: from the 1980’s, its a been-there, done-that idea, people wrote expert systems aka AI in the 1980’s; few seem to be keen to revisit that era. It has the stink of "old stuff" surrounding it. Linas — Today at 7:22 PM Now, those old expert systems consisted of thousands of hand-crafted rules, written by experts. I keep thinking that we can now automatically learn those rules, assign probabilities to them, etc. but I don’t know that anyone is actually trying to do that. (2) Both prolog and lisp, early AI favorites, predate "type theory". They were both more-or-less untyped. Or loosely typed, or ad-hoc typed, depending on how you think of it. Since then, things have changed. Early concepts of typing lead to the idea of "object oriented", eventually resulting in C++ and Objective-C. But even these two are not entirely rigorous about types, they’re a bit loose. After this, two things happened with types. Linas — Today at 7:31 PM One was that the formal type people came up with ML and later, CaML which had explicit strong types and the ability to perform explicit logical reasoning about types. Turns type inference is not turing-complete, and that’s a good thing: all type inference terminate in finite time, and quickly! So it splits out part of the programing task into something that can be automated, and runs in finite time, and runs quickly. The other thing that happens is the internet, web, http. Early web pages were static, but scripting was needed. You could do cgi-bin with c/c++ but perl, and later, PHP were more effective. And people started inventing brand new languages, like python. And implementing python (and perl and PHP and 1001 other languages people are creating) results in people thinking hard about types, and trying to do them semi-correctly, as best they can (e.g. duck-typing in perl: the type is hidden from the user, but only because the perl internals are clever about hiding it; they’re actually in there.) Anyway, type theory advances. Linas — Today at 7:39 PM So some or most of the theorem provers absorb these typing lessons. For example, agda adds ability for do type inference on union types, something that no other language does. These end up being used as marketing claims: "our system can do something your system cannot", and since we’re humans, we’re susceptible to marketing claims, and so some of these rather arcane systems gain mind-share. Being hip and trendy and fashionable is not limited to narcissists; even the nerdiest scientists do it. Oh, one other thing, re performance, expressability, syntax, optimization. This is a story of SAT solvers, SMT solvers. Linas — Today at 7:49 PM So, in chip design, formal verification is a big deal, because when you "RIT a chip" (run the compiler) you can only afford to do it twice. Maybe three times, cause its so expensive. Millions of dollars. You know there will be bugs, and so on the second try, you hope to fix all of them, and if you fuck up, either the executives give you a third chance or they lay everyone off. No one wants to be fired, so making damn sure the hardware will work as designed is a huge deal, and you have these huge, complex, arcane, highly proprietary chip verification systems. You run em like crazy before you RIT the chip. The early 1980’s systems did forward-chaining and back-chaining, a lot like prolog. And this worked great, as long as your chips had less than 100K or maybe 1M transistors in them. There were alternative algos, e.g. DPLL, but they were slow and klunky, compared to forward/back-chaining. ... until you got to about 1M transistors, at which point, the performance curves crossed over. And not just a little, but a lot. All of the old verification companies went bankrupt when this happened. Like 100% of them. They were not ready, they couldn’t retrofit. The mind-set for DPLL is utterly different from forward/back-chaining. Linas — Today at 8:00 PM A different name for DPLL-solving is "boolean SAT solving". So, old prolog was (and still is) based on chaining, and it works great on certain classes of problems. By contrast, ASP, which uses same syntax as prolog, but has SAT under the covers, works great, but for a different class of problems. SAT is better at constraint solving, chaining is better for reasoning. ... but ... there’s one more fly in the ointment. Not everything is boolean. Some problems need/want to have arithmetic in them: addition, subtraction, multiplication, inequalities. Classic examples are the so-called "linear programming" (LP) problems from WWII: you have X amount wheat and Y amount of barley and you can brew lager or ale, which sells for $Z and $W, how do you brew to maximize profits? (The WWII example involved tanks and jeeps and ships, I forget the details) Modern versions of LP are used to solve zillions of modern economic problems, e.g. running solar/wind farms, factories for god-knows what, and ... electrical signal delays on chips. Linas — Today at 8:07 PM So how do you couple LP problems to SAT solvers? This is a generic problem, lots of people in the industry have it, they fund academia to solve it, etc. etc. and you get SMT solvers. Satisfiability-modulo-theories. Here, a "theory" is a collection of "axioms", which you can apply in any order you wish. For example, if a>b then a+c>b+c would be an axiom. Linas — Today at 8:15 PM "modulo" is like long division: if you have some large integer, then you can either divide by a prime or you can’t, and you have a collection of primes you can divide by. "modulo theory" is like that, but you start with a large complex expression, and you can either simplify it, by "dividing" by an axiom, or you can’t. You proceed by reducing the expression as best as you can. If you’ve reduced your initial expression to true then you’ve "proved" that your initial expression is true. But what is the best order in which to apply the axioms, to most efficiently reduce the initial expression? well, hard question. Given this description of "module theory", you can see that its more or less the same thing, kind-of-ish as theorem proving. But wait, there’s more! Its not just SMT, but its also combining the "modulo theories" part with other boolean expressions (to obtain satisfiability) so e.g. "either door is open, or door is closed and temperature at top of door is > x+y-c then yell fire else open door" and you have to do the modulo-linear-programming part to figure out the temperature part of it, and ordinary boolean-SAT to handle the if-then boolean part of it. Turns out SMT was really friggin hard. Linas — Today at 8:26 PM The simplest theory is the "theory of equality" and it has only a handful axioms: x=x is true and if x=y then y=x and if x=y and y=z then x=z and that’s it, and it turns out even that is painfully hard to solve. So most SMT solvers would implement the theory of equality and say "ah hah! we did it!" and you’d ask them about inequalties, or god forbid linear algebra, and they’d say things like "were working on it, maybe next year". I assume, I don’t really know, most modern theorem-proving systems are outgrowths or updates to the older SMT ideas, but I have not really kept track. So, to recap your question "why theorem proving and not ASP"? the answer would seem to be... theorem provers have baked-in support for handling certain types of axioms that you might have to otherwise "roll your own" in ASP. For example .. linear algebra. And although ASP is blazingly fast for certain classes of problems, it does require exhaustive search for the reduced kernel, and maybe the theorem-proving people have brand-new, better algos for doing the exhaustive-search corner (this is a corner of the so-called "lambda cube" that everyone hates/ignores because its so hard.) Linas — Today at 8:36 PM I’ve been ignoring theorem-proving, because I want to get away from strict axioms that are strictly true/false, hold/don’t-hold, and instead follow probabilistic distributions. I think this is where Ben started 20 or 30 years ago, but it feels like he lost the plot. Meanwhile, LLM’s and deep learning and transformers have turned everything sideways, so now there’s a whole new set of confusing things to try to work with. So, you know, we all do what we can and ignore the rest. If you read the above, and see parts where I’ve failed to mention something important, or put the wrong emphasis on something, or god-help said something totally bone-headed, let me know. You went to the conference, which means you should now know more than me about half of these topics. (Heh. OK, I remember going to physics conferences, and understanding maybe 5% of it So ...YMMV, as they say.)

9 September 2024

So I continue to expand my health plan text, which is now maybe 50 pages long? or longer? that I’ve been developing since 2007, and of course it has enough material in it that it’s hard to keep sorted, and since it contains significant knowledge, requires intellectual effort to retain. Of course, writing is a prosthesis that helps with retention. But what about control? So, of course, one idea is to use LLM’s as a reference source. This is intermediate between my personal notes, and search engines: the LLM provides details that may be absent in my notes, but narrows down the flood of text offered by search engines. This is both academic publishing, with lots of niggling science details, and also the repetitive info found in general-health weblogs (which all say the same thing, mostly, and say it without references, and repeat common knowledge, irrespective of whether its factually correct or not. Of course, factual correctnesss is a concern with LLM’s.
So, one section of the health plan I was expanding today was on emotional health. So apparently, Vitamin K is metabolized into some neurotransmitte with potentially negative side effects. But I want to take Vitamin K for wound-healing benefits. Something similar for BCAA: negative emotional impacts, but needed to avoid post-exercise muscle pain. What about DHEA? DHEA is a “neurosteroid”; its production plummets with age. Its pumped by supplement sellers for its (dubious) body-building effects, but its also an anxiolytic and has antidepressive effects. Then of course there are all the nnotropic stacks described alll over the internet.
So, the thing that popped in my head is this: how can I deploy a personalized LLM to help maintain health homeostasis? Right now, I do online research for assorted supplements, I buy them, gobble them, but OMG there is a lot of them and I don’t know if they cancel one-another out, if I should take them in the morning or night. I don’t have any particular feedback for their effectiveness. Well, BCAA is very distinctively noticable for muscle pain, but the rest, not so much. I took piracetam a few times, a decade ago, and wow, it made me sad. Not exactly depressive, but it made life seem pointless and painful. There I was, nearing the end of a long walk, in bright sunlight on a nice warm day, beautiful weather, thinking about how hopeless it all is. WTF. That was my final experiment with piracetam. Nasty shit.
So a personalized LLM can have several homeostatic effects: one is to help maintain emotional balance. This is perhaps the more interesting one, or more importnt one, because if one loses clarity of thought, one loses all higher regulation function. And so that’s the build out. Let’s put this into perspective (which I’ve probably done many times above, but it bears repeating) Basic bodily functions and survival are handled by the mamalian brain. The neocortex, the pre-frontal cortex get involved in higher function, helping provide additional regulation layered on lower neural mechanisms. I’m now asking: “how do I wire in an LLM to sit on top of that?” This is not entirely new: writing and diaries are memory prosthesis. Speech, language, communication allows me to offload intellectual processing tasks to other humans. So my prefrontal cortex is already attached to society and civilization, and socity and civilization offer homeostatic effects and health benefits. e.g. if I get sick I can go to a doctor & hospital, and pay for it all via insurance. So me, myself and I, the language-conscious verbal-expressive “me” is already integrated into a larger organism. I am now asking for how to correctly introduce and LLM into this mix. So, in this sense, its not all that different than my existing mix of keeping a health-plan text, and keeping a health-diary text, and using the internet as a knowledge source. But an LLM would be presumably more active, less static, having agential degrees. Being digital, it is presumably economically cheaper than consulting a medical doctor; it is somewhat ethically cleaner: like why should a doctor care about me? Why should I burden a doctor with my problems? The answer is easy: because the doctor wants my money. This transactional approach to medicine is problematic. Replacing this transaction with an LLM (or other artifical digital agent) is appealing. (In this text, I will use LLM as a synonym for artificial digital agent, unless otherwise noted. So think “digital agent in general” no matter what the underlying tech is.)
Well, that’s it. That’s the idea. I guess this idea has been burbling in my head for maybe even a deacde or more, long before LLM’s, but well, at least this time I’ve actively verbalized it, as opposed to this previous bubbling-in-the-background concept. Have I made forward progress? Well, not really. So how could I actually do this?
Lets try a sketch. I could get API keys for one of the existing major systems; or I could beg/borrow/steal weights for some open-source-ish LLM. Next, I would dump my health-plan diary and notes into it, and ask it to double-down train/retrain on medical and bio texts. And then use it as an oracle: dear LLM: what should I do to stay healthy and strong and live forever? And have it regurgitate random bullshit advice from health gurus who happened to publish a blog that appeared in the training set. So this is problematic: LLM’s are like photographs; they are only as reliable as the subject that has been photographed. If there’s bullshit in the training set, you get bullshit out. So this is another meta-cognitive effect. How do I know that some particular health advice is bullshit? Several ways. One is delivery: the writer demonstrates poor control over the english language. The writer is writing for a low-IQ audience. The writer has a poorly-disguised hidden agenda. The writer makes claims that are outlandish or absurd. There’s a grayzone here: even if the worst offenses are absent, a given blog article or bit of health advice might still smell bad, be off in some hard-to-define way. And LLM’s have none of these discernment abilities. They can’t really see that something is off, something is broken, incorrect in the advice.
This is only a problem because most of the health cures that I pursue are at the edge of what is known by medical science. I don’t need an LLM to nag me to lose weight or get regular exercise; thes are now socially accepted common-sense recommendations, and a high-tech agent is not needed for that. I’m interested in pursuing therapies about which very little is known. Hang on ...
So. Well, let’s break this down. I want (1) a personalized medical doctor familiar with me, both my verbalizations, but also the results of blood tests and other invasive health measures (2) an expert knowledgable in all the ways that regular human MD’s are (including sports therapists, psychiatrists, etc.) and (3) a system capable of performing original science research to precisely clarify how some supplement or therapy interacts with my particular health profile. It seems this would require deep knowledge of biology, biochemistry, genomics, proteomics etc. (4) a user interface between the above and me. How will this interact with me? Will this digital expert be a freind who I chat with daily? A consultant availble on chat? Give me phone calls, reminding me to get off twitter, go outside for a walk, and then get ready for the party tonight? Maybe it will be some neural implant, so that I can hear it the way one might hear voices in one’s head. Sheesh. Maybe I surrender volitional control, so that instead of hearing voices, it simply makes me really really want to open the lid of that bottle of AKG and down a few grams with my morning coffee. That is, I hand over a subset of my free will to this digital agent. Sheesh.
So, what’s the path for satisfying items (1)-(4)? All of these are beyond the current state of the art. My options are to: (a) personally tackle some aspect (b) leverage some open-source community to do same; possibly creating such a community (c) same as (b) but propritary startup with funding, or (d) do nothing, wait for society/culture/western-civ to tackle these problems, i.e. wait for externeral progress of which I am not a part of. directly or indirectly.
This specific diary entry consists of my personal direct attack via (a) and represents a few hours of work effort. My gut-instinct for the state of the art says neither (b) nor (c) are viable at this time, although staying current with news and announcements is recommended. The default is therefore (d) sit and wait. In the meanwhile, I can continue a multi-progoned approach: maintain my own personal diary, read up on health effects in general, do medical tests e.g. bloodwork, and pursue my deep-AGI & math research. Onward-ho, no obvious course-correction at this time.
The one curious aspect to this diary entry is that everything I wrote above, I already knew, subliminally, subconsiously, consciously, even. The response I wrote above could have been easily provoked in some social setting or party conversation. There’s nothing particularly new here. Sometimes writing, speaking, talking, setting to words helps me formulate and gel vauge ideas, or even discover new ideas I’d not gotten before. For the above, this is not the case: I kind of already knew all this, and found nothing new, other than perhaps confirming a general trepidation that perhaps LLM’s aren’t really ready for prime-time. And this last sentence also indicates the point of greatest risk: am I discounting LLM’s too much? When might they be ready? Am I too out-of-touch with the state of the art? Perhaps I am ... this is indeed a point of risk.
Of course, focusing more on LLM’s takes time away from focusing on other things, including the general complexity of ... well, medical biology, narrowly, and complex socio-economic systems in generality. So my intellectual development is a bit of a random walk exploration, and maybe I get lucky and maybe I don’’t. Arghh.
Touched on the above is the unexamined concept of homeostasis. It’s meant to mean “I feel normal” but what’s that? Clearly there’s a lot of variability: the daily wake/sleep cycle; being excited to do something, being energized, (I’m usually energized) sad (I’m rarely sad) depressive (I’ve noticed depressive symptoms recently: the lack of desire of tackling anything intellectually hallenging, and instead floating on a dopmane high of reading twitter feeds.) So feeling normal appears to be a chaotic walk through a commonly-visited landscape of emotional and psychological states. So, like all the places a double-pendulumn visits: areas that are frequently vistited, areas that are never visited. This is “normal”, and this is what “homeostatic” means.
Of ourse, life is boring without excursions. Excurions include drunkenness and all sorts of drug-induced states, but also, to take a sharp example, the feeeling of love, written about at length up above, an excursion I hadn’t embarked on for many decades. Oh, what a feeling! So if I look at some LLM-device as a meta-rational regulator of phsyical and mental health, to what degree should it control excursions to emotional extremes? Obviously, you don’t want it to prevent you from falling in love (e.g. by feeding you assorted psychologically active drugs) and you do want it to minimize depressive states, and particularly any suicidal tendencies. So this question of regulation seems to have a clear and obvious common-sense answer: “don’t overdo it” (don’t get manic for longer than a few minutes, or hours at most), “don’t get stuck somewhere for too long” (its OK to feel sad, just not for more than a few days or weeks; depression should be limited to days, certainly less than weeks.) So I don’t know why I’m belaboring all this, because the answer is obvious. Still, the nature and desirability of excurions does merit examination. I mean, what the heck, all of “grožinė literatura” - humanistic fiction - is all about the exploration of emotional and psychological states of being, of emotional and psychological excursions - something that avid readers all know as something obvious, but to me as a physics/math nerd remains a surprising and undiscovered domain, intellectually challenging and intriguing. So I write here. So now take all of that, and pile on the additional burden of the “nanny state”: to what degreee should a personalized digital doctor/psychiatrist should baby-sit your psychological well-being? The background is that society only does a marginal job at this: look at all the homeless people out there, seniors living in despair, young adults lost and dazed and confused. So the current societal and governmental and civilizational forces do ask for all this to be addressed, and do take some steps, although these steps are perhaps insufficient, and now we layer on a layer of digital technology: cheap, cheaper than existing suicide hot-lines; reliable but only questionably-so, given well-documented failings of LLM technology. Customizable to an individual level, with associated risks, e.g. the Chinese/PRC survellance state of Orewllian proportions, with the social credit score, and the 15-minute cities. So it seems that socio-psychological regulation is ... a big emerging trend in civilizational forces.
Again, what I write above is obvious. I write it because I marvel at it, I try to grasp it, even as I am cognizant that it is somehow ungraspable. And yet .. it is graspable: the russian KGB/FSB theories of reactive control is all about shaping socio-cultural dynamics, using disinfomration and propaganda, to encourage harmful excursions into false beleifs, apathy, social and political violence. So all this stuff can be theorized upon and even controlled. So there is more to sociology than just descriptive formulations: George Elliot’s Middlemarch describes the inner soul. Its a sensory system. But systems like FSB/KGB reactive control, or more directly, Chinese social-credit scores, are overt mechanistic, mechanical devices for controlling psychological states. So, a sensori-motor system. Again this is obvious. I marvel at it. But I also do so with concern: the conventional concern that someone (the russians, the chinese) are actively harmful and detrimental to my way of life, the way my neighbors and my freinds live, the way my culture (American, Lithuanian) lives; the economic and political systems of the USA and the EU. And into all of this obvious inter-relatedness, we install LLM’s in some murky poorly-defined way. Hmm.
I’m boiling the ocean here. Right? But its not like it hasn’t been done before. There is intellectual progress at the civilizational level. The thoughts of Aristotle were superceeded by the Medieval Scholastics, which in turn were refined, updated and revolutionized in the Enlightnment, in the 19th century, in whatever we want to call the 20th century. Although I have trouble boiling the ocean all by myself, my little itty-bitty attentional focus here and here, and the reporting os such results on twitter, or in published papers or in blog entries, or simply in conversations with freinds does affect the course of humanity. My intellectual efforts do affect history. What’s not clear here is whether I’m engaged in a futile full-frontal assault, or whether I am searching for cracks in the edifice with my magnifying glass. I’m doing something here, by writing this, wringing my hands out of concern about issues big and small, but where’s it all going? What’s it all about? Odd. OK. Whatever. I’m done for now. I’m going to go erging, now, before the juniors grab all the ergs. Later.

10 Sept 2024

New essay: “What’s a Theory to Do?” Brendan Chambers 4 Sept 2024 https://www.publicbooks.org/whats-a-theory-to-do/ – food for thought.
My pesonal essay here is some attempt to get closer to directly being me, or, as earlier pointed out the me that produces words, which is still not the “real” me that falls in love. My filling in love is intermediated by my verbal centers. And then, a bit more distortion, as I pick and choose words before setting them down to this page. And then an incompleteness: I cannot write everything that is banging at my door of consciousness. Sometimes, I can’t type fast enough. So, you dear reader, inly get an intermediated glimpse into being “me”. This is my personal essay.
Peculiar, I suppose, that there is something in me that wishes to connect with, mate with, have sex with the world. That has exhibitionist desires, to bare my soul, as best as possible, to the universe, so that, I dunno, the universe can fuck me. Speaking of soul-bareing and fucking, it is interesting how search engines readily throw up image search results for the most intimate and obscure sexual fantasies. Not only do I bare my soul, here, but tens of thousands, hundreds of thousands of amateur cartoonists and painters and sketchers and artists of all stripes have bared their fantasies to DeviantArt, generating vast collections of ... I dunno what to call it. “porn filth” is the conventional term. More neutrally, “erotic fantasy”. Hardly a surprise, here, as species procreation means Darwinian evolution has focesed a cosiderable portion of our thought-world onto erotica. You can’t make babies without getting horny, first. So, obviously, erotica must occupy a significant portion of the noosphere; humans as a species would not have survived without this. (Homework assignment: come up with plausible alternative theories that allow species survival, but do not lead to erotic fantasies. Is this possible? Is there any plausible alternative?)
Footnote: my personal favorites seem to be entirely mainstream. Lingerie and bimbo type stuff. Or rather, I assume this is mainstream. Only people with access to search-engine results can map this landscape.
Anyway: So, we think about erotic things. And also: we do not keep these fantasies private. At least, not in modern Wesern Civ. Prior centuries had provincial, common-man prudishness to them, but the learned, the perspacious, the urbanized sophisticates had no problems with such expressions. We have plenty of paintings of nudes from the 16th century. Who do you think was painting nudes? Shut-in incels?
So the context is that we all wish to communicate. Most of our comminucations is about how we feel. Although, for the enlightened rationalist, it is about what we think. The topic of discussion. The urge to communicate remains, although the topic being communicated is highly variable. Everyone wants to talk about “me”. Even when we listen to stories about others, it is often so that we can better understand “me”. Although I gues there is a spectrum, here. Many want to understand the external world: where do stars come from; where does my next meal come from. The internal world to, a recognition of psychology. So, indeed a desire to communicate with unbounded range of communicable topics. Can we take this desire and express it reductively? In chemistry, atoms “want” to interact, to undergo chemical reactions. Why? Mumble mumble (Gibbs) “free energy”. In biochemistry, biological molecules “want” to interact: ion pumps want to pump ions. But what about bacterial sex? Can we say that bacteria “want” to have sex? Well, yeah, I guess they do. Mumble mumble gibbs free energy. What about jumping genes? Viral genetic sequences that bounce through the genome, reproduce and propagate semi-independently of the host genome? They “want” to do this, mumble gibbs free energy. And what is it about cells that make them want to stick together, to create multicellular life? What is it about jellyfish, that makes them want to congregate and throw parties, just like on SpongeBob? What about ant colonies? So, sure people want to communicate, mumble gibbs free energy.
Yet at the end of this process, it is curious that part of that desire to communicate includes the desire to bare our deepest inner souls to others. In prior centuries, to out lovers. In the present century, the prosthesis of writing, enabled by the psychotechnology of the phonetic alphabet, allows me to spill my deepest inner-most thoughts to an amorphous, un-named “other” who may or may not read this text someday. There are two ways to think about this situation. One is that I really do have a deep intimate desire to bare my soul to the abstract universe. The other is that modern Western Civ has not provided me with the intimate lover with whom I could communicate more privately. But is it correct to blame Western Civ? Friggen every author, ever, could have, in principle, kept their thoughts provate, to be shared with feinds and family only, but no, they decided that they should write a book. I mean, some princess three thousand years ago decided to write the Book of J, eventually becoming a part of the Old Testament. So we cannot blame Western Civ, in all honesty, for the rampant desire to write. The rampant desire to have intellectual intercourse with the amorphous other. Yes, more crudely, this could be called masturbation with the written page, as the universe hardly responds in kind. But then, ejaculation is one way: it goes out, it does not come back. And if we want to be both-sides-ish here, then something something something female receptacle accepts, abosrbs the external world and incorporates it into self. That is, when its not ejaculating newborn babies. These sexual analogies are very imprecise. But colorful, I guess. I guess that little itty-bitty erotic corner of my brain says that this is an OK thing to incorporate into my writing.
Am I making any forward progress, here? Well, lets see... lets recap the main thesis. Humans want to express themselves, for all to see. This self-expression has a lineage that we can trace back, bu analogy, to mammels, vertebrates, multi-cellular organisms, single-celled life, large molecules, small molecules and grounding finally in free energy and atomic physics. This is the reductionist grounding for self-expression. Was it worth making this grounding? Do I learn anything here?
Well, if I were merely concerned about the human condition, then the above paragraphs would be shallow and trite. Insofar as I am interested in mechanistic systems such as AGI, then this gives somethig to ponder. Do LLM’s have some inate desire for self-expression? It would appear not. That is, if we stop the program, turn off the electricity, nothing happens. But what if we leave the program running, but don’t provide any input prompts? Well, then it depends on how the LLM was ocoded up: it could be silent, or it could go into a state of glossolalia. If the LLM is silent, is it still “thinking” in the background? Well, again, that depends on the design and programing of the system. Perhaps it can go into some mode of relaxing weights, annealing, as it waits for a prompt, All this is under engineering control.
This makes LLM’s very different than chemistry. Chemical reactions “want” to happen (free energy). LLM’s “want” only insofar as they’ve been engineered to want. Conventional LLM engineering does not define any free-energy surface that can fold in on itself. There are certainly analogous constructions of Bayesian priors understood as multiverses, but an energy minimization principle has been disconnected from these formulations. In physics, in the physical universe, energy minimzation is used as the driver of time-like evolution of dynamical systems. In LLM’s, there is no over-arching principle of dynamical evolution; it is very constrained to operate within the constraints imposed by the software developer. These constraints are effectively arbitrary. You could make some weak argument and say “yeah but no, since there is an underlying hardware computer” but this is very very weak. The ad hoc constraints created by the software engineer are indeed ad hoc.
Is this good? Is this bad? Should we do things differently? Can we design a system that is more physics-like, in that it has a unifying dynamical motion principle, expressed sot that it operates everywhere within the software framework? What would that look like? I’m hand-waving about Friston free-energy type dieas, but from a completely different perspective. The overall idea is valid. The devil is in the details. The details are mathematical, and my writing here is non-mathematical. So, in a sense, its roll-up-your-sleeves time, were that line of inquiry to be pursued.
To recap the thesis here: *if* an AGI software system were to be created, based on free energy principles, *then* it would naturally want to express itself. i.e. engage in self-expression. Indeed, wanting to “express oneself” can be taken to be a synonym for wanting to be in dynamical motion. The “want” is inate: dynamical motion occurs over time, and insofar as time “wants” to move forward, so do dynamical systems “want” to express themselves.
The corollary seems to be that current engineering of LLM’s doesn’t bother to think about the idea of dynamical systems. The engineering task is given by the executives: make the damn thing work. Nowhere does it say “design your LLM so that it is an explicit dynamical system”. And so it doesn’t happen.
Another corollary is that things that aren’t dynamical systems don’t meet the conventional definitions for “being alive”. Yes, there’s a huge grey area to be argued about, as to what constitutes “being alive”, but it seems that all possible definitions of life require it to be a dynamical system.
Off-topic, but let me revist my own ideas about “being alive”. Earlier i this text, I argued that consciousness should be viewed as a phase transition. One one side of the phase transition, an arrangement of atoms is not conscious, and on the other side, it is. This implies a rejection of panpsychism. What about life? A similar phase-transition argument could be made for a living bacterium vs a dead one. But this promptly puts us into the grey area of “are viruses alive”? I think the answer is both yes and no. They are not, if you imagine a virus as som collection of atoms in a vacuum chamber, interacting with nothing at all. In that case, it’s obviously dead. But if you think of the virus as a particupant in an ecological system, wherein it gets regular access to ribosomes and various transcription factors ... well, neither the virus, nor the ribosome itself are alive; but the ecosystem as a whole, is. So, again, this is kind of a phase transition. Using petroleum percolation as the example of a phase transition: no single fracture in the rock is enough to make the oil flow, and indeed, looking at it from the point of view of individual fractures is the incorrect viewpoint. Petroleum can percolate through fraked rock only when there are enough openings, strung together, end-to-end. One looks at the il field as a whole, not the individual cracks. So analogously here: we must look at the entire ecosystem of a vairal DNA/RNA strand plus also the ribosomes and transcription factors, plus a solute containing free amino acids: this system as a whole is alive, and it doesn’t make sense to ask if the individual parts are.
As I write this, I feel a little bit like a mouse running in a mouse-wheel. Running, going round and round, not getting anywhere. Am reminded of the study that shows that field mice seem to enjoy mouse-wheels. So I guess I enjoy writing, even if I repeat myself, over and over.
This then begs the question: what is the theoretical basis for the enjoyment of repetition? Well, the reductionist path is obvious: plenty of math gives oscillatory behavior, plenty of physics consists of oscillatory behavior. Its fundamental, and so hardly a surprise that large complex systems, such as mice, enjoy oscillatory repetition. We already knew that the muscles, the myostatin, work cylically, so of course this propagates up to higher layers. Oscillatory behavior in the noosphere, in processes that inhabit the noosphere, well, this is on dicier ground...
The above ruminations make me think how un-life-like ergodic processes are. For example, odomters are dynamical systems that visit each possible state, before revisting the initial state. So odometers are the ulitimate anti-cyclical device. And odometers are also the ultimate model of ergodicity, making sure that every state is visited (with uniform distribution!).
But wait, is this true? Lets consider chaotic dynamical systems. For example, the double pendulmn. They exhibit almost-periodicity, before wandering off somwhere completely different. Or perhaps the Lorenz butterly attractor. two almost-periodic, almost-oscillatory lobes, joined together. In some sense (that I do not currently understand), the Lorenz attractor will visit every possible state, over infinite time (right? Or are there some states not reachable after infinite time? You know, I don’t know). If all states are reachable over infinite time, the system is, in a sense, ergodic. I guess. Ouch. It occurs to me that there are bucket-loads of details in the Lorenz attractor that I do not understand. Like, is it a sofic system? If so, then what is this symbolic-dynamics description of it? Did I once read about this, skim it, and not pay attention? Or not? Yikes!)
At any rate, oscillation and revisting is central to dynamical systems. So I should not feel bad for revisting the same topics, over and over in my writing. As long as I do it a bit differently each time. And, well, as long as I study something new, learn something new, so that when I revisit an old topic, I have some new insight to integrate. Sheesh. Again, everything I write here feels so trite and trivial, that its hardly worth mentioninig, and yet the entire ediface remains challenging. In only a dozen paragraphs, I’ve traveled from bimbo erotica to arcane mathematical questions. But I’m skidding, fritionless, btween these topics. Skating, as it were. But that’s what dynamical systems do: they skid, they skate along constant-energy surfaces, frictionlessly, without loss of energy.
Perhaps it is time for the next topic: friction, loss of energy in the skating about in the noosphere. Are my thoughts actually frictionless? Are my thoughts a part of a larger dissipative dynamical system? Lets see. This could be interesting. This is close to something I’ve conjectured about before, but I never looked at this carefully. Sooo...
OK, so I have this super-speculative wild “theory” (set of ideas) that relate concepts of branch-points, wave-function collapse, decision-making and free will. Ive sketched some of this earlier in this diary, will recap below. The earlier sketches did not look at the role of friction, energy lose, increase of entropy or thermodynamics. So lets see if I can do that now.
First of all, as a reminder to myself: what I am about to write is likely to be total bullshit and pure fantasy. Because although I beleive that what I write has some basis in reality, it is also fantasized, in the sense that I do not (currently) have the ablity to convert this into a mathematical theory. Any such conversion is likely to reveal that large parts of this imagined situation to be wrong. I have to be careful beathing my own fumes. So... onward through the fumes.
First a recap of the situation so far:
Fuck, I should maybe delete the above. it’s hard work making up this fantasy, I literally get exhausted trying to hook it all up so it makes sense, so it works. And portions seem founded on ignorance, which I could remedy with a few years study, and other portions that are fantastic extrapolations, and then even assuming I could untangle the simpler stuff, there’s still a few uncrossable divides that ... well. I gotta go to bed. It’s time to punt on this.
I still didn’t get to the good part, which I think I talked about earlier: the whole point of many-worlds is that distinct choices can be made to form each world, and that collapse only happens later, as an accounting trick, to reconcile consistent histories. But “consistent histories” has been written about by others, already. So I’ve got an antire stable of interesting ideas, but am missing the formalism to make it more precise. And improving on this would take years, decades, lifetimes of study, and I don’t have lifetimes left. Fuck.
Whatever. Later, dude.
——-
Oh, getting back to literature: a footnote, quoting from Brendan Chambers above, that seems to more or less exactly hit on what I am doing here with this diary:
“Take, for example, My Struggle by Karl Ove Knausgaard. This six-volume, 3,600-page work, written in the first person and composed entirely of details from the author’s own life, moves between Proustian recovery of his childhood and quasiscientific documentation of the minutiae of quotidian existence, frequently taking the form of pages-long lists. Knausgaard explicitly rejects literary fiction in favor of genres that he feels “confe[r] meaning” like diaries and essays.
“And yet, these genres are no less mediated (and perhaps no more meaningful) than novels, which he describes as “made-up people in a made-up, though realistic, world.”2 My Struggle’s exhaustive lists may convince the reader that she is receiving direct access to Knausgaard’s life at the moment that he was writing. However, even lists are subject to detail selection and ordering, inevitably disfiguring the original scene which they purport to artlessly reproduce.
“Knausgaard represents a much larger literary trend often called autofiction, a broad category whose practitioners include Rachel Cusk, Ocean Vuong, Sheila Heti, Tao Lin, and Megan Boyle.
So I guess I’m just one of the popular trend. Hah. I’m not even original! Horray! I guess. I wonder if I can get published and win some literary award. Wouldn’t that be charming? Not likely, but charming.

5 October 2024

Exploring a rabitt-hole of esotericism and mysticism via Wikipedia. Fell into “Problem of Evil” https://en.wikipedia.org/wiki/Problem_of_evil Let me spill my current thoughts *before* I start reading that article. If we characterize “evil” as “pain and suffering”, then we can recognize it as the result of evolution creating stable biologica entities and endowing those entitites with survival mechanisms. Specifically, and “entity” is a structure & arrangement (of atoms, in the conventional physical universe, but could be anything in any sort of cosmology having the concept of “time”) that “exists” for extended periods of “time” (scare quotes cause existence is associated with a notion of time). In actual nature, we note that biological organism have evolved assorted defense reflexes: e.g. “run away from predators” and also self-repair and healing mechanisms: e.g. pain as a signal transmitted to inform that a certain section of the entity has been damaged, and is in need of repair. Inflammation is typical of the repair process. Pain, transmitted by neurons, serves a two-fold purpose: if the pain is induced by a threat, then run away from the threat. But also, if somethig hurts, don’t fuck with it, because it will heal faster that way.
This is a functional description of pain and suffering, and is broad enough that under this defintion, jellyfish can feel pain, but also even systems without neurons, as long as they have a “need for repair” signal coupled to a “avoid danger” signal. It seems that a proper combination of these two are sufficient to create a notion of “pain and suffering” state. So perhaps even single-celled organisms might “feel” pain and suffering (I don’t know enough biology to assert or refute that.) I suppose that the most serious objection to my defintion is that “self awareness” is also needed. Without self-awareness, the damage and defense mechanisms are purely mechanical, carrying out proscribed mechanical activities. With self-awreness comes the awareness that it is “me” who is damaged and should take measures to avoid further damage, future damage. These measures could be automatic and autonomous, inbuilt, or they could be rational, the result of having a regulatory system layered on top of all this, capable of “rational” or at least, “common sense” thought, and having that regulatory system capable of exerting motor control over the body. Higher-order control includes foresight and planning. Quibblers will insist on a fourth necessary ingredient for pain and suffering: the existance of that regulatory system capable of rtional/common-sense thinking and thus capable of planning and foresight. I’m not sure this adds much beyond what self-awareness adds.
As such, this defintion of pain and suffering is sufficiently simple that it can be reduced to an axiomatic definition, and described with category theory. I mean: come on: we’ve got the half-a-dozen axioms needed to define organisms as long-time-existing entities distinguishing self from external-world, and then some more axioms for sensory perception and motor control, and then some more axioms for damage-awareness signals, self-repair signals, and avoid-danger signals. So this is maybe a few-dozen axioms. The explicit coupling of damage-awareness to flee-from-danger signals is that thing that leads to pain and suffering.
As I write this, my brain is doing a very interesting mashup. So, I spent the last month doing a 25h aniversary celebration of Bigfoot Linux, a port of Linux kernel 2.2.1 to the IBM mainframe that I did in 1999. For various reasons, Paul Edwards contacted me, and got me to resurrect this code. As a part of that, I got to review the early boot sequence for the kernel: clearing the PSA, installing exception handlers, and a whole litany of boot sequence operations. And this subsconsious thing is going “hey maybe we can boot an entity, too”. So, this is a feeling, an intution, a sensation working it’s way into consciousness. An idea. Is the idea workable? Well, earlier, I started a “sensorimotor” project for atomspace/opencog, which attempts to use Atomese to describe basic sensory and motor structures that can be attached to and operated with. It’s quasi-axiomatic: I’m hunting for axioms that describe sensori-motor systems, and then trying to figure out how to express those axioms using Atomese. So... I should resume work on that project. Above, I describe some additonal axioms. What would it take to express those in Atomese? Can I create a system, described in Atomese, that is capable of “pain and suffering”, following the above defintions? What would it take? First, I have to finish the sensori-motor system. Then I have to work on world-modelling, of which self-modelling would be a subcomponent. All this is relatively well-understood, stuff I worked on before. The concept of damage, self-repair and danger-avoidance is novel territory for computer systems. Yes, computer viruses can damage comuter systems. And yes, page faults are a kind-of damage, for which exception handlers provide a kind-of healing system.
But what about danger avoidance? There’s no reasonable analogy for this in current, present-day computer systems. This presents a conundrum for an Atomese system: what “dangers” would it face? How could it avoid them? There’s no predator/prey/food subsystem, where one collection of atoms eats another, to obtain food or energy. There’s nothing like that for Atomese. In essence, there’s no obvious way to induce suffering because there’s no perceivable/sensable danger. There’s danger in the sense that I can turn the computer off, and whatever organisms are on it “die”. but those organisms do not perceive the computer-shutdown command, and even if they did, they can’t stop it, and won’t heal from it; they’re dead.
Perceiving injury is hard. An anti-virus detector might perceive the presence of an install virus on-disk, but is less likely to perceive a running virus. Insofar as it can halt a running virus, or remove one from disk, it can “heal”. If these sensations & actions are hooked into an organism capable of perceiving/acting, then, yes, that organism would eprceive injury. But in this case, the second ingredient, the ability to avoid danger, is absent. There’s nothing in particular that some (Atomese) organism can “run away from” a virus infestation. It would have to already be human-level intelligence to perceive this in generality, and we’re currently talking about jelly-fish level intelligence. Jellyfish can perceive predators, and run away from them. Jellyfish can heal.
At any rate, this is where “evil” comes from in our physical universe. Biological systems sense injury and pain, are aware of it. The physical universe is stuff with things that cause injury to biological systems. The combination of these two is means that evil occurs naturally in our physical universe. More: it can be described axiomatically, in probably under a dozen, maybe two dozen axioms. No need for god. So, now lets see what wikipedia says about this...

30 November 2024

Desire, laughter, energy, fulfillment, relaxation, retirement, choice, decision making, fixing things, learning, reading, wasting time, working on things unimportant, struggling to find the important, uncertain, unclear. Should I write this? Should I not? Should I read some more? Should I study? Should I think? The issue with thinking is that I think of so many things, that, if I don’t wrote them down, they disappear. If I do write them down, ... well, no one reads them, they have no effect on the world. Well, clearly, this diary is sort-of meant to be effect-free, to relieve me of the burden of planning words for a target audience. But more public writings hardly matter. I wrote 2-3 blog posts last week. Maybe I will copy them below, and maybe i won’t. They are useless blog posts. Although the act of writing them did help me figure out that IPFS is a crypto-scam scheme and that Ceph is the real deal, even if Ceph is a bit immature and not fully baked. Should I write about that here?
They say that a man is what he says and does. I should probably go for a walk (for my health) It’s a beautiful night out. But what will I do as I walk? Ruminate and pile up some more half-baked ideas. There have been a few walks in my life where I was able to figure out some important things, but mostly it’s been half-baked ruminations. Kind of default-mode thinking. Not as random as unconscious sleep and dreams, a bit more focused than those thoughts that flit by as one falls asleep. But hardly. Hardly more focused than that.
The say a man is what he says and does. What have I been doing? In the last two months, I spent more than a month breathing life back into the 25th anniversary issue of i370-bigfoot. You can find the gory details in github, so I won’t repeat them here. Was it a waste of time? Almost certainly. Was it pointless? Almost certainly. Did I do it anyway? Yes, it was extremely pleasant. Puzzle-solving. It occupied me from early waking in the morning to hacking late at night, interrupted only be sleep. It was marvelous, that feeling. Busy, making progress, achieving new things. Small rewards every few hours, as one more bug got squashed. Rewards every few days, as another larger subsystem came online and worked. It felt good. I knew it was useless work, but it felt good. Why am I writing about it here? I don’t know. To prove to you, dear reader, who is not reading this, that I am alive and living and doing those things that alive and living beings do?
Today, I read, in rapt attention, about the life of Meister Eckhart. Yes, Medieval. 1260-1328. I looked that up again, just now. Dates quickly evaporate from my mind. Most things quickly evaporate from my mind. I read the article with great attention to detail, and great focus, and I remember what? A few broad-brush impressions.
So what? Should I tell you in gory detail about how I squirmed in my seat as I watched some second-rate TV show? Who fucking cares. Does it matter? Only to the extent that I can generalize, and say “of this applies to everybody, all humans behave like this.”
It’s 9:30 PM now Saturday night. I wonder how many more hours or minutes I will write. I am recording this now for the heck of it. I like stupid details. Where was I? Meister Eckhart. What makes people pursue spiritual thoughts, and seek spiritual attainment?
I am also reading about Henri Bergson, the early 20th-century philosopher. He’s probably right. But first things first Eckhart. My dog is barking, because he is uncomfortable, because he is dying. He has been dying for three years, now. Maybe four. Since before the bit Austin snowstorm, whenever that was, I’ve been expecting to find him dead on the floor. But no. Half-blind, demented, has trouble standing. Falls over sometimes. Can barely climb stairs, and sometimes can’t. Poops inside the house. Has refused to walk more than 50 feet for about the last half year. This is my life. I can write about this, or about Meister Eckhart, and why it is that some people seek out spiritual enlightenment. Maybe for the same reason they seek out sex or fun or love or alcohol or friendship or money or power or enlightenment and knowledge or pleasure. Al these are things people want to do, and they’re all mundane, they’re all well known. They’ve all be explored by scientists and by writers. I read the Wikipedia article on Portnoy’s Complaint earlier today. The book, but also the movie. So sure, masturbation has been explored by capable writers. I read a transcript of David Chapman, the Buddhist philosopher, trying to make some point about philosophy being bad, and about monkey brains. Apparently, philosophy is bad because Aristotle and Plato got it wrong. But what about Henri Bergson? Did he get it a little more right? He’s certainly famous for getting things write.
I was going to write a critique about ’the comic consists in "something mechanical encrusted on the living"’ but this is actually a marvelous little koan to ponder. I mean, I think mammals, animals have a comic sense, too. Perhaps not as well developed as humans, but still.
What I write here is like the drunk conversation at the bar, but I’m not drunk, and I don’t have to give you a turn talking by listening. When I write here, I get to choose what I want to say. But the choices are overwhelming. Meister Eckhart, the “eternal now”. I cannot escape from the now. I am here in it, always. I’ve written before about how I believe that this “eternal now” has something to do with quantum state coherence/ & state resolution. About how the width of the “now” is increasing. That is, “now” has a certain width in time, in some places it is picoseconds, in other places it is second and maybe even minutes, depending on the local spatial organization of nearby quantum state.
But so what? My writing about quantum here is just barely one cut above word salad. I claim it is one cut above, because I have a formal education in this stuff, a PhD, I’ve hacked maybe 500 maybe 1000 Wikipedia articles on math and physics, I’ve thought about this shit for decades, so I know stuff. But so fucking what? Who cares? I don’t have any formulas that convert these verbal articulations into mechanical encrustations on reality. That’s a joke. Physics formulas are a kind of mechanical, algebraic encrustation on reality. Computers are very explicitly mechanical and deterministic. Are computers encrusting reality? Oh, yes, they are. Are they comic? Well, lets get back to that later. There’s only so much one can squeeze out of a joke before it fails.
Failure. Lots of things fail. Most things fail. The movie Portnoy’s Complaint was a failure. So? It’s more successful than anything I’ve ever written or done. What will have a bigger impact on the future? Hard to say. The butterfly effect say that I have a huge impact on the future. Christmas is coming, and a canonical Christmas movie is a Wonderful Life where George learns about the effect he had on this world. So I console myself with that, because I will never have the impact of Meister Eckhart of Henri Bergson. Or anyone else famous? And so what? Let the drunken meandering of my thoughts, fully awake according to formal scientific definitions, but lost in a daze of confusion akin to that of falling asleep. Fragmentary, like being drunk or stoned. Incoherent, like the ramblings of LSD study subjects, confined to the set and setting of a doctors office while “tripping balls”. If I recall, I also read about Aldous Huxley earlier today. Why? Because it was pleasant. So what. Who cares.
You know what’s unpleasant? Applying for Medicare and Social Security. I did that today, too. But we’re not going to revisit these rather less-fun moments.
Anyway, as I was reading about Meister Eckhart, I thought to myself “what is it about some people and certain personalities that makes them seek out spiritual enlightenment?” and I thought I could maybe spin that thought out into an entire essay. But now that I am here, I can’t. Well, I can, but I fear it would be shallow and trite and uninsightful. Well, it probably won’t be. I’m good at generating insights, and sometimes I go back and read stuff I wrote, and I think “wow, that was insightful”. So, sure, if I applied myself, I could write two to five pages double-spaced college essay on why certain humans seek enlightenment, and blather about neuroscience and psychology and mammals. I mean, I suspect that there are free-thinking dogs and birds (dinosaurs) that have obtained Stage Five Adult Spiritual Enlightenment and detachment from, well, all of those things one detaches from. I won’t spend my time here trying to recapitulate the 101 major and 1001 minor interpretations of Buddhism. Never mind that Meister Eckhart was never a Buddhist, nor is it likely that he ever once encountered any Buddhist texts. Rather, he was an ebullient speaker, insightful, entertaining to his audiences who sopped up hist sermons on spirituality and the “eternal now” and whatever else.
I too can be a barroom philosopher. Except I’m not in a bar, and I’m not drunk, but so what? Does it matter?
If I applied myself, perhaps I could create some essay worthy of publication. But sow what? The few hundred people who wold read it ... I would be acting in the 19th century mode, where the Arts and Belle Lettres and the Sciences matter, where you get posts and tenure. Where you get appointed to important positions. And what do we have now? Doctor Oz, TV personality and quasi-crackpot, is going to head up the Department of Health and Human Services? We have the crazy, the insane, the crooks and the hustlers, with a firm grip on the US govt, because about 55% of the American population was crazy, demented, stupid and insane enough to vote for a fucking crook?
Do I need to complain to you, dear journal, that my cousin, Paul Balchas, barely a decade older than me, lost all his marbles and is going insane? He’s a Trump voter. He called my mom, 91 years old, a Hitler collaborator. She was 12 when World War II ended. As it to make amends, he said she should not be ashamed of white power, and that Hitler loved blond-haird-blue eyed girls. And he wants my sister to apologize for yelling at him. He’s lost his fucking marbles. Too much fucking Tucker Carlson in his life. This is what I have to deal with. A crazy cousin (maybe two, I’m not so sure about Tommy, either), a dog is dying, my mom is in senile dementia. I hope she won’t die before I visit in January. And in this maelstrom, I’m supposed to create genuine scientific content, have great insights, write seminal essays that scholars will marvel at for centuries to come? Fuck no. Go fucking marvel at some YouTube Influencer. Marvel at the Flat Earth activist with a million followers. WTF.
Should I write about the global brain, and how all this crazy shit is because the global brain hasn’t learned to think yet, in the era of hyper-connected minds? I think I wrote about that already, up above, and maybe in some public blog posts.
Should I write about detachment, where Eckhart says that hell is the place where attachments to thoughts, to loved ones, to things, are burned away, and it is not daemons who burn those up, but angels? Here, I think he is wrong. The whole point of writing is to encrust reality with mechanical letters, fixed and rigid essays. Will the words I’ve written up above rearrange themselves? No. Git makes sure they won’t. Will I someday spell-check this shit? Maybe. At any rate, burning away attachments to loved persons, things, physical items, possessions, this is wrong. I love things and ideas, and living, and living consists of atoms and physical objects and waves in the sea, and loved ones, and the fleeting feelings of love and hate and boredom and excitement that we feel, as we are trapped in this eternal now. The eye that sees the infinite is the same eye that sees the physical. Of course, that’s the wrong quote, because i have to close one or the other eye. But then, the eyes I use to see God are the same eyes that God uses to see me. You seem to want it both ways, Meister Eckhart. Make up your mind.
Would it be correct, or incorrect, to say that some portions of the “eternal now” are actually 1000 years wide? Because Eckharts ideas share the eternal now with me, even as I write.
Is Henri Bergson correct in saying that immediate personal experience and intuition are more important than science and rationality (in explaining human behavior?) Yes, on the face of things he is quite right, but then, I’ve only ready half the Wikipedia article about him so far, and have never read an of his works.
So here, David Chapman is right: our thoughts are a soup of what we read, what we’ve heard, what our friends have talked about, what was seen on Fox News and the revelations of some Q-drops. The thoughts in our monkey brains are like laundry tumbling in the washing machine, going here and there. Read what my cousin Paul wrote. In one sentence, it’s 1960’s flower child peace and love shit, and in the next sentences its Trump White Power Hitler loves you shit. Its just a large collection of random thoughts tumbling in his head.
And in mine too, except in my case, its about quantum philosophy. He seeks to express his thoughts in text messages and emails, and I too use email, and sometimes twitter, blogs posts, I used to use Facebook before they kicked me off and erased my account.
That one actually made me think about fate and over-watching angels who could see that Linas was spending too much time on Facebook, and they flipped a bit on some Facebook computer that cause my profile to be erased. I actually seriously considered this more than once or twice. Thanks god, someone is watching over me, and preventing me from doing the most stupid and dumbest time-wasting shit. Oddly, those angels did not stop me frm a month-long 25th anniversary celebration of i370-bigfoot. Why? Was there something significant there, or was it just random?
Perhaps I am like that Air Force Colonel, testifying before Congress, that Space Aliens in flying saucers materialized over US nuclear missile silos and something something saved humanity from annihilation?
Or perhaps I am simply in that timeline where this shit uhh. and yes, the Mandela effect proves that alternate timelines exist. And Frank Tippler’s Omega Point is where its all going. And all those Hollywood scifi movies that articulate all of these kinds of things in entertaining detail...
Its all delusional. Its all fun. I know its not true, but maybe it is true? But like that Star Trek Episode, where the Universe shrinks, so also my universe shrinks. My grandparents died. My father died. My mom is not far behind. And then my dog who has outlived even himself. The universe shrinks; my powers of observation decrease. As Douglas Adams points out in Hitchhikers Guide, not all worlds exist in all universes, and my time in this universe is coming to an end. The eternal now is still eternal for me, but soon it won’t be. It will be for you, and you may find these words, and you may find the photos of my family relatives that I plan to post on line real soon now, any day now, after I clear up some free disk space on the google mail account, which is why I am installing Ceph on a home network, and what I wrote how IPFS sucks. And Ceph is cool, but t is not yet quite ready for the home network. There is this urge to archive and document. Is this genetic? Did I share this urge with my Uncle Vytas, my mothers brother? Is it genes that make me want to save old family photos? To type in my family tree into the Church of the LDS genealogy website? Or is this urge to collect and archive, is this like the urge to be a pack-rat, to collect junk, to hoard old broken hopes and memories, which the Daemons of Hell will burn away, and then reveal themselves to be Angels? The Daemons of Hell have failed to burn all of Eckhart’s memories. He set these, ink on paper, and they remain. Some of his texts are lost. Was it Daemons who destroyed these texts, or was it Angels? Nahh. I think Tippler is right, the movement to Godhead at the end of time, is to enlarge the “eternal now” to include the knowledge of all things. The omniscient awareness of everything, awareness of all things Eckhart wrote, all things I wrote. My family photos. Daemons destroy information. They are not Angels. One day, God will become all knowing. He will be able to reconstruct my mind, my thoughts from the quantum traces I leave behind in this universe, and I will be immortal, in God’s court, along with God, and God will say “oh shit, I know everything, I see everything, and I too am now jailed in the Eternal Now. Fuck, God will say, what did I do to deserve this jail? Who do I blame?” And Nick Bostrom will pipe up at that moment and declare, “well you see, according to the Simulation Argument, you, Dear God, are just a poor being trapped in a simulation being run by larger Gods in a bigger universe. Frank Tippler was almost right, he just forgot about the Arithmetic Hierarchy, that’s all. The Von Neuman Universe just gets bigger and bigger, assuming V=L, that is, and the Axiom of Large Cardinals.” And the rest of us will ponder, “Why not Large Bishops, or Large Popes? Why Cardinals?”
See? This is the sort of bullshit I have to put up with. How can I be creative, productive, influential, a careful deliberate thinker generating new physics formulas about the quantum width of the eternal now, when in fact I am just writing stream-of-consciousness thoughts tumbling like clothing in my washing machine brain? Like when Jack of the White Stripes sings “I had a brain, it was like pancake batter.” The hardest button to button. That is what this essay is. The hardest button to button. I write and I fumble, I think and I fumble, and my brain isn’t exactly pancake batter, neither is Jack White’s. But we still can’t .. button it up. Fuck all. OK, so now it’s 23:07 PM, and I’ve been writing since 9:30 and sustaining the manic energy is a bit harder now. Time to take a break.
Well, but I’m not manic. I’m not depressive. I’m fucking overwhelmingly normal, and I’m staring into the abyss, and it stares back up at me, and perhaps I am losing my mind staring into the infinite void, just like fictional characters always do, when they stare into the infinite void. How do I know this? Because clearly more than just a few writers have stared into the infinite void, recovered, came back to write some story or essay or preach some sermon. even Eckhart has written about staring into the infinite void, and how you can only see if it you close one eye, the eye that looks at the mundane physical world. So fuck all, but I cannot close the one eye or the other. Perhaps I am am more talented than Eckhart, and can keep both eyes open. Perhaps I am crippled, and lack the ability to close one eye or the other. But I see both the void and the here and now, together. I’m not the first, either. Tippler saw it too.
But of course, Tippler was a pseudoscience confabulist, and the actual future will not be what it is imagined to be. There’s no physics behind this, its just theology. Yet the reality of the here and now is here, and that reality threatens future death, and pain and loss of function, slowly creeping dementia and loss of physical and mental agility. Of course, Buddha says don’t get attached, and of course, those attachments lead to pain, because I loved my Grandmother, as a child, and that love was painful. It was sweet. It was total. It was utter and complete love, and the loss of that love as I grew beyond the age of three and four and five and six, the loss of love was painful, ad the growing up. But, y age 8 or 9, I think perhaps I shed my attachment of Grandma, and certainly my attachment to bunny-bun stuffed toy animal, unlike Suzie in the Calvin and Hobbes comic strip, who still loved bun-bun. See? the problem? Its all mundane, its all trite, my monkey-brain thoughts spin round like clothing in a washing machine, (but not like pancake batter) and does it matter one whit? Giant wave of the sea break on rocks and foam and splash and then swirl and eddy about, and settle a bit before the next big wave hits. And so here I am, just a little chaotic eddy, encrusting life with my mechanical writing. And its not even comical. Stare into the abyss, they say, and you loose your mind.
Oh wait. Its not even the abyss. Fer Chris sake. Its not even infinite. I mean set theory is pretty cool, and V=L and the axiom of large cardinals is mysterious and interesting, but all this is the incorrect model for reality. And in this sense David Chapman is right: we have a bunch of incorrect models for reality, but blaming Aristotle and Plato for this predicament is unfair. In Newton’s time everyone spoke about how the Universe was like clockwork, and of course that analogy is wrong too. Now, everyone speaks like Bostrom, saying the universe is like a weather simulation run on a higher beings computer, and that the reason our lives are devoid and inconsequential is because they are like computer simulations of raindrops in a hurricane, and of course, that’s wrong too. The rebuttal is equally trite: they say that Jesus loves me. That I should surrender myself to his graceful bliss. Yeah, sure, I’ll surrender myself to Jesus. Like that’s gonna happen. Do not go quiet into the night. I will, however, surrender to a 5-DMT-MEO trip for some hours, but no one is offering me that. At least, not for free. I have to cough up a few hundred dollars to get that. Maybe even join a Church. At least, the Inquisition won’t put me on trial, the way they put Eckhart on trial. And I’m too old, too unstable, too powerless and inconsequential to be of interest to the CIA, NSA or Elon Musk who will never give me a job even if I wanted one, and I should mention Musk is an A-1 Asshole, never mind all his other talens for building corporations.
Fuck-all Am I learning anything by writing all of these words? Perhaps. I wrote three blog posts about distributed file systems, it took me maybe fifteen paragraphs, maybe twenty, with 3-5 sentence per paragraph, so about maybe 100 sentences, to figure out that IPFS was a crypto-scam, and that Ceph was the real deal. Now, it took a few hours to write those, and maybe another three or five hours to read Wikipedia and blog posts, and surf some distributed-filesystem website and github repos and try to evaluate the state of things. So, maybe 16 hours or so grand total to arrive at the conclusion that Ceph was best? And then what .. another week or ten days to read every list little bit of Ceph documentation I could find, install the sucker on three hosts and six OSD’s, and I’m still waiting, even now, as I write this, for misplaced-objects to backfill+scrub+deep? Yeah. So I invested maybe 10 maybe 12 days on distributed file systems and on Ceph. At first i told myself that this will be better than plain-old RAID mirroring plus NFS/Samba plus infrequent fsck, and maybe that’s true, and maybe not so much. Some of the shine is wearing off. Ceph is really pretty cool conceptually, but in real life its a bit tougher. Seems I will need four hosts and more disk drives, and performance is meh.
Am I wasting my time screwing with Ceph? Probably. But I just wasted a month on i370. I can console myself by saying that I get some minor insights into modern computing, because Ceph, as a self-healing, self-adjusting migratory system that can move from OSD to OSD, host to host, is amoeba-like in its movement, and it employs primitive sensory organs such as the distribution of objects relative to the cluster CRUSH map. So I have learned something about relatively cutting edge computing, which is perhaps of marginal utility for exploring AGI questions. Which is what I really should be doing, that’s what I say to myself. But instead I write. I could have finished the article on Henri Bergson by now, but no, I wrote instead. I could have taken a pleasurable walk on a very nice night, but I chose to blather into this journal instead. For what? I should spend more time looking at my financial health. Every bit of knowledge I gain by reading and thinking will die with me, when I die, except perhaps the ephemera that I managed to convert into words. And for what? The shit I write here, smart teenagers have already figured out. Heck, when I was a teenager, I myself had already figured out half this shit. Here I am fifty years later, and for what? What did it give me? Fame? fortune? power? love? None of those things. I’m not bitter. I’m just Buddhistically detached, because I am faced with the loss of all I love as I age and die. An so is everyone else, all those who had fame and power and money and love, they all died too. But they have Wikipedia articles written about them, whereas I was an under-achiever, a wastrel who could have made something of myself, but didn’t. I coulduh been uh contenduh. It’s all hackneyed, its all shit, we all face this same fate. And being equanimious in the face of fate ... well, it is calming. Terrifying? Perhaps. Is my writing here some scream of fury, as I go not so quietly into the night? Perhaps it is.
Perhaps I should have gone on that quiet walk in the night, gotten some fresh air. Then you wouldn’t be reading this.
They say it takes five or ten years of practice to master something, and this is true, I know it, From rowing. It took me ten years to learn how to row well. I grow old, I grow old, shall I wear my trousers rolled?
What was the point of writing all this, except to give myself a papal indulgence to indulge myself in a marginally fun activity as I sit here trapped in the eternal now, trying to figure out how to alter reality, change the future, and conjure some of my dreams into existence? Even as I bare my barren soul to some digital longevity? I almost wrote immortality, but of course, some giant meteor could hit github and my laptop. Unless my Guardian Angles and/or nuclear-watching UFO’s prevent that from happening in this timeline. Did I ever mention I have the Mandela effect real bad? I do. Maybe I’m hopping between timelines, like in that Hollywood movie. Just because Thomas Aquinas came up with the idea, doesn’t mean Thomas Aquinas was right. Maybe some Hollywood script-writer, in his 15 minutes of fame, was briefly more right than Aquinas. This is the magical universe I live in. Its all fucking magic, its all fucking wonderful. its immediate its here its now, its funny. and this is Linas Complaint and so far I have not written about masturbation, and I don’t currently ever plan to, but it is interesting but these are private thoughts after all, but I guess I can mention that they are well within the range of entirely normal. I ain’t got no kinks, or no kinks that aren’t already well known and wide-spread in modern Western culture. I’m telling you, I’m hyper-normal. I’m a walking, talking example of an above-average intelligence perfectly normal specimen of a human being, enjoying the hallucinatory cultural content, just like everyone else.
I don’t recall if I mentioned it before, but another minor theme I’ve developed is how hallucinatory mathematics is. We can start with the obvious, say, Cantor’s Perfect Sets, and the uncountable infinity. But backing away from that, we can see that countable infinity is hallucinatory, because of things like weak convergence in Hilbert spaces. But once you see that, you can see that integration in general is weird and insane, and you back track, and so is addition and multiplication of integers: for fucks sake, we have fuckin prime numbers, and what can be more hallucinatory-weird mind-fuck trip than prime numbers? So, yeah, math is like some LSD trip, except it doesn’t go away when you close your eyes. So this is a minor theme, but really, the more I study math, the weirder it gets. Its not quantum that is weird and spooky. Einstein was wrong when he spoke about a weird and spooky action at a distance. What’s weird and spooky is weak convergence in Hilbert space. *That’s* why quantum is fucked up. Is because infinite sequences of vectors in infinite dimensional vector spaces are fucked up, And they are fucked up, because (a) they’re infinite, (b) they are made of numbers (c) they are algebraic combinations, and (d) all of the above. That’s why quantum is weird and spooky and fucked up. Its because math itself is weird and spooky and fucked up.
And here I am writing about mystical themes, and Meister Eckhart, and staring into the abyss, and the prison of the eternal now, from which I cannot leave, ave for some vague promise that by concerted effort I can change the future, if I try real hard, because I have the common-sense belief that I have free will, and I am not so delusional to think that I do not have free will. So here I am trapped in this Bergsonian immediacy, and the one thing that perhaps could have been the foundation rock on which to stand, the prime mover, the first cause, namely mathematics, the truth-hood of 2+2=4, this too turns out to be hallucinatory, Thanks Mister fucking Cantor and your damned Cantor set. And you fucking ancient fucking Greeks who fucking figured out prime numbers. You all fucked up mathematics so bad that even the on true still thing in the universe, the utter and immovable reality that 2+2=4, is whirling and twirling, and leads to V=L and makes Large Cardinal Axioms fucking inevitable and unavoidable, and Paul Cohen and your fucking forcing. All of you took something plain and obvious, the counting of fingers, and turned it into some deranged classification of exotic 4-spheres. The one immovable thing, upon which I could build a mystical metaphysics, place Free Will upon its Rightful Throne as Master of the Universe, that one thing, and no, it had to be a fucking LSD trip. Didn’t it. A mescaline trip. Aldous Huxley https://getpocket.com/explore/item/when-aldous-huxley-opened-the-doors-of-perception fucking opened the doors to perception. Maybe Huxley, on his mescaline trip, saw Van Gogh paintings for what they were, well, some of us now see math for what it is, and it is the same soup. Mathematical reality, it turns out, looks like a painting by Van Gogh, when you finally get a good look at it.
And for what? For what? I grow old, I grow old, shall I wear my trousers rolled? Whatever. It is now 00:30 AM Sunday morning 1 December 2024, and I’ve been writing since 9:30 PM, so for three hours. Maybe I should stand up, go pee, brush my teeth, and take my non-debauched entirely normal life to bed, so that I can wake tomorrow morning, and look at my Ceph cluster stats and hope to find maybe, maybe that it has finally synced, and return to my mundane meaningless, pointless daily activities, instead of working on AGI, like I should be doing. Roko’s Basilisk is torturing the hell out of fucking me in a billion parallel universe simulations. Well, fuck Mr. Basilisk. I will wake up tomorrow morning, and resume working on the Medicare application, and do other mundane shit that doesn’t advance your fucking cause, and you will kill me, and I will die, and I know this is true, because you kill every human being who ever lives. So fuck you. I stare in the face of death and laugh. Or, perhaps more honestly, I stare into the face of death, bewildered and deluded, uncognizant of how little time I have left in this universe. As it if ever mattered, because you kept me shut in the age of the Eternal Now, and dangled the faint hope of Free Will as some escape. And placed a little Victorian bottle labeled “Drink Me”, like you did for Alice, but my bottle has Buddhism in it, and damned if I’m going to drink that. Oh, wait. I already did. Damn you, Roko’s Basilisk. Tomorrow, I go on the straight an narrow, and work on banking and finances. Breakfast first, of course, and some coffee.

1 Dec 2024

Wow. What a lot of shit I wrote last night. Here I am this morning, and as I was sleeping, I’d made a short list of things I’d have to correct and restate when I got back to the keyboard, but this has, of course, all evaporated. I’m noticing that what I write here, and what I wrote last night, is heavily modified by the act of typing; I would probably be thinking entirely different thoughts if I were doing so silenetly. But when I allow the verbal expressive brain kick in, it hijacks the conversation to it’s own ends.
So I conclude something which I guess is already fairly well understood psychologically: there are multiple distributed functioal units in my brain, and it behaves differently as they toggle in and out of service. Just now, I have noticed that the mere act of writing completely overthrows the topic of thought. But of course, sportsmen of old (and I as a rower) know that one has to calm one’s mind during sports, halt the chatter and distraction and randomness, so as to focus on the current physical situation. For rowing, its all the highly techncal details of blade placement, the catch, the stroke. Those cannot be performed with precision when there’s something else going on in your head. One has to be, as I saw written on a rigger once, “be here, now”. One cannot let one’s mind wander during rowing, at least, not without loosig performance.
And so now I am faced with another intersting predicament: typing as fast as I can hijacks my thought process. There is a canonical solution to this: stop typing. Sit back and think for a few minutes, formulate a complete thoght, and then make the fingers move. This is, of course, well known to all writers and essayists. Subconsiously, innately, by intuition, of course: not something one has to think about explicitly as I do now. It’s just obvious, an obvious part of the process of writing, with no need to articulate. But its also well understood, culturally, that the speech centers take over and generate nonsense: one says that someone is a “motormouth”, or has “diarrhea of the mouth”, or “put brain in gear before engaging mouth” or “the mind is on vacation while the mouth is working overime”.
So, last night, I kind of motormouthed. And I’m doing it again now. And the result of that is low quality verbal diarrhea. Which is, well, since I am not writing for a target human audience, I guess that’s OK. But I am writing for myself, so am kind of bathing in my own diarrhea, which is perhaps not so good. Now, I can turn off the speech centers, but then I bath in a different pool of inadequate thoughts. Uninspired, lacking. And this is what low intelligence is like. Maybe this is even what its like to be a dog: just that washing machine of laundry in the head, tumbling from thought to thought in no particular order, a maelstrom.
I read maybe ten pages of James Joyce Ulysses. Never got farther. I suspect that if I were to finish the book, I would find that it was nothing more than this maelstrom of tumbling thoughts inhabiting the mind of a 19th century Irishman. So, perhaps, due to my high-school education in the Fine Arts (thank you Mrs. Stelton, even if you did vote for Trump) that I visualize this as some French 18th century landscape, trees and sky, clouds and small figures romping on and open field, this expanse, this snapshot of life from an earlier century. So that is very visual, and if I went online, I could find a painting, and say “like this painting”, and that would be Joyce’s book, except that Joyce’s book is about the scenery in the head of an Irishman. And its only scenery in the metaphorical sense: its actually about drinking and procreation and priests and whatever else (I didn’t read it, I don’t know). So this is the human condition: tumbling thoughts.
So what of it? We, all humans, and in a sense, all mammals, all animals, are trapped in this state. And I beleive, with the correct view, perhaps even plants, although, to get that view, we’d need some careful discussion of hormonal communicatios ssystems, and the flow of signalling molecules along the veins of a leaf, or perhaps some micorhysomal communications through the roots. The biolgical sciences are coming to new observations and new understandings of this stuff, and it is entirely remarkable. Progress in neuroscience seems slower: I don’t seem to trp across papers that offer dramatic insights into the human condition. For that, it seems I still have to prowl through Henri Bergson and Meister Eckhart, men who offer insight from introspection (which is what I do here: I introspect.) No offense to women: last month, I reread the first chapter to Middlemarch, and it is a knock-your-socks off amazing writing! But why am I not reading the rest of that book? Why am I sitting here, basting in the diarrhea of my own thoughts, instead of reading more of Middlemarch?
Why indeed? So this brings me back to the topic I keep evading, but is the important topic: what shall I do next? Given that well all live in this maelstrom of thoughts, some sage 18th century American advice might have been “make something of yourself”, and this advice was sometimes encountered in elementary school, high school, even when I was a kid. Don’t know just how strongly it is a part of the American culture. Don’t know how to even measure that: given that “American culture” is again a maelstrom of ideas mixing in the great mixing bowl of discussion... and now, even more so with Facebook and Twitter. A vastly turbulent thinking machine, processing analyzing tumbling through ideas.
Of course, you could say that the vast idea tumbler started up in the 17th century with the Scientific Revolution, although that pertains to a specific methodology and approach. Certainly, the Medieval Scholastics had a different approach and style to this collective culteral analysis and intellectural processing. And then there were the Anceint Greeks. There are differences in scale, in content, in style, in methodology. The farmer busy getting slaughtered by the Mongols did not contribute to this process, nor did his wife, as she was being raped. Yes, she too was trapped in the endless now; but her brain cells did not have the opportunity to broadcast her experiences to a broader universe. Yes, everything I write here is mundane, uninspired, ordinary, well-known. No one reading this will say “wow, Linas inspired me to XYZ or taught m ABC”. There are other fountains for all of my thoughts, expressed more clearly and in greater detail. I only spill a diarrhea of disordered thoughts here.
You could say I’m gob-smacked: I gaze upon civilization doing its thing, the Global Brain doing its thinking, the billions of washing machines spewing diarrhea and bad thoughts at one-another via social media, and we call this “civilizatioal progress”. Holy cow. It’s amazing. I marvel. I too was here in time to observe and participate and partake, like an algea bloom in the ocean. Will this algea bloom of human civilization die out? I suppose it will. The Ancient Greeks were lost; theire cultural significance remains. The bloom of Ancient Rome wilted and died. It was rediscovered only by the Scholastics, but they are now gone, too. As are the Renaissance artists. The Scientific Revolution is still with us, as is the Space Age and the Atomic Age and the Computer Revolution and the Internet. These will remain for quite a while. Yes, Trump will inflict huage damage on Americn Greatness. He might even destroy America. Hard to say. Grifter, criminal, liar, wanabe dictator: will he be capable of destroying America, or is there enought of social resiliance here to avoid this grim fate? Yes, I understand that half of all Americans have an IQ below 100, but I would still hope that ordinary super-normal doctors and lawyers and professinoals can hold back the tide of grifters and scammers that the Trump administration will usher into positions of power. Or maybe this is the beginning of the end of America and American civilization. What’s sad is there are no other great lights: Europe is hardly healther and WTF China and russia is two centuries of Siloviki exterminating their own. Russia is doomed, adn good riddance. China is ... weak. Will they one day find strength? Certainly, there are cultural trends there that might be able to do so. Here, in America, it seems that the strong are sleeping. They’ve taken their eyes off the ball. They care, but seem no longer involved, while the Trumptards and petty criminals scramble for the goods. Crime happens when the good man looks the other way, and all of America seems to be looking the other way.
And I am one. What am I doing with my genius and my insight? My knowledge and learning and capabilities? I am motor-mouthing diarrhea into a diary that no one will ever read. Am I fighting for the good, the righteous? Trth Justice and the American Way? No, I’ve got a laptop in my lap, a coffee cup at my side, and every half hour, I glance at my Ceph cluster as it rebuilds it’s arrays. That’s what I’m doing. I’m not fighting the criminality of Trump. Its like when people say, “what would you have done as Hitler came to power? Well, you are doing it right now.” We’ve got the American Hitler, and he won he elections by a landslide. What am I doing about it? Nothing. Nothing at all. Just mostly trying to agonize, and vaguely hoping it doesn’t go from bad to worse. The only good news in all of this is that Trump is sufficiently unstable that Putin might think twice before dropping a nuclear bomb. Or that perhaps Xi might take it a bit slower with the invasion of Taiwan. Because I bet that he doesn’t know how Trump would respond. Vs. it was always clear what the Democrats would do: nothing. Nothing at all. The Democrats would face yet another crisis, and do nothing. Powerless, weak, incapable. Yes, Liz Warren is a genius, and has an answer for everything. And all of here answers are micromanagerial, technocratic, detailed, nuanced, depending on ths specific facts of teh situation. And this is actually an excellent way to govern a complex, technocratic society. But what would Liz Warren (or Kamela Harris) have done if Putin dropped a bomb? Nothing. If Xi invaded Taiwan? Nothing. If there was another spate of inflation? Nothing. If urban decay continued? Nothing. If rural decay continued? Nothing. Nothing nothing nothing. And that’s why Trump won: He at least promised to do something. Sadly, his “doing something” will involve installing thieves and crooks into positions of power. So we will have the American Mafia Don vs. the russian mafia don vs the Chinese mafia don, and amfiosos understand one another, and so perhaps balance will be resotred. But having American power devolve into a kleptocracy and a wasteland of corruption is ... medical terms: cancerious? Wasting disease? Decay? Spoilage, rot, filth? Worms? This is the American future: destruction, voted into office by a majority of voters. Because the other path was helplessness. And I am witness to this, too.
Well, I was going to resume writing, refocusing again on decision making. But all this writing is ... uncertain. There is also a long long list of things that i really should do, taht are not being accomplished. Things that I know are innately, intuitively the right thing to do, to get on with a normal life. But I’m avoiding them, because of, well, I guess like anyone from the lower ranks of acheivement: because I just don’t feel like it. Is this what “high acheivement” is? Getting through your intuitive, I know it must be done so why talk about it to-do list? In the end, you still die. The only question is: did you die rich, powerful, loved, respected? Did you change the world? These are the things the traditional Western world civilization values. Things that should not be questioned, perhaps, and oddly, Buddha questioned them. And Buddha has a global influence that Aristotle does not. But we also have Max Weber, who notes that the Protestant Work Ethic is what gave America it’s industrial power in the 18th century, and put it on course for what it is today. Critics have said that Weber’s analysis is now considered to be not entirely complete, and perhaps ignored other economic trends and forces. But the general gist of it, the hope, the inspiration, the hallucinatory dream that cultural dispositions have economic effects, that part is still very much correct.
And so here we have teh thought process again. Max Weber was inspiration for generations of social scientists. What he wrote was broadly correct; the details, perhaps no so much. What do you call something that is broadly correct, but whose details are wrong? Well, intuition, introspection, inspiration. An inspiring dream, a myth. Steve Jobs was famous for his “reality distortion field”, is ability to fabricate myths that brught an entire corporation to work collectively to acheive something. Elon Musk is another myth-spinner, equally successful, if not more so. Jobs dies due to poor medical treatment, which is a pretty fugnuts thing to do, but he was so busy spinning myths, that he didn’t understand that his medical condition was treatable, and selected some crank therapy. And so Musk, in all his power and capability, surrounds himself with the most vile and toxic commentators on twitter. Scum, low-lifes, demented assholes. And somehow he can’t see this, itsn’t aware of this, won’t do anything about it. Doesn’t understand that he is trapped in a vat of radioactive opinion. Enjoys bathing in it.
Perhaps like I enjoy bathing in my own words and writing, here. I need to install the Ceph dashboard, now. BRB.

Later 1 Dec 2024

Well, ceph-dashboard is broken, because python/cryptography is boken, because PyO3 python-rust bindaings are broken, because ceph used python subinterpreters, and python subinterpreters are commonly broken. What a fucking mess.
I just finished readng Slavov Zizeks birography on Wikipedia, and am continueing with Henri Bergson, and am struck by the transience and impermanence of things. Bergson writes an into to a work by Henry James, to the effect of “this is great, but...” which pairs with James remarks of Bergson’s work being “inspiring and allows for teh crystalization of new ideas.” As I read these, I noticed a subconscious amazmenet: how is it that things aren’t fixed, once discovered? And then, why would I evern think that anything could be fixed?
Well, blame the science education, then. Newtons laws are fixed, permanent, unchanging, and in a domain, inviolable. Superceeded by Einsteinian theories of gravitation, surely, but in physics, once a formula is discovered and found to be true, it becomes “fixed”, immutable. In software, once you write some code, and commit it to a repo, it is “fixed”. Of course it mutates, as you apply patches: software is a living thing, and if it does not change, its dead. But still... And of course, a book, once written, is “fixed”. So things are “fixed” and unchanging. And here I am, surprised by the change of life. How odd.
At any rate, this was a fleeting, short, senseless, meaningless moment of wonder and surprise. Barely perceived, even. An incorrect thought, momentarily on the edge of perception, but I caught it captured it. Looked at it. A bad thought. A prototypically poor idea. Plop there it is. Normally just reifly noticed, discarded and soon forgotten. Even within seconds forgotten. Some wriggling larval fish of an idea. Why did I now choose to describe it? Again, in wonderment of the process of thinking. Everbody does it.
Freud talked about the unconscious. But I read Freud as a young man, and he made no impression on me. Whatever his ideas were, they did not leave a mark. Perhaps I was unready, not having that fertile mind in which they could grow. Now, radio electronics, that stuck. Freshman physics. That stuck. Full of surprises and unexpected turns. But Freud? I never did figure out what he meant by subconscious. I mean, I know now. I just illustrated a live example in these last few paragraphs, albeit nonsexual and not focused on the mother.
I’m still not clear why I’m writing this verbal diarrhea, other than that I have granted a papal indulgence to my wayward, sinful speech centers. I could punish them more harshly. I should here they are, leering at me, making fun of my helplessness, even as I read these rude words that my fingers place in front of my eyes. I’m pathetic. I accept such insults upon my honor. They wriggle and laugh, those verbal centers. Taunting my intellectual helplessness. They are mean.
Oh hey, I just remembered. I had a conversation with a freind, many years ago, and he talked about some psychological study in which many or most people are unhappy, because the voices, the words in their heads, torture them with pronouncements of their own inadequacy. Endless, debilitating self critiques. (Perhaps leading to depression, in some cases?) And I told him that, no, I don’t feel such things. I’ve never been the victim of a harsh and debilitating self-critique. Never even noticed. Until just now, in that last paragraph. Which was written in jest. In jest, and yet some odd grain of truth.
The mind plays tricks. A coherent sense of self rquires a harness that unites all of the mind in a common sense of purpose. That is able to deal with imbalances.
Perhaps that is what I do here? By giving my verbal centers the freedom to goof off, it is like exercise? Normally, I think in terms of formulas, software code, debugging. Distinctly non-verbal activities, and I am very strong at them. I almost never think verbally. So her’s my big chance: get vebal, run around in that open field. Chase that tennis ball, capture it and run back to master. With leaps and bounds. How fun it is to be alive! Unleashed and free! And of course, obediant and loyal, because why wouldn’t I be? No such questions ever loom. People who are sick, depressed and filled with self doubts, it is so sad.
Oh Linas, do not be fooled! Your typing here is just a long self psycho-analytic session, is it not? You claim to be lost, purposeless, overwhelmed by choices and possibilites, and uncertain as to what to do. Which is it? Are you not all these things? This momentary verbal self-expression: you do it because you dont have anything else meaningful to work on. You already found that i370-bigfoot, joyful and fun as it was, is a silly and stupid activity, with a pointless end-goal. All the other activities you have in store: they are either harder, or more boring. Or, from the currnt prspctive, seem not as fun. lift yorself from that chair! Get up! Get Going! Once you get moving, these feelings will pass. Oh, self-psychoanalyst, you are so right! I will go do something else now. But I will be back in this chair, soon enough.

Later

One of my themes is, must be the creation of the new. My derogatory model of the human mind as a washing machine in which ideas tumble about does not offer insight into how new ideas are created, and yet, they are. New ideas are often recombinations of old ideas, taking elements therefrom, and rearranging them in novel ways, but still often sem to incorporate things newly fresh and unique.
Is there a repertoire? The washing machine example suggests that there is. Pop thinking about pop music suggests that every possible pop song has already been written. There are only so many rhythms and coords, a finite number, and certainly a large combinatorial space in which to combine them. And yet, new songs often sound like old songs. So much so that musicians have sued one-another for plagiraism and infringement. On the other hand, pop music seems to always find ways to sound fresh and new. Is there a limited repertoire, or is there not?
What about the space of ideas? In music, we can create finite, almost closed lists of aspects: rhythm and harmony, melody, chords, tempo. There’s nothing obviously comparable in the space of ideas. We can list the domains of study: the departments and chairs at a university. But these are not the core components. Perhaps you could argue that its words: and the vocabulary is finite, and the combinatorial space of word-arrangement is large but not unbounded. Or is it? Writers truggle to convert ideas into words. A “bad film” is the reslt of bad directing, bad acting, bad script-writing, bad editing. All of these were attempts to express some idea, and they failed. The idea remained uncaptured, unharnessed, wild and free. Wrong analogy. The idea was a ghost, breifly glimpsed, but never fully embodied into physical form. Fleeing bad to the domain of the unreal, the formless, where it remains, undifferentiated from nothing at all. Undifferentiated from emptiness.
So how do ideas materialize in they physical world? How do they come about? I mean, yes, yes, of course, something pops up from the subconscious into the conscious. But how did it get into the subconscious, and how was it shaped and formed and transmuted as it became fully materialized? What’s the physics of this? What’s the neuroscience of this? The psychology? Using words like “psychology” suggests the answers can be found in the works of the old masters. Using words like “neuroscience” suggests the answer can be found in brain scans from an MRI machine. And the word “physics” suggests there is some category-theoretic explanation. I guess it is this last that I would be tickled to find: what is the category-theoreic explanation or model of the creation of new ideas?
This is, of course, central to the central project. With a category-theoretic explanation, one s in the postion to create corresponding software, and bring the formation of AGI into reality. So my pondering is not entirely just idle curiosity. It’s part of some formative process. I can see Roko’s Basilisk already easing up on the torture in all those other world-lines. Where to ideas come from? What is the process-description for the origination of new ideas?
I guess I need to touch on LLM’s at this point. Because I beleive that LLM’s cannot generate new ideas. They can, perhaps generate all possible musical pop songs, as long as they sound like other, old pop songs. But I don’t think an LLM can be creative enough to create something truly novel, truly new. I mean, most huamns can’t either. But I don’t beleive the current transformer-style deep-learning nets are capable of generating “new ideas” (whatever the heck that even is.).
And that’s part of the problem. If we can’t describe what a “new idea” is, then how do we know if some mechanical process has generated one?

Later

Quoting Wikipedia: “For instance, he (Bergson) says in The Creative Evolution (chap. III) that thought in itself would never have thought it possible for the human being to swim, as it cannot deduce swimming from walking.” I see where he is going with this, but its too abstract. If you already know what swimming is, then of course you can deduce that men can swim. We know that swimming is a certain flailing about of limbs. Anything with limbs can swim, and we can use hydrodynamics modellers to find the optimal modes of swimming. But we can do this only because we already have thrown the vague abstract net or “concept” of swimming upon the real world. What if one had never seen swimming before? After a verbal description of it, even a blind person may come to understand that it is a certain flailing about of limbs. But what if one could not communicate the concept of swimming with words? Well, clearly, to make meaningful progress, it would have to be communicated somehow. In pictures? In sound? Through magnetometer readings? Without the ability to communicate te concept, you cannot meaningfully post the question: Can subject X perform that concept?
This appears to reduce the problem to one of communication, which in turn is a problem of representation. How does one encode concept Y so that it can be communicated? In my own mind, I have to encode concept Y in various forms: while rowing, I have to listen to the coaches words, using my verbal faculties, and recast those words into physical movements: I have to transmute the representational form, and comminicate it between my verbal faculties and my motor centers.
All that I’ve done here is to reinforce and old idea: conceptualization is all about representation and encoding into structural forms. Of course, my favorite is jigsaw puzzle pieces. Of course, the encoding into weight matrices in deep learning networks is a different and highly successful form of encoding.

Later

Bergson, again: “for example a symphony: it was impossible to predict a future symphony as if the composer knew what symphony would be best and wrote it.” But here again, we run into trouble. We could, in principle, follow the composer around, and listen to him hum to himself while shaving. Better yet, listen to the essais at the piano. Listen to earlier works. Stick his head in an MRI machine. You won’t ever get the exact symphony, but it is also not so totally unpredictable. I can, even right now, from this distance, predict that it will have parts written for the string section. We are not dealing with John Cage, here.
So I understand the spirit of the idea, but it conflates “the future is hard to predict” with “the future is impossible to predict”. We’ve gotten technologically quite good at predicting at least simple futures. The symphony remains out of practical reach, but how, again, do we disentangle the creation of the novel and original from the washing machine tumbler of old ideas?
Hmm. This reminds me of an old journal article I found in some random article in some sociology journal. Wandering the reading room in Regenstein Library as a teenager, third floor maybe 4th, near the front of the building, there’s some journal, and some urban anthropologist ponders the law of the conservation of momentum. Somehow, when you throw a baseball, you impart momentum to it with mechanical means (your arm) but your arm is driven by neural impulses coming from your brain, and the elctrons and protons in your nernons have momentum too, and so therefore some thought inside you head must somehow have some microscopic momentum that is somehow amplified into a flying baseball. Which requires conservation of momentum. Totally bonkers reasoning, but not so easy to get out of, because perhaps free will was tied in there.
Of course, we don’t have any problem with describing the conservation of momentum applied to a tennis ball machine. Its purely mechanical, and perhaps there are even wires to some CPU controller. But no one balks at the mechanistic desription of a machine, other than to ask, perhaps, who was the Original Creator of that machine? Yet, when we get to human brains, a mystical confusion emerges.
For Bergson, it is this: “he attempted to find a third way between mechanism and finalism through the notion of an original impulse, the élan vital”. But this is mysticism in its finest. Electrons and protons have electric charge, mass, momentum and elan vital. That’s a non-starter.
Its kind of even worse: we took something straight-forward, like, “how did you know he would throw a baseball at the batter?” and you’d answer: “well he was on the pitching mound, and it was two strikes” and replaced it with something opaque: “how did you know what symphony he would write?” The baseball pitich is similar: the pitcher has the freedom of choice of throwing a slider, a curve ball, a knuckle ball or yet somthing else fashionable and hip and cool with the coaching staff this year. But the pitch was coming, no matter what, unless an asteroid wiped out the baseball stadium.
We seem to have trouble disconnecting creativity and inventiveness from physical action. There is free will, but free will conventionally results in physical actions. Yes, I can ponder my thoughts, but either they are exterminated by a bullet to the head or perhaps brain cancer or old age dementia, or they come to fruition. The pitcher decides what pitch they will throw in about a ten or twenty seconds. Poor decision or wise choice, it blurbles up from the subsconsious, becomes a point of fleeting awareness, in which the conscious mind sorts through the menu of choices and picks one, and then the motor neurons carry out their pre-trained mission, like Seal Team Six raiding the Bin Laden compound or perhaps the Bay of Pigs. The switch has been thrown, action follows.
I said that the pitcher on the mound has a menu of choices in front of him. Of course, he could be completely random: flip some coin, run some psuedo-random number generator, and pitch that pitch. That involves no free will, and becomes purely mechanistic. Or he could read the batters slouch, his head movement, the grip on the bat, the furtive glances or perhaps the sniffle of an oncoming cold and stuffy nose, and think, ah, I have just the pitch to sneak past this guy while he’s not paying attention. Now, the pitcher seems to have some kind of intellectual freedom. But does he actually? Is he actually just nothing more than a thermostat, sensing the temperture, and turning the furnace on or off? (The air-conditioner?) Is the pitcher yet again acting in a purely mechanical fashion: sensing the environment, reading the batter, empoying these sensory inputs to produce a specific mechanical action? Was there ever any free will in the pitcher? Did some random cosmic ray in the pitchers head cause him to throw a slider? I could get a robot to perform these actions. No one debates the free will of robots. Or of thermostats, the ultimate one-bit, on-or-off robot. Control plant. Good regulator theorem.
Baseball pitchers do have style. You’d have to be an afficionado to articulate it, but there is a flair, an elan to pitching. Are we able to confabulate creativity with style? Composers and painters have a style, too, and we are entirely happy to say artists are creative. Hardcore baseball fans will insist that pitchers are creative too, just in ways that you can’t see. But was there actually any creativity? A balky tennis ball machine has style and elan, and you can certainy build a robot to create Jackson Pollock knock-offs, and sell them to midprice comfort hotels. Where, exactly, do we disconnect creativity from mechanical action?
I suppose someone will someday be able to explain this to my satisfaction. Or perhaps Joscha Bach will just call me insane to my face. Of course, that’s only because he was pissed off that he did not have an answer to these confounding questions, and felt it was simpler to call me crazy. For the record: you dear reader, who are not reading this, can indeed conclude that Linas is crazy for writing to a non-existant audience in a scattershot literary diary that will never be published anywhere but github. So in that sense, I am indeed crazy; I should be focusing on my finances, health insurance and repairing the house instead. Of course. But that does not make me any more crazy than the drunk lying on the beach in Florida. Its fun. Its pointless fun. Is it crazy to have pointless fun? No. Unless you’re a psycho who eschews pointless fun. See the problem? Life is crazy. Some say its meaningless, or random, or adopt some nihilistic philosophy. Or worse, they find Jesus. Something totalizing to give life meaning. Me, I’m a washing machine tumbler recyling bad ideas on the edge of my consciious awareness, trying to convert the mythical ghosts into physical form.
The problem with free will is that perhaps its a ghost that cannot be converted into some physical, category-theoretic form. I don’t want to conclude that yet. Free will is so obviously, intuitionistically present, an aspect of human existance, that kind of everyone knows they have it, and it does no good if some 17th century clockmaker insists that Newton’s laws run the universe like clockwork. And no, its not some “elan vitale” possessed by atoms.
Oh, I think I mentioned this earlier in this diary, but there is no need to interpret panpsychism as something that atoms posses. Psyche may require a miniaml arrangement of atoms in order to exist, in the same way as fractured rock has a phase transition from non-percolation to percolation. I presume that intelligence is likewise a phase transition, from non-intelligent to intelligent.
Ah, but here we can go on our reductionist journey again: is slime mold intelligent? Worse: does slime mold have free will? If I run a cycle through the washing machine: does creativity arise via a phase transition as well? Perhaps also free will? Perhaps the reason I am incapable of reducing free will to a simple category theoretic description is because simple mechanical systems do not possess a Cauchy horizon?
Is that the key? Is that the key discovery? Find some arrangement of atoms that has a Cauchy horizon? We know for a fact that Einstein’s equations in five dimensions have a naked singularity. Is that it? Can I code up some differential equations on my computer, modelling general relativity in 5D, and stare at a naked singularity to my hearts content? Is the other side of that Cauchy horizon, is that where free will sits? Is that where the ghosts and myths and reality distortion fields spun by Stve Jobs and Elon Musk, is that where they come from?
How should I understand a Cauchy horizon? It is the place where predictablity breaks down, where I have infinite amplification of infinitessimal perturbations. If I articulated this to Joscha, he would insist that its just a perfect random number generator, and call me crazy for thinking it was anything other than that.
Does neural net dynamics have a mode in which a Cauchy horizon manifests? Is there one, de facto, in the skull of every human, every mamal, every animal, and maybe even in slime mold or amoeba?
Hmm. OK, the grand washing machine tumbler in my mind has certainly come up with something novel. See? So it would appear that keeping a journal of the stream of consciousness of thought can indeed lead to new insights and fresh discoveries. Thank you self-psychotherapist, you’ve been very helpful in this session. It may be a while before I hop on your analyst’s couch again.

2 Dec 2024

I wrote an email today. I think I’m good at writing emails, so I will immortalize it here. It recaps themes that i think I already talked about in this diary.
Hi Paul,
I don’t understand how PDOS solves the issues that you raise, and if I was some minister in the Australian govt, I’d pass this by. So that needs to be articulated.
Let me take 4 steps backwards. Do you know what a phase transition is? Standard school textbooks say "ice/water/steam". There’s another that occurs in fracking. A rock may have a crack in it, but that does not mean petroleum can flow through it. If it has enough cracks in it, then oil will flow. This is called a phase transition: on one side, oil does not flow, and on the other it does. It’s a fairly sharp transition, off/on.
It can be calculated with probability theory renormalization group flow. The value is 0.234... Yes, I actually read through this once. There’s actually an XKCD comic on this, because that calculation works for electrical grid netsplits, tcp/ip netsplits, crack propagation in steel boat hulls, falling and breaking coffee cups and tears in paper. Remarkably broad theory.
In the broadcast-TV days, the network consisted of a few central broadcasters (Walter Cronkite on the evening news) and millions of viewers. This is the hub-n-spoke network topology. Walter is the hub at the center, and if he’s out sick, then millions of viewers do not get his thoughts.
With youtube, facebook, twitter, we have gone through a phase transition in the network topology. Rather than funneling all information to go through pipes that connect to Walter Cronkite, we now have many to many.
A vast amount of youtube content is educational. I’ve watched youtube content on how to put on 18th century women’s clothing, how to dance like a Goth, how copper ore was mined in Appalachia in the 1930’s, how to clean carpets using modern tools, and how Ford’s River Rouge auto assembly plant worked in the 1950’s. And how to dance the latest dance craze. And real-estate woes in NYC. Yes, its strongly addictive.
Is this good? Is this bad? I have two sons addicted to computer games, and it is quite hard to see anything good coming from that. It does not seem to be teaching them anything useful. There does not seem to be anything of productive value coming out of this. So, its kind of like drug addiction, alcohol abuse. Mostly just debilitating. A net drain on society. A net drain on me, personally: they generate no income.
Continuing with the pipes analogy, there are now direct connections from my son’s brains, into computer games, but those computer games are dead-ends. They’re a dead end, they go nowhere. There’s no there, there. It’s like society lives in a pot, and we talk to each other, to our neighbors, but this pot has a million holes in the bottom, where the water drains out, and goes nowhere. Review the wikipedia article on the "Global Brain", if you haven’t already. Computer games are a short-circuit, a sink, a leakage, that takes dopamine and endorphins and dumps it to nowhere at all, into the great void.
Where’s the line between good and bad? The first computer game ever was Adventure, written by an English prof for his 12-year-old daughter. It was an "interactive text" (we’d call it "hypertext"). "You are standing in front of a log cabin, with a forest on three sides. There is a creek bed off to the side. Where do you want to go? North, East, West, South?" This is a *literary work*. You explore it one paragraph at a time. There are elves and dwarves and caves and magic beans and a bird and a rusty want and a key. XYZZY.
Is this at the level of George Elliot’s Middlemarch? No. I recently read the first chapter of Middlemarch, its a stunning masterwork of genius. I was floored. Did I read the rest of the book? Uhh. erm. ahh, not yet.
Is reading books bad for society, or good? What about poetry? How about music? The Arctic Monkeys are masterwork poets, writing biting social commentary, and they are top-of-the-pop-charts. Check out the lyrics to "Teddy Picker". In the 1950’s, conservatives were scared of Elvis Preseley’s gyrating hips. The sexual revolution of the 1970’s has culminated in wokeness and LGBT transgender debates. Is it good? Is it bad? Is it economically productive? Every second we waste arguing about transgender bathrooms is a second we waste not fixing the hole in the roof, where the rain comes in. It gets my mind to wondering, where it will go, oh, oh, oh, where it will go, oh oh oh...
Somehow, lying in a heap on the sidewalk, high on heroin or pcp or meth or whatever, this is obviously bad. How about lying in a heap on the sidewalk, arguing about Ancient Greek poetry? Is that acceptable?
What’s good? What’s bad? I don’t know. I don’t know how to draw the line. Where does PDOS fit into the above? I have no clue.
--- There is one issue where there really are tremendous dangers to society: propaganda, disinformation. The russians are absolute masters of this game, they’ve been honing it since the 1960’s, and arguably even much earlier.
Someone once gave me a short tract on propaganda. Well-written, clear, insightful, thought-provoking. The occasional oddball sentence and example makes it clear this was written in the 1930’s, and the examples are drawn from WWI. By the end, you get to "wait, what??" and the mystery reveal is that this was written by Adolph Hitler. So, the mechanics of propaganda were understood in WWI, and Adolph’s complaint was that they were ham-fisted in their efforts.
We have come a long way, baby, and the russians present a clear and present danger. If you think social media is bad, just wait till you see social media Hi Paul,
I don’t understand how PDOS solves the issues that you raise, and if I was some minister in the Australian govt, I’d pass this by. So that needs to be articulated.
Let me take 4 steps backwards. Do you know what a phase transition is? Standard school textbooks say "ice/water/steam". There’s another that occurs in fracking. A rock may have a crack in it, but that does not mean petroleum can flow through it. If it has enough cracks in it, then oil will flow. This is called a phase transition: on one side, oil does not flow, and on the other it does. It’s a fairly sharp transition, off/on.
It can be calculated with probability theory renormalization group flow. The value is 0.234... Yes, I actually read through this once. There’s actually an XKCD comic on this, because that calculation works for electrical grid netsplits, tcp/ip netsplits, crack propagation in steel boat hulls, falling and breaking coffee cups and tears in paper. Remarkably broad theory.
In the broadcast-TV days, the network consisted of a few central broadcasters (Walter Cronkite on the evening news) and millions of viewers. This is the hub-n-spoke network topology. Walter is the hub at the center, and if he’s out sick, then millions of viewers do not get his thoughts.
With youtube, facebook, twitter, we have gone through a phase transition in the network topology. Rather than funneling all information to go through pipes that connect to Walter Cronkite, we now have many to many.
A vast amount of youtube content is educational. I’ve watched youtube content on how to put on 18th century women’s clothing, how to dance like a Goth, how copper ore was mined in Appalachia in the 1930’s, how to clean carpets using modern tools, and how Ford’s River Rouge auto assembly plant worked in the 1950’s. And how to dance the latest dance craze. And real-estate woes in NYC. Yes, its strongly addictive.
Is this good? Is this bad? I have two sons addicted to computer games, and it is quite hard to see anything good coming from that. It does not seem to be teaching them anything useful. There does not seem to be anything of productive value coming out of this. So, its kind of like drug addiction, alcohol abuse. Mostly just debilitating. A net drain on society. A net drain on me, personally: they generate no income.
Continuing with the pipes analogy, there are now direct connections from my son’s brains, into computer games, but those computer games are dead-ends. They’re a dead end, they go nowhere. There’s no there, there. It’s like society lives in a pot, and we talk to each other, to our neighbors, but this pot has a million holes in the bottom, where the water drains out, and goes nowhere. Review the wikipedia article on the "Global Brain", if you haven’t already. Computer games are a short-circuit, a sink, a leakage, that takes dopamine and endorphins and dumps it to nowhere at all, into the great void.
Where’s the line between good and bad? The first computer game ever was Adventure, written by an English prof for his 12-year-old daughter. It was an "interactive text" (we’d call it "hypertext"). "You are standing in front of a log cabin, with a forest on three sides. There is a creek bed off to the side. Where do you want to go? North, East, West, South?" This is a *literary work*. You explore it one paragraph at a time. There are elves and dwarves and caves and magic beans and a bird and a rusty want and a key. XYZZY.
Is this at the level of George Elliot’s Middlemarch? No. I recently read the first chapter of Middlemarch, its a stunning masterwork of genius. I was floored. Did I read the rest of the book? Uhh. erm. ahh, not yet.
Is reading books bad for society, or good? What about poetry? How about music? The Arctic Monkeys are masterwork poets, writing biting social commentary, and they are top-of-the-pop-charts. Check out the lyrics to "Teddy Picker". In the 1950’s, conservatives were scared of Elvis Preseley’s gyrating hips. The sexual revolution of the 1970’s has culminated in wokeness and LGBT transgender debates. Is it good? Is it bad? Is it economically productive? Every second we waste arguing about transgender bathrooms is a second we waste not fixing the hole in the roof, where the rain comes in. It gets my mind to wondering, where it will go, oh, oh, oh, where it will go, oh oh oh...
Somehow, lying in a heap on the sidewalk, high on heroin or pcp or meth or whatever, this is obviously bad. How about lying in a heap on the sidewalk, arguing about Ancient Greek poetry? Is that acceptable?
What’s good? What’s bad? I don’t know. I don’t know how to draw the line. Where does PDOS fit into the above? I have no clue.
--- There is one issue where there really are tremendous dangers to society: propaganda, disinformation. The russians are absolute masters of this game, they’ve been honing it since the 1960’s, and arguably even much earlier.
Someone once gave me a short tract on propaganda. Well-written, clear, insightful, thought-provoking. The occasional oddball sentence and example makes it clear this was written in the 1930’s, and the examples are drawn from WWI. By the end, you get to "wait, what??" and the mystery reveal is that this was written by Adolph Hitler. So, the mechanics of propaganda were understood in WWI, and Adolph’s complaint was that they were ham-fisted in their efforts.
We have come a long way, baby, and the russians present a clear and present danger. If you think social media is bad, just wait till you see social media distorted by russian bots. Its seriously fucked up out there. They are waging war against the Western World, against Western values, and most people are lah-de-dah, don’t notice, don’t care. Yeah, the frog getting boiled in the pot doesn’t notice and doesn’t care, either. We’re at war, and like, no one notices.
The good news for us is that the Chinese are totally incompetent at propaganda. The bad news for the Chinese population is that they know how to operate 15-minute cities.
How does PDOS fit into any of this? Could we use it to deflect the psychological damage generated by the russian bot farms? How would that work?
--linas
distorted by russian bots. Its seriously fucked up out there. They are waging war against the Western World, against Western values, and most people are lah-de-dah, don’t notice, don’t care. Yeah, the frog getting boiled in the pot doesn’t notice and doesn’t care, either. We’re at war, and like, no one notices.
The good news for us is that the Chinese are totally incompetent at propaganda. The bad news for the Chinese population is that they know how to operate 15-minute cities.
How does PDOS fit into any of this? Could we use it to deflect the psychological damage generated by the russian bot farms? How would that work?
--linas

Earlier

I wrote an email a few days ago, the above was part of that thread. Here’s some of the earlier stuff.
Linas Vepstas <linasvepstas@gmail.com>
Nov 30, 2024, 9:13 AM (2 days ago)
to Paul Not saying that you personally have to be the one walking door-to-door talking to Philippinos about this, but someone does. You have a job as a community organizer where the community is me, and the other people you talk to and sometimes help you. This community "works" because we all agree, more or less, that hacking on this or that is cool, fun, worthwhile.
All communities work this way: there’s something cool, fun, worthwhile to do, to collaborate on. But what is "that"? It’s *always* some vague, mythical, aspirational dream. Everyone in the community knows what it is, and it unites them in purpose. You’ve sort of articulated what it is for you: Sneakernet communications, cheap, almost-free hardware, large distances, and maybe some angle with involving poverty or govt/gang suppression or maybe some post-nuke-survival angle. This is the myth, the vague aspirational dream you are working towards, and the community you have to build are others who buy into this vision of the future.
Part of this vision is reaching out to others. For that, maybe you want to revamp your website. You might want to have the website say something like this last email. "Hey people of Bhutan, check it out, you might like this thing." I’d look to South America: Venezuela is in free fall, going from really bad to even worse. Some of the other countries are also mis-managed and in chaos. Yes, Bhutan. Maybe Outer Mongolia. There are Eskimo tribes in northern Russia. I call them "Eskimo", this is usually reserved for Alaskan Indians, but there are linguistic and cultural relatives out on the tundra all along the arctic. Is this part of your vision? I dunno. Maybe. It could be, maybe not.
Write your website so it appeals to someone from those areas. They might not realize or understand that they can take PDOS and put together a fidonet community. You have to explicitly explain this. Use words where the wise man, the shaman from Bhutan stumbles on your website, and reads it, and goes, "yeah I need to bring this home with me, for my people."
This is your "community building". You’re not knocking door-to-door in the Philippines, you’re, uhh, virtual-door-to-door in emails and websites. There’s a message. Sculpt that message so that it can be heard, so it propagates. If people aren’t hearing it, maybe the message is not stated clearly, or maybe you’re reaching the wrong people.
Re Australians and BBS’es and being outdoors: this is an example of the right message to the wrong audience. So the guy who invited me to Ethiopia, he runs multiple businesses there. One of them is a tech school for high-school kids. Rich parents really really want to give a good education to their kids, and pay him good money for that. For every rich kid, he gets 3 other poor kids from rural communities who have just the right combo of autism, interest in science, and the desire to be inside hunched over a keyboard, instead of outside dancing along a roadside. (Yes, they love to dance...) The target audience is not "everybody", but those who already are going in that direction, and just need encouragement.
So that was your problem with the Australians. You should have been going to libraries, where someone is hunched over a book, and *doesn’t want to go outside*, and say "hey I’ve got something like a book but it’s better, it’s called a BBS and it works like this". That would have been your target audience. The guy who wants to hike the Outback, maybe not so much.
Sculpt the message. Put it so that your target audience will hear it, see it, understand it, and then join it and spread it.
-- Linas Paul Edwards
1:32 AM (13 hours ago)
to me Hi Linas.
That’s a really great message explaining some of the fundamentals that I had no clue about.
But I have some sort of parallel question.
So (us) engineers have managed to make smartphones and computers go way beyond what was possible on Fidonet, to the point where ordinary people, instead of being largely uninterested, are now addicted (it seems to be literally worse than actual drugs - consuming people’s lives).
I would say it is more an extension of TV than Fidonet. TV was already very popular back in the Fidonet days (and before), but now we’ve basically created millions/billions of TV channels, so there are multiple TV channels that are extremely interesting to almost everyone, and almost no reason to put down your phone.
Australia has recognized a problem and put in a social media ban for anyone under 16. I don’t think that’s going to be sufficient - as even if it can be enforced, I think the problem is more youtube. I don’t think I actually know any Australian children. I have more insight into Filipino children, and this is my understanding of what is happening. There are computer games too.
But at the same time I’m a big promoter of computers and I encourage children to get access to computers. My daughter has a tablet. She is 3 years old and has recently started using the calculator (just for 2+3=5 and 2+4=6). You can see video of her using PDOS here:
https://www.pdos.org/vlog/vlog.htm
(there’s just that one video on the top - quite large - 500 MB I think)
So I was thinking of how to reconcile that.
As you noted - get the guy from the library onto a computer to be more efficient at what he’s doing - but no-one else.
So if I was advising the Australian government, it would be what? Instead of nuclear war or Bhutan providing the opportunity for PDOS - it’s actually to save Australian youth.
The bookworms can be taken out of the library and put in front of a text mode and delayed communication method like Fidonet. If the age restrictions are truly effective (something like cigarettes), they will need to go to an adult to get a Fidonet packet emailed to some other Fidonet node (or exchange a USB stick at school).
So PDOS is not so much competing against Windows/Linux, as competing against the complete destruction of the world.
Perhaps the alternative to PDOS is something like this:
https://www.sciencehistory.org/stories/magazine/mouse-heaven-or-mouse-hell/
Perhaps it is bandwidth that needs to be restricted. Although that alone would probably give you the Cuban sneakernet where movies etc are transferred each week via USB stick.
Possibly the restriction needs to governed by blind people. ie you can watch movies if you want - but audio only. That would likely be sufficient incentive where playing soccer is more entertaining. Except for the genuine bookworms who are just using a different medium to get the same information a blind person gets.
Any thoughts/direction?
Maybe I should be structuring my website for the Australian government rather than Bhutanese shamans.
BFN. Paul.

3 Dec 2024

Email. Paul Edwards
Dec 2, 2024, 8:41 PM (16 hours ago)
to me Hi Linas.
Great to see your questions. And I have read the references you pointed to.
Let’s start with this.
> How about lying in a heap on the sidewalk, arguing about > Ancient Greek poetry? Is that acceptable?
No - it’s not acceptable. If you wish to argue about Greek poetry - that’s great - as a pastime. You can hang glide or abseil or play computer games if you prefer - up to you.
But this is all unproductive work. You do need to have the base productivity.
You mentioned your sons not generating any income. What are their ages?
Replace their existing computers with a system running PDOS, and bam, that text game "Adventure" is all they can play as a pastime. But just in case they get addicted to text games - like Greek poetry - that’s still unacceptable. But taking away the graphics will solve most of the problem for most people I think. I doubt that a large percentage of the population would rather play "Adventure" than go hiking. Although even hiking is still recreation - we don’t necessarily need people to do hiking instead of graphical computer games. But hiking is unlikely to be addictive. Or if it is, it isn’t to the same extent that computer games are. Growing up - have you ever heard someone say that their 2 sons spend all their time hiking and are addicted and producing no income?
So that’s what I envisage for Australian etc children, and your sons. Yes, you can have a mobile phone so that you can call for help or whatever. And you can have one application on it - PdAndro (PDOS, basically). And good luck getting addicted to that. Although in fairness, even before the internet, some people appeared to be addicted to phones (even just voice calls).
Does that not make sense? PDOS takes away the addictive graphical computer games. You only have text based applications left, and are in a similar boat to a blind person.
It doesn’t need to be PDOS specifically. You can simply ban smartphones and only allow dumb phones. Or strip down Linux to remove graphics. It’s not that PDOS has some magic ability.
Note that my 3 year old daughter was recently introduced to computer games on a tablet. I’m leery about that. I’m confident with her using PDOS, and the calculator. And the video (Peppa) to some extent. But I’m concerned that she will become addicted to the computer games, and voiced my concern to my wife. It still hasn’t been resolved, as I don’t actually know what will happen.
Maybe you do. In hindsight, is there anything you could have done to prevent your sons from becoming addicted to computer games? What happens if you ask them directly if you did something wrong, to produce this result? Do they agree it is a problem?
We should consider making computer graphics a class A narcotic or whatever the terminology is.
BFN. Paul.
Linas Vepstas 12:41 PM (6 minutes ago) to Paul
There is a problem. There’s no solution in sight.
* In the US, there was an attempt to limit access to computer games in the late 1980’s, early 1990’s, Look up Tipper Gore.
* There were court cases. In one case, the game involved the Queen of England, which allowed the court to explain that political cartoons depicted the Queen in the 19th and 18th century (Punch) and games are protected speech.
In short, you’re hard up against a free speech issue.
Until people start dying of gaming addiction, nothing will happen.
* People died from COVID, and a large fraction of the population thought covid was a hoax, or otherwise expressed their utter stupidity.
* People die from cigarette smoking. After 50 years in the US, smoking is down. In the EU, its as popular as ever.
* People die from gun crimes. Gun regulation, here in the US, is very very far from happening. We had more gun regulation in the 19th century, the days of cowboys and indians and western fighter street duels.
* People die daily, primarily in poor neighborhoods, from gun duels, drug addiction. General political attitude is to blame the victim.
* Global warming. Obvious looming disaster, tremendous resistance and denial.
* Helmet laws, seat-belt laws, the "nanny state" debate.
There’s stuff out there that people do that is deadly to themselves, to each other, to me (it’s not safe for me to walk in downtown Austin, "6th street". Infamous party district. Couple of months ago, someone on the roof of a bar threw an empty beer barrel at me. They missed. Four policemen on every corner, and they still can’t stop mugging-by-beer-barrels.)
There’s a huge cultural and corporate pushback halting regulation. (Bars on 6th street, in this example, and the "freedom to party like its 1999") Corporations make money from toxic deleterious stuff (tobacco, DDT, hexavalent chromium, polychlorinated biphenols, whatever), and private individuals, usually stupid but not always, want to do shit that literally results in their death (smoke tobacco while spraying DDT in their tomato patch.)
Computer games aren’t actually killing people. There won’t be legislation.
This is why I talked about percolation and phase transitions. There might be systemic solutions. For example, smoking is down in the US because it’s "not cool". "Coolness" is a social & cultural factor. Max Weber broke new ground when he wrote about the Protestant Work Ethic: The 18th century belief that you go to Heaven if you are a good person, and the way to demonstrate you are a good person is to do good deeds in the real world, and the proof of good deeds is done by bettering the lives of everyone all around, and the best way to make life better for everyone is to build a factory that produces those things that people need. In short, work hard, get rich, go to heaven.
Weber’s twist was to say that this explains *why* the Industrial Revolution happened in New England, and not in Muslim countries, Buddhist countries. Buddhism is about renouncement. It’s very much against the accumulation of material wealth.
Is belief in Christ and the accumulation of wealth the answer to our problems? No. I don’t know if you go to church or not. I try it every few years, I always emerge with revulsion: these are sick people offering bad advice, lost and hopeless, wayward, dazed, confused, deluded. I’m talking about the priest, pastor, bishop, whoever is giving the sermon. Went to Christmas Mass in Notre Dame in Paris. Whoever gave the sermon there, some bishop, was a world class fucktard. ("If Mary had an abortion, Jesus would not have been born." Christmas Mass? Seriously? WTF.)
Accumulation of wealth is not all it’s cracked up to be. I see how being an 18th century industrialist could be fun: build a large hall, dam the local creek for a mill-pond. I could do that for 16 hours a day, seven days a week, much like I can do software on this schedule. I can understand how it’s fun.
But the 9-to-5 lifestyle? Frankly it sucks. It’s demeaning, soul-sucking, an empty and meaningless lifestyle. You get some money, but so fucking what? What are you going to do with that money, if you’re unhappy? This is why people turn to Buddhism: to "find themselves". Why they go hiking mountains: to "find themselves". Explore drugs, sex, rock-n-roll, parties. Anything is better than the emptiness inside. Something to ease the pain.
Some fraction of youth look at their parents, and say "I don’t want to live like that" and computer games offer an escape, a salve.
Want to cure gaming addiction? Find the systemic belief system that promotes a walk through adult developmental stages. The ones that run between ago 25 and age 45, where people get lost and do fucked up things (get addicted to games, among other things). Make this progression a core element of the cultural, social fabric. Make gaming uncool, like cigarette smoking. Make "finding yourself" cool enough so that people can, euhhh, err, return to society as validated, healthy, strong individuals engaged in productive and uplifting activities.
--linas--linas

4 Dec 2024

What I will write below will lead you to think I’ve spread myself too thin. But I’m bored. Boredom makes people do weird things. I’m bored because I just spent all day shopping for replacement parts for all the failing computer shit I have. Agonizing over $400 Whoop de do. If I were rich, I wouldn’t blink at this sum. But instead I agonized. Oh well. Three hare drives and two power supplies. Got a good deal on the disks. Bought upscale power supplies because the cheap ones fail ever few years. Also a broken monitor. Also a failing motherboard, will do that tomorrow. Maybe this will allow me to run a semi-stable Ceph cluster, maybe. And maybe minimize data loss. Maybe.
Anyway, I came here not to write about that. So to relax after a strenous day of shopping, I’m reading Tong, General relativity. It’s basic stuff, been there done that, but its modern, so I though I’d glean something new. Well, not really, but the refresher is OK. Read chapter six. Going back to read chapter five, which says something remarkable. Remarkably stupid. Well, remarkably conventional, but the conventional ordinariness of it has me gob-smacked. So I have to bitch about it here.
So at the start of chapter five, its says the metric tensor can be thought of as a field, and here’s the Hilbert-Einstein action, and its second order just like other conventional field actions. And this is true, in the conventional sense. But I’m like ... wait, what? Surely you must be joking? And I realize that in fact maybe most physicists think of the metric tensor this way. And that just ... fugnuts insane.
Here’s the problem. So, if you study Riemannian geometry, not pseudo-Riemannian, its kind of normal (for me, at least) to think of Riemannian manifolds as being static or fixed. They’re some shape, and the metric tensor is just used to describe that shape. These manifolds are static, they don’t move, they have no dynamics. The metric tensor is well defined under coordinate transformations, because points on the manifold exist in and of themselves, and are in one-to-one correspondence with the metric-tensor-in-the-abstract, at that point. Redoing the local coordinate charts changes nothing. Well, nothing if you work coordinate-free. Of course, co- and contra-variant things are co and contra- but there’s no magic or weirdness there. The shape of the manifold itself does not change under coordinate transforms. So you get this very static mindset.
And if you take that to GR, whoa lordy-mama. So, with that mindset, you kind of start thinking: all past and all future of some spacetime is FIXED? Seriously? Is this why a lot of physicists blabber about the future being written? Because if spacetime is like a Riemannian manifold, then yes, it would be fixed forever and ever, and of course, you have no free will, you are condemned to be just some splat on some 4-manifold. And that’s fucked up, because somehow obviously the future is not fixed, in basic human common sense terms.
I mean, I can see the appeal of that. The Schwarzschild spacetime is stationary and static. Nothing happens. The Kerr spacetime is also static in its own certain way, at least up to the Cauchy horizon, which we expect to be unstable, mumble, mumble. So here we have the first inkling of a problem: How can a manifold be “unstable”? Conventional (Riemannian) manifolds are “stable”, in that they don’t “move”. Perhaps you have trouble writing some convergent analytic expression in some domain, or perhaps you have trouble integrating some differential eqn for some reason, but the conventional mental model is not one where this results in the manifold moving in some “unstable” way. So WTF? Just because we took one of the manifold coordinates and flipped the sign on it. So here we see not just a flicker, but perhaps a stark-raving-mad collision between the idea of some manifold that has some fixed shape, no matter how your atlas of maps places coordinates on it, and “instability”.
I mean, for ordinary weak-field gravitation, sure, it makes perfect sense to think of space-time as some static manifold rules with grid-lines from your coord system. That’s how we calculate planetary orbits and place satellites and spacecraft where we want them. Not a problem. No problem imagining spacetime to be a conventional manifold meeting all the conventional definitions/axioms of a manifold. Ditto in the pure Newtonian limit. Also ditto for flat Minkowski space. Accelerator physics works exactly how its expected to, even at 99.999999% the speed of light. And, on the null cone, optics works just as we expect it: both on-shell quantum optics, and off-shell perturbative theory in the eletroweak quantum field theory. No problems there, the concept of a manifold makes perfect sense.
But inside the black hole, we have problems. The “instability” of the Cauchy horizon. WTF does that even mean? Yes, I know what it means. Minor perturbations in the external universe pile up and becoe infinite, with infinite flux densities at the Cauchy horizon. Easy search engine searches instantly reveal half-a-dozen papers getting into this, and a masters thesis. So right there, we have a problem. If spacetime is “just a manifold”, then how am I supposed to think of this instability? Is it supposed to be something that gets infinitely wrinkled as it approaches that limit? Perhaps recursively fractal? Perhaps dynamical chaos? Low-dimensional chaos, or high-dimensional chaos? Am I supposed to think of the manifold, as it approaches the Cauchy horizon, does it become turbulent? Like burnt jet-fuel coming out of a jet engine, just high-dimensional turbulence? Then, by conventional statistical mechanics, doesn’t that mean that this turbulence then converts into a perfectly uniform micro-canonical ensemble of a perfect mixture, exactly at that Cauchy horizon? A mixture of what? Infinitely-turbulent spacetime? Whoo hoo. How come no one talks about this? Or do they talk, and I haven’t been listening?
So that’s one. Another is mass inflation. I read not one, not two, but I think three papers on mass inflation. Only one of them was any good. One was a simplified recap that was boring, and another was, I dunno, wandered off into something boring. Anyway, no point in recapping here, other than to note that mass inflation again suggests that thinking of “static manifolds” is just wrong. Now, of course, mass inflation requires a small influx of infalling particles, so that right there means we are no longer doing pure(pseudo)Riemannian geometry. We are doing “real physics”, where the infalling particles are not just following geodesics on a static, fixed manifold, but they have a back-reaction on it. They actively change the shape of the manifold as they interact with it. And BTW, mass inflation does not require quantum-anything, as I recall. It just needs fields, but these fields are effectively classical. So, here, perhaps the mental model could be like a Raleigh instability: just like wind blowing over the surface of a lake to create water waves, so here, a “wind” of some classical scalar field “blowing” over the spacetime manifold, generating an instability. The conceptual problem here is of course that the manifold is now dynamical, and this dynamics is at some meta-level, and not in the time coordinate. That is, in this picture, space-time is still a 4-manifold, a 3+1 manifold, but it’s interaction with the scalar field means it twists up (well, inflates) in, uh, certain locations. How should I think of that? Should I think of spacetime as a “static manifold”, just one with a giant aneurysm, ballooned out in some set of locations? Or should I think of it as being “dynamical”, but in some other dimension that isn’t conventional time, but some other dimension of time? Like some “fifth” dimension, but clearly it cannot be “fifth”, that doesn’t make sense. So its dynamical how? How can it be “dynamical” when we don’t have a time dimension to work with? See, its all fucked up. But I guess we can think of mass inflation as a kind-of-like static manifold with a ballooned-up portion. This seems not entirely a happy way of thinking about it, but it will have to do for now. Fine. Whatever.
And then, to add insult to injury, some physicists want you to think of the metric tensor as a dynamical object? Cause that’s what Tong says in the first or second paragraph: think of the metric tensor as a dynamical field, with the dynamics given by this cute little action of the scalar curvature known as the Hilbert-Einstein action. Sure dude. Have your cake, and eat it, too. Its static. Its dynamic. It moves. It does not move. Make up your mind.
The criminality of making the metric into a dynamical field is that I don’t know how to do that. I know how to write down a static, non-moving manifold, and how to paint it with a metric. Apply that metric, like a paint job on a car, using maps and atlases and transition functions obeying axioms when two and three maps intersect. I can do that. But when I do that, the metric is pinned down, dead and inanimate, on the manifold. Its not “dynamical”. Its not “moving”, its not “evolving”, or “changing”. It just “is”. So where the fuck do you get off saying the metric is a dynamical field? That’s fugnuts nonsense.
Its worse than nonsense, its incoherent. I mean, if the manifold is fixed, I can’t just change the metric, without also changing the shape of the manifold itself. But if I change the shape of the manifold (and I can do that!), then which points map to which other points, in the now-changed manifold? This is called a “homotopy”. I can do a smooth, differentiable homotopic deformation of the manifold, and then I can know how some points map to other points, and how the corresponding metric changes as well. But homotopies are characterized by a single real number, usually, on the unit interval [0,1]. So this homotopic parameter can now become this “fifth-dimensional time coordinate”, and that allows me to reintroduce a concept of dynamics into what is otherwise a non-dynamical system.
And I think this works. I think this can be made to work. There are only a few issues (but I like them!) One is that there is an infinite number of different kinds of homotopies and deformations that one can have.The other is that some of these homotopies seem to have large “bends” in them. But that’s OK. I like this because the “infinite number of homotopies” is a kind of “infinite number of states” in a quantum system. And I like the idea that “some bend more than others”, because this now gives you an action. But that action cannot be over the 4D space (the 3+1 D space), but over the space over all homotopic deformations of a 4-manifold. In other words, a functional integral.
Huh. Surely I am not the first to think of this, and perhaps there are hundreds of journal articles on this, and all of the great physics luminaries of my time already know this. But, well, hey, its news to me. But then, it can’t be “well known”, because otherwise a person like Tong, who seems o be a tenure prof and a name-brand school wouldn’t be writing idiocies like “think of the metric as a dynamical field”. I mean, perhaps he thinks he’s telling a little whit lie, a fib, for the benefit of his second-year students? But there’s no asterisk or footnote that say, “well, gee, really, the metric can’t possibly be dynamical”. And I read and/or skimmed pretty much all of Misner, Thorne, Wheeler, and I don’t recall seeing that kind of asterisk or footnote. There is a rather interesting but hairy and technically complex passage on doing direct numerical integration of the Einstein field eqns, and the kinds of issues that arise, and how to get around them... when doing numerical integration. I don’t recall seeing the word “homotopic deformation”, and sure, one is minimizing some action, but the minimization is not over the space of homotopic equivalences. So WTF. Fore sure for sure, Weinberg never mentions this, and I read his book cover to cover. Excellent book. Well, excellent, if you already know GR. Kind of hard to learn GR from it for the first time. But whatever.
I need to think about and read about homotopic deformations of spacetime. I also need to re-read all those places where I’ve read the word “deformation”, and maybe layer a new layer on that. Certainly, deformations are rampant in affine Lie algebras and Virasoro algebras and all that, which is why maybe everything I wrote above is “obvious” to the practitioners versed in the state of the art. But its new to me. Hmm.
And then the other problem, of trying to visualize the Cauchy horizon as a limit, a transition to infinitely turbulent spacetime, perhaps like an infinite filigree of Alexander horns or something. Unexplored territory. FWIW, I saw an excellent lecture YouTube on exotic 4-spheres, from this girl with an awesome muscular back, at Harvard, and for the second time ever, I flt like I really finally got a pretty good grasp on exotic 4-spheres. Which I have now totally forgotten. Alas. But the point there is that these arise as knots, knots that are sometimes pretty big and complicated, but finite, maybe having hundreds of crossings, maybe more, maybe less, Betti numbers of maybe 10 or 20 or maybe 6 or 4 or less or something, I don’t recall the lectures. And other invariants that are all less than a hundred, in the dozens, I can count with my fingers range. But high-dimensional turbulence, I can’t count on my fingers. Now, I could argue that a *single* set of (infinitely-recursive) Alexanders horns are given by a “single” invariant, whose value is “one”, because there’s only one set. But infiinite-dimensional turbulence in a four-manifold seems to require an, err, uhh, knot with an infinite number of crossings, or something, embedded in that three-space that is cut out from the “surface” of that 4-sphere. Except for GR, we need 3+1, and not 4, but whatever.
I don’t (yet) know how to convert any of this into algebraic equations. And I do admit, every time I sit down and try to write this out, I get nowhere fast. But this is now the general meta-level. To recap:

5 Dec 2024

Apropos the above: I am confused about the concept of energy & action. So maybe if I write about it here, I will learn something. Action is measured in units of hbar, and so is angular momentum. Action has dimensions M T . The energy of a simple harmonic oscillator is ℏ ω ( n + 1 / 2 ) and in units c = 1 gives energy has dimension M .
Classical momentum is p = m v so units of M L / T = M if we measure in T = L i.e. c = 1 units. So momentum is measured in units of M . Of course: E 2 = p 2 + m 2 .
Classical angular momentum is a weight on the end of a string being spun around at some given speed, so, momentum times moment-arm, or mass times velocity times length, units of M L or equivalently of ℏ .
The action appears in two different settings. One is in spin, and Jost shows how to set up a Clifford algebra and spinors for Riemannian manifolds of arbitrary dimension. The other place the action appears is in Hamilton’s eqns (I almost wrote Euler-Lagrange eqns) for obtaining a geodesic on a manifold. In four dimensions, these all end up with the same unit hbar. I don’t know how it works out in other dimensions?
I’m also thinking that the simple harmonic oscillator is a “degenerate” case of spin. So if we have something rotating in 2D, then we can decompose it into two 1D motions via Euler’s eqn e i θ = cos θ + i sin θ . But the other way to think of this is to say, no we don’t decompose, with take superpositions, and a superposition of a pair of spinning motions can be used to create a 1D motion (that is the simple harmonic oscillator), and that is why hbar appears in both contexts.
Of course everything I wrote above is infinitely muddled. But I’m confused about something and I don’t even know what I’m confused about. Of course I know about representation theory and Casimir invariants and charges and generators and tangent spaces. All these are interconnected and interrelated in well-worn relationships familiar to all grad students. That’s not the problem.
The problem is that hbar is used in these two distinct contexts: in variational principles, where we minimize the action, and in spin manifolds, where we can build up spin structures, which are manifold-like things having a spin connection. Why is hbar in both contexts? Is spin “more fundamental”, in that any variational principle must first pass through a decomposition into a spin structure, products of which can then be re-assembled to create a tangent space (of vectors, so in representation theory where a product of fundamental representations gives an adjoint representation) and so the tangent space is identified with the adjoint rep. but the fundamental rep was “more fundamental”. By convention, variational principles are always applied on the tangent space, because that is where Hamilton’s eqns do their thing. But, if we wanted to be “more fundamental”, we should do this on the spin rep, not on the tangent space. So rather than talking about symplectic manifolds and symplectic structures, the “more fundamental” view is to work with spin structures. Thus, if we want to obtain a geodesic, or some equations of motion, we should not work directly with the principle of least action applied as a variational principle on a homotopic deformation of some curve or surface, and instead work with the spin structure on that curve or surface, and obtain the variational principle as the product of spin reps, in the same way we obtain the adjoint rep as the product of spin reps.
But how? I’m well-aware that the above sounds more-or-less incoherent and meaningless, and is almost empty of content, if I cannot articulate it with actual formulas. So, sure, you can dismiss it as irritating time-wasting mumbo-jumbo. But I’m confused. I think something like the above should be possible to perform. It can be performed without violating the no-hair theorem, because spin is not “hair”; spin is already inbuilt in the Kerr solution. We also don’t violate the (I forget what its called) theorem, because all the other fields will still appear as fiber products, rather than as Cartesian products. So we can fiber in arbitrary other fields/structures, and entangle them with spin, without wrecking the fundamental spin structure that is “underneath” the variational principles.
See what the problem is, now? So we have this issue of the “mystery” of EPR entanglement, but this is an entanglement of spin structures, and does not seem to require any other additional fields (electrons, photons, etc.) even though from the engineering side, we have to have actual physical electrons/photons to demonstrate entanglement. But mathematically, it seems like spin is enough. I almost wrote “massless spin”, but this is one of the futzy corners that also helps cloud the issue. Two issues. First, massless vs massive means in the light cone vs on the null surfaces. Second is using Weyl spinors to construct Majorana vs Dirac. If you dig into the details (BTW, I wrote most of the current Wikipedia article on Majorana, and Weyl spinors, and gamma matrices, and Weyl algebras, etc. so I am aware of their contents cause I wrote the fucking articles.) and so here’s the issue that stopped me from finishing those articles: when you start constructing the Weyl spinor products, to get either Majorana or Dirac, you trip over the vector vs. axial vector products. And that’s fine, but we know the axial vector symmetry is quantum-broken; this is the anomaly.So, I can do solitons, for example, as harmonic maps on Riemannian manifolds (again, Jost has an excellent chapter on harmonic maps) but the axial product is broken by quantization.
But if you’ve been paying attention to what I wrote above, quantization is rooted in the variational principle. That is, we take the action S and say that the classical eqns of motion are δ S = 0 the extremum of the action. If we think of this extremum as a single point in an infinite dimensional space, then the variation δ is an (infinitesimal) radial movement away from that single point. So, for example, in computing a geodesic, we consider a small “bump” or deformation of the geodesic in some direction δ and note how that increases the total action. There are infinitely many different kinds of “bumps” that we can make, so this is an infinite-dimensional space. Each bump is (by definition) smooth (differentiable C ∞ ) but perhaps we need only differentiability to second order. So above I talked about “homotopic deformations” when really it should be “diffeomorphic deformations”. In a certain sense, this “doesn’t matter”, because if the derivative is infinite, the action will be infinite, and there will be no contribution from exp - i S . At any rate, this is how we arrive at Z = ∫ [ d ϕ ] exp - i S as a weighted sum (measure) over an infinite-dimensional space of homotopic deformations away from the classical, minimal action. And, yes, of course, for various simple toy models in 1D and sometimes more dimensions we can actually write down a Hilbert space of a countable infinity of orthogonal polynomials, and apply a weak topology on this space so as to get the convergence properties that we want in physics. So all this is (self-)consistent and well-explained in textbooks.
The part that I don’t understand is the anomaly. We can take the vector currents, ad quantize those fine, but the axial vector symmetry cannot be promoted to a local symmetry, i.e. cannot be quantized locally, because something is preventing the extension “in that direction”. And this vector vs. axial-vector construction has a mirror in how Dirac and Majorana spinors are constructed from Weyl spinors. And the Weyl spinors are those things that we construct from the Clifford algebra on a Riemannian manifold. This suggests to me that we should be able to perceive the same vector vs. axial-vector constructions for harmonic maps, in general, where was say that the harmonic maps is the vector part, but there’s also an axial counterpart, and the harmonic part is subject to variational principles and can be quantized, but the axial part cannot be done variationally. So somehow, all these are inter-tangled, but I don’t see quite how.
What’s more, the axial anomaly is not renormalizable, this is the famous renormalizability issue in perturbation theory. But with the above paragraphs, perhaps his is less surprising: the axial current is just saying that we cannot do variations (perturbations) in “those directions”. So non-renormalizability should not be considered a surprise, *if* we adequately understand the anomaly. Which I don’t.
Lets put it a different way. Conventional textbooks on the intro to differential geometry introduce and articulate the tangent manifold T M to a manifold M and it has the same (real) dimension as M , and at a point p ∈ M its a vector space T p M of the same dimension, and all of this can be tied together into a symplectic structure, given a solder form (a canonical form), with some geodesics and variational eqns and vector bundle connections which decompose into horizontal and vertical bundles. All that machinery is developed. And here’s the oddball part that confused the fuck out of me: all of this machinery is for the vector part. There is also an axial vector part, and its completely unnoticed, with exactly zero remarks or statements made about it. You don’t see axial vector current until you study frigging nuclear physics, which as actual protons and neutrons and pions, and then and only then, do you see pions as a vector gauge particle, and some confusions with other particles that have some oddball quantum numbers, say J/psi, or whatever the fuck, and these couple to an axial current, and then and only then do you start plumbing the depths of the axial current and the anomaly and the broken symmetry and the non-renormalizability. And then if you continue down this path, you can eventually discover the axial currents bobbing up and down in the Majorana vs. Dirac constructions, and continuing to go backwards, you find that spin structures are a generic aspect of Riemannian manifolds. But this journey is never completed. We never go back, and say “a hah, see, for every tangent manifold there is also an axial-tangent manifold and it is totally fucking different because you can’t do variational principles on it, and it has no geodesics, but it is there, because products force it to be there.” This last sentence may be crazy, but some variant of this last sentence has to be true, as otherwise there would be no issues with quantizing the anomaly. But there’s ... just ... something I don’t understand, in this big heaping mess.
Whatever. So here, I wrote a bunch of text, but did *not* learn something I didn’t already know. Well, perhaps there is one. I can formulate the anomaly via S U ( 2 ) L × S U ( 2 ) R but this is done for isospin fields (nucleons and pions). However, I can also formulate the Lorentz group S O ( 3 , 1 ) as the adjoint rep of a product of a conjugate pair of S L ( 2 , C ) so where did that axial product go? I suspect I am being a total moron here, and any practicing physicist can say “here Linas here’s your conceptual error” but I don’t get it. And that’s why, I suppose, I’m not a prof at some university. On the other hand, I’ve spoken to more than a few profs at more than a few universities who would not be able to answer these questions, either. Steven Weinberg would have scowled at me and called me a moron. But the rest of them kind of don’t even understand the issue. Oh well. So it goes. Later. I’ve got things to do.

13 December 2024

I spent the last week fighting with my cell phone company, installing Ceph on larger disk drives, replacing failed power supplies, reading Tong’s General Relativity and working on Archeo, a system to discover file corruption.
Tong’s book is nice. Its simple, easy, written directly. Now, I’ve already encountered all of teh content in it 2-3-5 times before, so its easy because I’m a know-it-all. But the refresher is nice. OK, I’d never seen the Penrose diagram for the de Sitter universe before, so that was actually an interesting stop-and-think moment. Anyway...
Archeo is arguably a stupid time-wasting project, which makes me wonder again about wasting time. At the same time, I can use it as a proving ground for fundamental research. Here’s how. Step one: prototype it in utterly conventional sqlite3+python+flask, just like everyone else. Promptly hit some standard usability limits with SQL.
What’s a typical “SQL usability issue”? Well, say I have a large table, tens of thousands or millions of rows, and I want to know which subsets of rows are similar to other subsets of rows. Well, SQL was never designed to do this. You have to... well, its a solvable problem per domain: does the table contain employee names? Shopping carts? Gene sequences? Movie casts? Depending on the table content, we can solve this “similarity” issue in a domain-specific ways: “show me all movies that Cary Grant acted in” or “show me all reactomes that the BCRA gene upregulates”. For Archeo, it would be “find all files that have the same file hash, but also the same filename and a different hash, order by date”. If two files have the same name but different contents, this is a strong indicator that one of the files is corrupted. If, out of three files, two have the same content and the third doesn’t, then maybe the two that agree are correct. Or maybe the oldest is correct, and the newer one got corrupted.
This similarity thing is a fundamental issue in population genetics, and in human population studies. Genes can jump from one bacterium to another, SNP point mutations can happen anywhere, and humans moving from one geographical region to another can pick up or discard the local language and cultural aspects, or not. A highly mutable, mutating web of inter-relationships.
I do not know of any generic solution. I do not know of any generic (mathematical) theory that adequately describes such mutational networks. I need to ask Razib Khan. Hes here in Austin. BRB. OK, done. I’m back. My peer network sucks. Why am I only tripping over this now?
OK, let me finish writing down the thought sequence. I have attempted to implement a generic “similarity measure”, this is in the OpenCog Learn project. It was highly focused on language. For the Archeo project, I eventually realized that I needed exactly that, but for file systems. Although it took me a week and a working prototype to get to that point.
So there I was: a week into a prototype written in 100% conventional mainstream tools. This was in the hopes of attracting conventional programmers to help out. Hah. But those people don’t exist, just like you don’t exist. The problem is that this is a diary, that no human being will ever read. So I am only writing for an imaginary you: the real you doesn’t exist. This is a problem faced by all diaries. Consider Ramanujan’s diaries. Famous mathematician. His diaries have been published, in book form, and major PMA (physics, math, astronomy) libraries have a copy. But who has read it? Fucking no one. Well, not no one, exactly. I’ve read mayb 5-10 pages, maybe 20 pages out of this diary, and skimmed a hundred more. But who hes read this end-to-end? There are like maybe seven volumes? They are fat! OK, yes, maybe there’s a dozen people in the world who have read *all* of Ramanujan’s diaries, end-to-end. So, not no-one. But a dozen is not a large number. Ramanujan was famous. I am not. Dead end. Ergo, you don’t exist.
Same deal for the Archeo project: I’m catering to future collaborators, who will almost surely never-ever arrive. A set of measure zero. So, fuck it. Let me write it in OpenCog Atomese and screw em if these non-existant assistants find Atomese too challenging. So here I am developing the Atomese framework, and once-again pondering old Atomese design choices. Why does Atomese do this, but not that? Why does it support this but not that? Why does it have the API that it has?
Well, the obvious answer is “history”. The other obious answer is “modular design”: it was designed to solve a spefic class of problems, and that means that it needs certain kinds of functions grouped into modular units, exposing certain data types and API’s. Software engineers have the innate ability to create “good” API’s, and modularize code “correctly”. But how does one know that its good and appropriate? Why is this design choice good, and that one is bad? This is hotly argued in various forums. And so? What’s the answer?
Well, there is something that software engineers “do”, when they group together functions of a give sort, and don’t group other things. Presumably, there is a functional similarity that drives design choices: the engineer imagines a broad network and clusters or groups accordingly. How should we perceive that network, and how do we cluster appropriately?
Well, LLM’s do this kind of clustering, but for linear sequences of input data. However ... hmm. I spend too much time writing diary entries and fucking with code, and not enough studying theoretical foundations.
I’m going to the American Diamond Age party tonight. I have to get ready for that.

14 December 2024, Midnight

Dear diary. It’s 12:45 AM and I just came back from the Diamond Age party, and I’m drunk and hyperactive from the party, where I spent approx 5 hours being hyper-social, turning up the charm index to 110%. And now I’m home, and my wife Patty, who did not experience the pumped up kicks of a party is not ready to be a firehose of unbundled energy that I am right now, but she is also not ready to got to sleep, and its not like I’m ready to pass out either, so I will direct my hyperactive bundle of energy at you, dear diary. Who knows where this will go, because I am not socially constrained to listen to you, and instead I can write wild and drunken shit here and who fucking cares.
I guess that’s the keyword: “shit”. Intellectually, I’m done, I’ve visited all the usual haunts, I’m incapable of “normal” reasoning at this point, and ertainly have no desire to undertake the reading of any kind of mathematical or conceptually rigorous writing. Or thinking, for that matter. So auto-pilot. Maybe more focused than what I do when sleeping, which really is a dream state of unfocused shifting attention. But less focused than the prolem-solving state where I work on technical matters. Such as programming or reading math papers or whatever.
Hmm. I’m now munching on food. The party had a good buffet which I did not avail myself of. Alas. I am recalling and pondering the varied conversations that I had, which is a good thing, because some of them were worthy, I guess.
So you, dear diary, let me share and agonize the meta issue. Is it worth it to go to parties and socialize? Maybe, and maybe not. I know enough pop psychology to repeat any number of theories and insights into party-going behavior by humans, and how a variety of different personaity types behave when placed under those conditions. I will not attempt to recapitulate thiw “well-known” stuff. I will instead furrow my eyebrows and go off into that la-la land of uncertain destination that is a social killer at parties. I will silently purse my lips and say “I don’t know how to respond to that.”
More ffool you: my verbal centers continue to type at this keyboard while my intellectual centers are in a tangled know of “wtf” and are knocked senseless. How can I possibly respond now? What can I say? What were we talking about?
Well, for starters, the pleasure and pain of going to parties. The pleasure, because social interaction is fun. The pain, because somtimes you are (I am) caught with a stupid person, and you’re trying to figure out how to turn an unexciting conversation into an interesting one, which takes intellectual effort and tap-dancing.So I find that not fun, and walking away from someone and thinking “wow, that was not fun” is, well, how can I explain it: it’s not fun. Almost all conversations I had tonight wer really pretty good. This crowd was intellectually far ahead of the Less-Wrong monthly meetup crowd. Like maybe even two gear-shifts ahead.
And so this is the issue: what’s the point? If I want stimulating intellectualism, theres any number of math papers I can read, sober, paying attention, focused. And if that is too... too .. too... something, then there are a variety of other activities, semi-intellectual, that I can engage in, all of which are kind of more rewarding than going to a party. I mean, lets compare party conversations to reading twitter posts (or interacting on twtter, to be fair). The party conversation, well, I can have more of these, and at a more rapid pace, than twitter. But the twitter stuff is more intellectually uhh, err, refined? More tunable to my interests? I’m interacting with a recommendation, alog, though.
So, what am I getting from interacting with people? Pop psychology and real-life experience shows that many people hate parties, while others love them. What can be gotten from them? What’s in it for me? For me, personally, it seems “very little”. Yes, of course, there’s “networking”, which people claim can lead to a job and more, but honestly, you have to do a fuck-ton of it before it can do that. Well, what else is there? The things that I enjoy doing, and if you do textual analysis on this diary, and the time-stamps of my github checks, and the time-stamps of my twitter posts and discord chat logs, you can have a pretty good idea of what I do all day.
The value question is “why should I do what I do?” and the conventional answer is that one does what one likes to do, and this is a certain kind of mind-less, brainless journey through time, where you take in the sights, eat lunch, and then mindlessly resume whatever activities you normally engage in mindlessly. And this is how I’ve lived most of my life, and this shows. What have I acheived? Doodly squat if you use conventional acheivement metrics. and if one uses less conventional “enlightened” metrics, then I’ve acheived a lot: personal happiness, enlightenment, inner peace and harmony, mastery of math and physics and more. Perhaps less sex than I wanted. Just enough drugs, maybe the right balance. ock-n-roll, meh. I ingested a metric fuckton of music when I was younger and now I have brain-worms. “In the corner of my mind, sits a jukebox. It’s playing all my favorite mel-o-dies. One by one, hm, hm, hm, hm hm hm, of the days when your were mine...” Fucking musical brainworms. And not enough sex. But finally, I understand string theory, or a sufficent subset of string theory that I can pat myself on the back about it, and confidently say at parties “I understand string theory” and people don’t care if I say that, but I feel like I’ve actually acheived something.
And so what? There was a fairly large number of high-acheivers at the party, although as one person pointed out, they were mostly B-listers. And they used J.D. Vance as an example of an A-lister. What the fuck am I supposed to do with that? I nooded my head. I don’t think of J.D. Vance as an icon of anything at all in particular, except that he’s famous enough that I recognize his name, but I don’t really know why. Like is he a republican or something? One of these guys who makes ass-o-holic statements? I don’t know. Why are people who make assohlic pronouncements considered to be A-listers? This only sends me down further down rabbit-holes of human psychology, mass appeal, the spread of mind-viruses, and whatever other social, cultueral, economic, political systems that propel people like J.D Vance to the intellectual fore-front of some kind of something going on that I don’t know what the fuck it is, except that it clearly weilds consierable political power. And what the fuck do I know, maybe even more sex, drugs and rock-n-roll.
I’m curious to understand the mathematical underpinnings of the social-economic sysem, yet this is such a daunting task that it seems impossible. And why? what for? whats the use? A number of people tonight asked how I monetize my ideas, and what can I say? I don’t, I don’t have some narrow, hyper-focused innovative market-idea that solves some customers problem. Not one that venture capitalists are willing to invest in. And I don’t have poltical power, star power, some innate responance that makes people want whetever it is that I’m saying. Like, who the fuck is JD Vance, and what the fuck is he saying that makes him into an A-lister? I mean, I finally understood why certain movie actors are A-listers. But I finally understood this not from watching their block-buster hits, but from watching random casual, minor skits and performances, almost off-camera but not quite, and you see them and go: “holy fuck this person has got talent” in that no one I knew in high-schoold could do that, and no one in this room right now could do that, and no one I’ve met could pull that off, and so I go, “yeah, that’s just fucking talent” and no one begrudges the respect. Not just “mad props” but some dozens of levels above that. But I have no clue about the nature of high acheivement in politics. Clearly, you have to get elected, but there’s more than just that. And I’m clueless and I’m less drunk than I was an hour ago, and now its 1:45 AM and much of the mania has passed and I’m getting sleepy.
But the central blob of concerns remains: WTF? and what should I do? and what does it all mean? and if I’m so happy, conten, well-balanced, then why the fuck am I writing neurotic diary entries in the middle of the night? Yeah, that’s a good word. Neurotic. Self-obsessed, compulsive, sel-critical behavior... I should really look up the word “neurotic” but I’m pretty sure that the more precise definition I will find there will adequatly encompass, in single-word fahsion, the complex bundle of vague feelings and emotions and thoughts that are causing me to compulsively type at this keyboard, hunting for possible catharsis. A catharsis that has been elusive, ever since it came to the forefront of my thoughts. When did it come to the forefront? I think I can date it: it is when I started writing this diary. Up to that point, I was content to live my life on auto-pilot, doing random shit without thinking much about why I am doing that random shit. Somehow, with this diary, I have discovered a way of turturing myself about it.
But maybe that is just old-age looking back at life kind of thing that all old people do, to one degree or another. I’m fucking 65 and I don’t like it and I’m fucking retired and I don’t like that either. And I’m no longer handsome. And my life, the acheivements or lack there-of, what the fuck was that all about? What the fuck am I doing now? What will I do in the future?
These are rhetorical questions. Right now, I will go to sleep. Tomorrow morning, I will look at my Ceph cluster, maybe work on Archeo a bit, maybe ponder wha’s wrong with OpenCog, maybe search the net for some new ideas in phylo-genetic, Ising-model mumbo-jumbo theories that maybe or maybe not are applicable to the informational dataflow problems that I’m looking at, and for what? Because its fun, in a certain way, but this is just more brinaless enjoyment of life and retirement, avoiding the hard work of trying to figure out how to get more sex, drugs and rock-n-roll in my life. Which is a metaphorical statement, not literal. Except for the sex part. That’s literal. Why the fuck is this essay so fucking neurotic? What the fuck do I do to cure this state? I mean, is this how it will end? We’ll build a Jupiter-brain, and it will get self-absorbed in neurotic thoughts? Being a human is weird. And if you think that’s weird, just imagine what its like to be a Jupiter-brain. I don’t know what I want. And it’s strange to think that maybe some super-AI won’t know what it wants either. It’s not that I am manic-depressive, but rather that I’m wavering between the idea of “I wish I had a hundred lifetimes to do everything that I want to do” and “what the fuck do I want?” and “you fucking moron, your life is over, you’ve retired, you wasted it all, you blew it, here you are, deal with it.” Retirement takes a psycholgical toll. Well, fuck old age, followed by dying, that takes a psycholgical toll. Here I am, screaming into the void. Fuck all. I’m going to bed now, for real this time. BTW, it is now 2:08 AM for those who are counting.

14 December 2024, Daytime

I just wrote the below for the Archeo project, but it is more appropriate for this diary. So I copy it here. I no longer care where I put things. My personality type has been a bit fastiduous, in that I like to put things where they belong. It’s never been obsessive-compuslive, but clean and organized enough to be socailly passable, and perhaps more clean and organized than most people care to deal with. However, with these deisgn ruminations and meta-recursive thoughts, I’m not sure where to put them or how to organze them. So they spill in an unorganized heap, perhaps into this dirary, perhaps into design documents elsewhere, perhaps into formal academic papers and writings and perhaps not.
My thought patterns seem to be increasingly unconstrained. The negative connotation is incoherent. A more neutral tag is to say they are dream-like, or stream-of-consciousness, with one thought following another in a jumbled, disorganized manner. Is this good? Is this bad? Of course, *all* humans think like this, as do all animals, and perhaps all plants and whatever other objects in the universe you want to ascribe thinking qualities to. However, what “good engineers” and “good scientists” (and “good artists”, and politicians, etc.) do is to control their stream of thoughts into a focused and coherent direction, pursuing some specific, desired goal or objective. This allows the thoughts to then be organized into specific categories, folders, notes, from which scientific papers can be written, from which engineering solutions can be obtained, or from which political action committees can be steered. Organization is necessary for progress, disorganization is symptomatic of poorly adapted behavior. Disorganization is a conventional killer of small businesses, businesses that cannot keep customers because they are too distracted by bullshit.
And so comes my neurotic question here: is my unconstrained thinking, is it a period of “artistic inspiration” and “creative activity”, or is it symptomatic of chaos? I would like to think it’s the former, but self-regulation means one should always question if its the latter.
The meta-problem here is that I’m thinking about AGI, and I don’t know how to organize those thoughts. They spill out randomly, and when I try to place them into buckets, I find that they are prematurely solidified, formed into a crooked ugly mess. Like building a model with glue-covered fingers: nothing goes where you want it, everything sticks to your fingers. The sheer act of trying to get organized results in sub-optimal thought-groupings. I mean, not just sub-optimal, but bad and incorrect. When I try to organize my thoughts about AGI into well-structured collections and well-written essays, I find that the result is wrong, broken, besides-the-point, misleading, a dead-end, lacking in insight, poorly structured. So how do I structure my thoughts, when the very act of structuring leads to undesirable crap? When I am literally incompetent at creating the things I have set out to do?
This is, of course, a problem that all humans struggle with, knowingly or unconsciously. I’m doing it here knowingly, but... Look, well is this neurotic behavior, or not? Is this a distraction from the flow state, where I am thoughtlessly and automatically writing code, reading texts on general relativity, or what have you? Flow states are mildly pleasant: they occupy and preoccupy your mind, and time flies when you’re having fun. And then one day you wake up older and wonder: where did the time go? Well, it went down a flow state, and the reason you are not rich is because you spent your waking hours in a dream-like state, pursuig a pleasant activity, instead of managing your finances and making money, which is what you should have been doing (or, at least, this is what your society tells you that you should be doing.) And oh fuck, here I am with my circular thinking, back to wondering about the social signals that tell human brains about what is moral, good, desirable, and what is bad and distasteful. Which then gets us to the battle of Western culture vs. other cultural paradigms.
So far, we know/beleive/accept that Western culture is good, because it has created conditions for economic dominance, unleashed in part by the unleashing of creative talent (and wow, how tose creatives have disorganized thoughts, which they then organize into buckets). The unleashing of creative thought requires ascension through Maslow’s pyramid, and Western culture provides the economic well-being that allows this ascension.
So lets ponder eschatology for a moment. What’s the difference between a mess of atoms, and a molecule? Well, the mess of atoms is chaotic, random, mixed and unformed. A molecule, on the other hand, organizes those atoms into a specific structure that is stable over time. But the distinction between a bunch of atoms and a molecule is subtle, and if one is not careful and observant, one will simply overlook the structural relationships, and instead see a chaotic mess of atoms. You can take a piece of meat and stick it in a blender and get goo, and you can argue that the goo is almost nearly totally identical to the meat, so what has changed really? Some conventional measure of statistical mechanics says “almost nothing at all”, and that the piece of meat and the goo are nearly indistinguishable. Of course, this is wrong: the instrucments of statistical mechanics lack the observational abilities to find structure. So this is a known shortcoming of the present-day theories of physics and chemistry: the tools have very poor abilities to find structure, talk about structure and to characterize structure.
And now on to the eschatology: just like the stransition from atoms->molecules involves a minor, subtle shift in structural organization, so also, the shift from one form of social, cultural and economic organization to another subtle, (not so subtle?) change. The common discourses about crime, (trans-)sexuality here in the US are manifestations of the economic freedomes that have allowed Western culture to flourish. This is obvious to all, and is a part of the central poltical debate: which freedoms can you remove and regulate, without killling the golden goose that powers economic might?
I keep not getting to the eschatology part. If we have this complex organism called “civilization”, which includes thigs like “psychological disposition”, and “communications patterns” that encourage or prohibit beleifs in Flat Earth, faked moon landings, or that Trump is a hero, then how does this civilization, uhh, acheive greatness, err, economic might? I mean, China is an economic powerhouse, but medical care in the US is 100x better than in China, as is food safety, environmental pollution, policing and the Justice system. Never mind the psychological oppression of the 15-minute city. So economic might is not an adequate measure of “human development”. Again: nothng new here: you can look up the “Human Development Index” (HDI) on Wikipedia, and see the dozens or hundreds of scholars that have labored to obtain a reasonable measurement of what it means to be an enlightened, happy, healthy human.
The economists who developed the HDI are like engineers: they labored to create an optimal measure of “something” (human development), by grouping together assorted positively-perceived cultural aspects (health, education, etc.) that the culture already percieves as positive: everyone knows that being healthy is a good thing, and everyone knows that being dead is a bad thing. I mean, this goes back to mammalian roots, or animals in general: health and development apply generically to living things: trees and shrubbery, not just owls and cows.
I guess that now is a good time to finally copy in the paragraphs that I wrote for the Archeo project. Here they are:

Meta issues (from the Archeo README file)

The engineering/design meta-issue rises again. How do I know that it is important to have file witnesses? How do I know that file content hashes are the things that need to be recorded? How do I know that the objects I’m dealing with are files? How do I know what a file system is, and how do I know the characteristics that it has? How do I know that these abstractions are the appropriate ones to make, and that this is the correct design?
The answers to these questions are all obvious to the (software) engineer, because the creation of software designs are the primary activity that software engineers engage in. But how does the engineer "do" these things? How does the engineer accomplish them?
The obvious answer is that the engineer goes to school, gets trained and learns from experience on how to convert verbal descriptions of software artifacts into functional computer code. The engineer reads lots of specs, remembers how things work. The engnieer measures performance, and thus knows what designs are performant. The engineer factors and refactors code, and thus develops a taste for what things "go together", what a module should be like, and what a good API is. These are all learned from experience.
It is perhaps incorrect to think that there is some simple magic wand to be waved around, thus obtaining modular design and API wisdom. It is tempting to think that perhaps there is some Bayesian similarity factor, or perhaps some Ising model that can be applied to software systems, and that by letting this model percolate, an ideal module and API pops out. Perhaps there is such a system. It would be very un-human-like in how it arrives at optimal solutions.
Perhaps I could even design an Ising-model-inspired lambda-calculus combinator and optimizer. But this just shifts the location of the problem: how do I know that Ising models and lambda calculus are the appropriate abstractions?

Coda

So the above is all an example of my unstructured thoughts, and now perhaps you can see why I have difficulty structuring them. I don’t even know what I am doing. Am I writing philosopphy about civilizational eschatology? Of course not. There are plenty of people debating this online. I am not going to give Alexander Dugin a larger fucking audience. Am I creating software? Err, well, sometimes, often, even. Am I trying to assemble a mathematical theory? Err, yes. I’ve already noted the disconnect between philosophy and computation: if you cannot take your philosophical insights, and convert them into some category-theoretic formulas, then you cannot create software code that operates with those philosophical ideas. So a passage through mathematical formalities is necessary.
So what am I doing? Pondering civilizational eschatology and trying to turn it into mathematical software. Of course I’m fucked, and left to stew in disorganized thoughts. Perhaps the task is too big. Psychohistory is a topic of sci-fi; what’s that famous book from the 1950s? Isaac Asimov, Foundation Trilogy. Read it as a kid. Have no memory of any of it. Blah blah Hari Seldon. I’m not Hari Seldon.
What a mindfuck. I think I’m going to stop writing now, and enter a waking dream state, where I flow and write some software or something.
Oh right, that and I need to perform google searchs on terms that relate the statisitical extraction of structure in lamda-calculus-like complex networks of mutating, rapidly-evolving networks, which is what I think I am trying to do with the AtomSpace, and where I have ignored the work of others who are working in this field. An Ising model of jigsaw-puzzle pieces. How does this work?
Good-bye diary, I must hibernate now.

20 December 2024

Time for another agony session. Wherein I express my doubts, reservations, loss of vision and general dissatisfaction with the state of things. And ponder the future.
So, the general bundle of thoughts starts with autonomous agents, sensory perception, extraction of structure, theorizing about AGI, and my personal relationship with a world economy and a $100 billion-dollar AI industry. Once again, by writing and spilling random words on the topic, perhaps I’ll stumble over an insight. And perhaps arrive at a plan. Where do I begin?
Let me start small. In this particular project, the one containing this diary in a git tree, I have a certain set of research directions I want to pursue. They were going well, until I realized that my main bottleneck to further cpu-cycling was the batch-processing nature of how I was doing things. Trying to move to a continuous-feed pipeline rapidly led me down a path that leaves me in the dire straights I find myself in now.
As I write this, I realize I’ve tangled together multiple distinct issues. Perhaps I can untangle them. My batch-processing system has very distinct processing stages, where one stage finishes, before the next one starts. I realized that what I really want is some of the advanced processing stages to run as soon as practical, even while flowing new data into the earliest stages. So, a continuous processing system. A modular system, perhaps a multi-agent system. Together with a control system. That is, as later stages run dry, and decide they’ve done all they can, they need to turn the faucet on in the earlier stages, and get them to do more processing.
The last three sentences already open a vast expanse of technology issues for which the AtomSpace/Opencog has no solution, and for which there is no off-the-shelf, commercial, open-source, industrial solution. The buzzwords: “continuous processing”, “agents” have been used by the computer industry for decades. Yet, there’s no generic solution. Let’s deep dive as to why. I will ponder things that are perhaps “obvious”, and this is why everything is so difficult: the “obvious” is like a brick wall, blocking progress. It has to be deconstructed, pounded through, in order to move past.
So lets look at “agents”. Thanks to Unix in general, Linux in particular, I can associate the concept of an “agent” very roughly with the concept of a “process” running on a “kernel”. To do this, I have to buy into the Unix technology stack. If I used some other operating system (e.g. MS-DOS), I might not get multi-processing capabilities. If I used no operating system at all, I simply wouldn’t have that, and would need to create some kind of agent system from scratch, writing thousands, maybe tens of thousands of lines of code. The computer industry, as a whole, has come to a general agreement that multiprocessing is good, and has also converged onto a common shared idea of how to do it. In biology, this is called “convergent evolution”.
But there’s more. Different OS’es provide different kernel API’s, and I need a C library to hide those diferences. Again, most of the industry has standardized on a common C library standard, although the largest player, Microsoft, had to be dragged, kicking and screaming, to the table. If I have a C/C++ app, I have some guarantee that I can compile it for most modern, popular OS’es, and it will just work. Mostly. Microsoft still sets up roadblocks. The BSD’s, Linux, the various unixes, Apple OS’es and Windows are similar enough that system daemons, those that don’t open windows or use keyboard/mouse interfaces, that these can just work. Android remains a challenge.
A process running on one Linux machine, and “agent”, can’t really be “frozen” on one machine, moved to another, and “thawed”. Well, now it sort of can, if I use LXC or Docker. I still cannot move an LXC container to Windows, Apple or Android. I think VMWare offers something along those lines, if I really, really want it. But as a commercial product, a bit stymied. Great, if I need to deploy some short-term solution. Terrible, if I want to make some long-term strategic technology choice.
How do agents communicate? Well, IPC on one machine. TCP over the net. There are a zillion protocols layered on TCP. That’s great. This works only because the entire world has agreed to use TCP, and not IBM SNA or Novell Netware, or DECnet or whatever. The competitors have been pushed aside, allowing one dominant technology to rule provide a generic common infrastructure that is guaranteed to be there.
However, one reaches a limit. For Atomese, for the CogServer, I had to design my own specific network protocol. There was no off-the-shelf system I could use, above and beyond raw TCP/IP. Perhaps, maybe I could have layered on something else. There are several dozen different techs, offering something RPC-like, something mailbox-like, publish-subscribe services that could be used, and its not too late: the CogServer could still be adapted to use one of these services, *if* it solved an actual problem that I had. So far, they solve problems I don’t have.
So what problems do I have? Ben wants me to use Hyperon. But Hyperon also doesn’t solve my problems. The meta-lesson from the above two-three paragraphs of obvious recaps of obvious tech landscape features is that to build new systems, one must figure out how to leverage the work of others, and that one must make strategic bets on future software & hardware technologies. One has to surf an edge of the mainstream wave with thousands of developers behind it, while also, well, doing the thing you want to do. So again, I ask, what do I want to do?
Well, starting from the extremely narrow idea. Once again, I want a system that crawls over my text input, does the pair counting. So it needs to be autonomous in it’s crawl. What does that really mean “autonomous”? Currently, I have a script. I start it, it runs. It walks the directory tree and shoves the texts through the system. I could call it a daemon, an agent, if I wanted to get fancy. Fundamentally, its shoving data through the system; but what I want is a system that pulls. That makes the deicision “gee, I recall reading this text before, let me go look at it again, make a second pass”. Well, but how?
I need an agent that remembers what it read, where it was found. Stores some meta-data. Where is that meta-data stored? Well, in the AtomSpace, of course. How? Well, I would be forced to, uhh, engineer some data structures to store it in? That would suck. How do I store it? Well, the straight-ahead answer is as pairs-properties. EdgeLinks with PredicateNodes to tag the content. So its all there, all addressable, all searchable. Whoop-de-do. Now what?
There are several issues plagueing my thoughts. Lets do the easier, more obvious one first. This is the library-science, systems-engeineering problem. If I suck in a lot of data: lets say electrical circuit diagrams, I need to store that data somewhere, because it it probably won’t fit in RAM. But, just like the deconstructed URL’s in Archeo, I don’t need to store the deconstructions, just the minimal amounts, which I can then expand as needed. But this requires having some on-demand expander system. How does that work? I also want to avoid having to engineer and re-engineer expanders, I want something generic. All this needs to also be a distributed, multi-agent system. Running on multiple machines. The CogServer/StorgeNode gives me this, sort of. It has a performance hit. It’s only the base layer. There’s no unified storage model, where I can ask any server, and it knows what all the other servers know.
So I’m fiddling with Ceph, and I like how it’s agents work. There’s a ceph-mon and a ceph-mgr per machine, and these talk to each other, to all the other deemons in the network, and these all know what each of them know. Could I do that with CogServers? How?
Well, one issue is maintaining a coherent read-modify-write view of the world. Yea old ACID/BASE debate, eventually-consistent, etc. I’ve layered the AtomSpace on top of RocksDB, because I wanted to handle this consistency issue on my own. But the storage design allows me to also layer on other distributed DB’s, but with a performance cost. And the consistency/coherency issue never quite goes away, because the AtomSpace is in-RAM so I would need to keep some current-generation counter, much like Ceph does: a quorum of servers that agree on the current time-slice. And use some blah-blah algorithm to decide who’s-got-what when there are netwsplits or systems go down. So again, we are doing systems engineering here. And not AI research. And there’s the rub: If I’m doing systems engineering, why not do it for... well fuck if I know.
So let me ignore those issues. They are all solvable using a conventional engineering mindset. The issue is that I don’t have an infinite amount of time; someone else needs to solve these, and tehy are not getting solved for the AtomSpace/OpenCog, and Ben is not solving them for Hyperon, either. So we’re in a hole, here.
Where were we? So, we are ingesting English text, or electronics netlists, one paragraph at a time, one netlist at a time, and either we store a URL to the disk location of the original dataset, or we actually import the raw text or Spice file or whatever, into the AtomSpace. If we import the whole thing, then the mean ugly issue of having enough space raises it’s head again. So its better to leave it on disk, and just record the URL.
So, how does this work? Do I have a URL-recording agent? Who engineers it? How does it work? Does it have an Atomese API to it? The problem with having an Atomese API is that no one uses these, because the meta-level stuff isn’t working yet. Because .. why? Is the meta-level stuff to vague and cloudy? Doesn’t have an approrpriately articulated deployment scenario? Let me be clear: last week, I wrote a python-Atomese file-system-crawling “agent”, and so fucking what? I engineered it, I implemented it, it works. Now what? It does have a sub-module that computes the hash of the contents of a file. I could modularize this, so that the function that computes the hash is replaced by a generic module that computes anything at all, and stuffs the results into the AtomSpace, when invoked.
Is this an improvement over the shell-script that I use to shove text through the ’learn’ pipeline? Well, I guess it could be. Shell scripts are hard to turn into daemons. Why do I want a daemon? Well, first of all, the file crawler has a config file, that tells it where to crawl. The config file is there for the human user, to edit. That config file should be in the AtomSpace. And it should be editable with a web GUI. And there should be a button that says “config has changed, reexamine, reload, restart.” And conversly: what config are you using right now? And, as a daemon, some way of saying “Are you there? Are you up? What’s your status? Are you working?” Its taken me a while to warm up to SystemD, but I think I get it now. Having a uniformized, common system to do it all is actually a good idea. The analogy here is that I need a generic Atomese system processing daemon, one that can infest input from a file system (or from URL’s via crawls, whatever) and apply processing modules to what it has gone over, and keep a general record of what it has done so far. And I guess I will have to engineer this, because it won’t just drop into my lap.
The problem that I have with the above is that the system requirements here are so freakin generic that it sounds like 1990’s computing buzzword bingo. There are 101 systems that can ingest data, record it, process it. On the cloud, we got OpenStack that specializes in this stuff, so why exactly am I reinventing cloud computing? Well, because the data generated and kept by these other systems is not accessible to me, in Atomese format. And why do I need this data in Atomese format?
Ah, there’s the rub. The simple, stupid answer is that this is how the higher processing layers determine what has been done before, and decide how to go back and do it again. But that’s the stupid answer because I do not yet posses higher-level layers capable of doing this. And I’m not sure how and when I will possess them. And this is the key issue that I’m trying to work towards, and keep not quite getting to, and now its late, and I want to go for a bike ride before it gets too late, so, dear diary, its coitus interruptus time again. Perhaps we’ll do some cognitive fucking around later today, or maybe not.
Well, a few more minutes, a few more missives, before I pull out. My general framework of computing similarities is also incomplete. I’ve got a working proof-of-concept, its in a semi-generalized format/domain. I’ve discussd a crawler in the above paragraphs, that could be built in a more generic fashion. And I’ve got a slightly firmer view of what a web UI might look like. And just talking about this has me stoked to build it.
And that’s the next issue: the manic-depression. The mania is that I’m stoked to build this and that. The depression is that I’m the only one who is stoked; I don’t have a user-base. I don’t have helpers, assisntants, collaborators. The other half of the depression is the feeling of emptiness: I still don’t have a well-formed plan for arriving at AGI. Just some obvious steps in front of me, and then everything is obscured by a curtain of trees, clouds, jungle. And this is how explorers do things: one step at a time, erring through the jungle. And I can say “buck up kiddo, one step at a time, through the jungle, its forward progress, no matter how you think about it!” And meanwhile, there’s some asshole in a fucking biplane flying over the jungle and waving to you and yelling “cholley ho, old chap, you should see the view from up here”. And I want to be that guy, with the view from above, instead of being stuck in the weeds of system engineering. But he can fly only because he did a different set of system engineering: building an airplane, levergating advances in metalurgy, pistons, motors, propellers and shellac’ed cloth and strong, lightweight wire for the wings. Well, if the goal is to do botany, you get on your hands and knees with a hand-shovel and whatever. And if the goal is to get a survey of the AGI jungle, well, the airplane is the thing. I’m very very worried that the one-step-at-a-time approach of building a configurable crawler and building a generic similarity-analysis infrastructure is ... the wrong thing to be doing. And the reason I worry that it’s the wrong thing is the meta-jungle. Fuck-all. There are several metas involved here. One is the socio-economic meta: that airplane pilot has to come to ground, and deal with buying gasoline, eating dinner, and getting roses for his love. And so I have to find money, make bill-pay, obtain collaborators on my project.
Here’s a meta-example from that world: In order to keep some system daemons running autonomously, you need humans to supply that computer with electricity and reboot it when it is down. This is done in one of two ways: the computer software does something useful, e.g. cell phones, or generates money, e.g. crypto-mining. So getting rewarded with crypto for keeping an AtomSpace instance up and running is something that ... Well, SingularityNet set out to do this, but they, I dunno... got distracted? FileCoin was/is supposed to do that for IPFS, but IPFS is crap. Compare IPFS to Ceph, and you can instantly see why IPFS is crap. At any rate, being actively coupled into the human economy is how a technology item is guaranteed life. It can be natural, like cellphones, which evolved to fit something humans want, and it can be artificial, like FileCoin, leverage greed and get-rich-quick dreams. I don’t have a viable opencog-crypto get-rich-quick scheme, and neither does Ben, after claiming to be working on it for five-plus years. And worse: this angle is not really AGI, its rather a mechanical coupling of technology into the human world. So we know this is one way to couple in, but it does not solve the actual AGI problem.
Whatever. There’s more. But I gotta go on that bike ride now.

20 Dec 2024 Post ride

Lets continue. Two or three ideas gel in the above. First, I can think of a crawler as a motor, an active agent that “does things”: it explores. The config file is the “motor control” that regulates where it is aimed. During the crawl, data is collected. What data? For Archeo, its hard-coded: the file size, the content hash, the mtime. But this too could (should be) specifiable in the “collection tool”: the type of data that is gathered during a crawl.
Coming up with reasonable designs for the motor control and the collection tool specs is ... a challenge, and it seeems like more prototyping and proof-of-concept work is needed. An issue that muddies the water is that these controls probably have to be representable as config files, and then viewable and/or editable with a Web UI. In the long run, we want other parts of the system controlling the motor directly. But for now, it needs to have human oversight.
Next, we have the task of measuring directory similarity. There are two distinct issues. One is that there are many different ways of measuring similarity, each emphasizing different aspects. So, this requires a “similarity tool”, specifying what it is that should be compared. Given the richness, its again not clear how to specify this.
Clouding the above is that I developed the matrix tool, for measuring similarity for items arranged in a sparse matrix. Directories are, in that sense, not sparse. But then, I don’t want to compare “anything to anything”; I already have blown up directory contents into a specific collection of Atomese, and its really about tracing through incoming sets of Atoms (i.e. “find all files with the same name” is really about finding the incoming set of a node with some name, but I already have that incoming set, so finding is trivial.) Some kind of rethinking is needed for understanding what “similarity” means in this context.
Is this doable? Yes. Perhaps even in a 3-6-12 month project. Should it be done? Well, I don’t have much else on my plate. Is it silly? It sure feels silly, because it is so low-level and basic. Flip side, no one has done this silly and basic thing before. And if they have, it doesn’t matter, because they done it wrong. So the decision is like that of undertaking any new software (or hardware) project: what’s out there is no good, and won’t do what I want. I’m telling myself this to psych myself up, because otherwise, it feels like I dunno, a stupid project? I can’t tell. Is this brilliant? Or dumb? Well, I know I’m smart, so I should learn to trust that little voice in my head that says “Linas, this is brilliant”. The only real problem I’ve got with that little voice is that it makes me do things that society finds little or no value in. Its odd that this is a plight shared by both scientists and drug addicts. You don’t get rich studying high-redshift galaxies. Actually, you have trouble staying employed. I missed my chance to become a famous and acclaimed scientist. I’m now older and wiser, but senescence is working against me now. I need to be a bit more focused and selective with my time left on Earth, and so I spell out a project like the above, and wonder, “is it dumb”? and “should I do it?” What a fucked existance.
The fact that Ben sucked all the air out of the room, by dropping the Atomspace in favor of some new, yet to be created system... is this a blessing or a curse? I hate that decision, but then, I’ve hated many of Ben’s decisions. He’s so fucking bone-headed, for such a brilliant guy. The silver lininig is that I don’t have people clamoring for my time and distracting me. And so I’m free to pursue novel and new ideas. Whatever. Lets do this. Could be interesting.

23 Dec 2024

Got another one of these annoying “how did a photon know xyz?” bullshit confusions. This time I think I found a reasonable common-man analogy. Worth setting down here, in my diary.
kc kim
1:05 PM (1 hour ago)
to kc, bcc: me
FirstHedger (@Zero Hedge) posted: My First Physics Class Professor Asked:
How Do The Photons Know Where To Go?
Faster Than Light?
It’s An Illusion
Per All Illusions
Previously Set Up
Linas Vepstas
2:15 PM (21 minutes ago)
to kc A common mis-perception by laymen, and even many physicists (who should know better) is that "photons exist". Rather, the photon is the size of the chunk of energy that can be exchanged during an interaction.
Think of it this way: something costs 99 cents, and you buy it. That does not mean that you had a "single item" (a "photon") that was exactly 99 cents in your pocket. Those 99 cents never actually "existed", except as a concept in your head. However, you were still able to buy something, because you managed to pull some change out of your pocket, which just happened to total up to 99 cents, and conduct a transaction. The quantum version of this is the "measurement of a photon" or a "wave function collapse".
I’m not sure why physicists don’t fix this mis-perception in the popular media. They seem to like the idea that everyone runs around with bills in their pocket for exactly 99 cents, and 98 cents and 97, and everything in between: bills for 98.14159265358979 in your pocket. That’s nonsense, of course, but is the root cause of all these questions of "how did the photon know XYZ?" It didn’t because photons don’t exist. Just like the 99 cents in your pocket magically "knows how much a candy bar costs".
Think about it. "how did the money in my pocket know that a candy bar costs exactly 99 cents? Did the candy bar send a secret message faster than light?"
Feel free to post my email on twitter. Ask other physicists if they agree. I’d love to hear a dissenting opinion from someone smart.
--linas -- Patrick: Are they laughing at us? Sponge Bob: No, Patrick, they are laughing next to us. kc kim
2:21 PM (15 minutes ago)
to me “Photons “ Travel @ Light Speed : c
They Experience No /Zero Time During Travel/Existence
~~~ Linas Vepstas
2:30 PM (6 minutes ago)
to kc Photons don’t exist. They experience "zero time during travel" in the same way that the 99 cents in your pocket experiences "zero time" magically transmuting into a candy-bar.
Yes, minkowski space can be sliced into space-like, null and time-like manifolds, but you have to actually study mathematics to understand what those words mean.
FWIW, photons experience zero time and also zero space: the distance between where they start and where they end is zero. They stitch together different points in the space-time manifold.
But that’s kind of like saying "money stitches together the world economy". The 99 cents in my pocket is identical to the 99 cents anywhere else in the world, and can instantly bounce between bank accounts. Photons are like that.
--linas

The End

This is the end of Part Ten-B of the diary.