File-based commerce borrows from the STR (shared transaction repository) http://www.gldialtone.com/STR.htm such ideas as intrinsic reconciliation between parties, and invitational subscription.

The FBW is a client specification which talks with other parties by putting triple-entry accounting transactions up in the sky, on any filesystem, with GUIDs as filenames. 

Triple-entry accounting is a gimmick which makes shared transactions look like double-entry accounting to the unrelated parties in a trade.  Parties always share and agree upon a transaction amount, date and description. This shared entry is like a convergence of the mutual account receivable and payable into one entry. The shared entry plus two private entries (stubs) dangling off of it for each party's private offsetting GL entry is triple-entry accounting:

 PartyOneStub      SharedEntry      PartyTwoStub
(private data)    (agreed data)    (private data)
------------       -----------      ------------
My Rev/Exp         PartyID's        His Rev/Expense
classification     Amount           classification
(etc.)             Date             (etc.)
                   Description
                   (etc.)     

In the FBW, each transaction is an encrypted file, having a B2B message in it, and a private stub for GUIDs pointing to other parties' stubs to the transaction (which are optional, obviously since they are private). Security is controlled by controlling the keys rather than access, or data.  Look, people---all computer security is based on encryption. Either encryption is good, or it ain't.  If it's good, we can park our transactions in public, and escape from the "user account" which delegates power to the host.  User accounts are infantile now that we have bigtime, reliable SANs.

Files on this FBW spec would be read-only. They multiply forever in the filespace. Good use for all those SANs, or the inevitable publiuses and freenets and mojonations on large commercial ASPs. Proliferation of read-only files provides total backwards audit trail to the key holder. Nothing is ever ambiguous. Who needs two-phase commits.   George Guilder says storage is infinite.  Bandwidth is infinite.  Let's use some.

Freehaven, Freenet, MojoNation, Gnutella, Tipster, and Publius offer great potential as decentralized transaction platforms for consumers and Small/Midsized businesses.

The existence of unconstrained copyright violation and other prohibited content on these filesystems suggests the systems are potentially more private than mainstream B2B systems. AT&T seems to be involved in building Publius http://www.cs.nyu.edu/~waldman/publius/    P2P storage has wide appeal in moving corporate communications beyond possibility of subpoena in civil claims. http://www.zdnet.com/eweek/stories/general/0,11011,2618363,00.html  

Platforms that are beyond possibility of regulation, battle-tested by having prohibited content, are secure enough for Small/Medium businesses. The corollary to this is that no B2B service can be regarded as private or secure unless  there is an outcry from the government to shut it down. If the government can crack a system to regulate it, crackers can also crack it to steal.

In an odd turn, thieves may be shifting our core money systems towards diseconomies of scale, towards cell architectures, geodesic and headless, where intruders would need to crack many links undetected, to find value.

Banks evolved for centuries as 'the ledger of record' for people's mutual balances, because thieves made every other system
too risky, regardless whether it was books or tokens. Now, thieves make banks and other large-scale book-entry systems
riskier for small business, than just doing it ourselves. 

* Todd F. Boyle CPA http://www.GLDialtone.com/
* tboyle@rosehill.net Kirkland WA (425) 827-3107
* XML accounting, web ledgers, BSPs, ASPs, whatever it takes