Re: Distributed distribution
For the next generation of underground Internet marketplaces, distributing the computation and storage among several servers ought to mean no individual sysadmin need ever be in possession of enough information at any one time to identify any activity positively; although there would still be room to incorporate some measure of redundancy.
Implement the same database schema on multiple server instances; then corresponding records on each one contain some mix of real and fake information, chosen to provide a sensible balance between plausible deniability and multiple redundancy (so a complete correct record can be assembled, even if one or more servers becomes unreachable; but it is not obvious, even with the full set of data, which one of several equally-plausible possibilities might be correct). The distribution of real and fake data among the several servers need not be the same per record.
The client's ordering system would distribute records, each containing a mix of some correct and some bogus data, among several database servers.
The vendor's dispatching system would query the details of a transaction from enough servers, at the very last moment before they are required; separate the real data from the false data; and then ensures the details are forgotten again immediately after they are no longer required (for instance, once the shipping label has been printed, there is no need to keep a copy of the purchaser's address anymore).
I think this idea actually has legs.
Note: In the event of any related patent dispute, the foregoing, being published and date-stamped on a public Internet forum, may be considered evidence of prior art.
Biting the hand that feeds IT
