Tim May <tcmay@got.net>
Here's a meta-question: Suppose one holds highly secret or sensitive data, for which one wants to use an Eternity service to ensure the information is not suppressed by some government or other actor.
Why centralize the data at all?
Why not just use the "pointer" to the data and offer to provide it?
Which is what Blacknet was all about. Instead of focussing on a data base, focus instead on an untraceable market mechanism.
I am more and more seeing the similarities between BlackNet and Eternity USENET. The only real difference that I can see is that for E-USENET I have been talking about periodically broadcasting the data to allow very high security for the reader, and also the idea of keeping a local copy of the documents so that data is pre-fetched to speed up accesses. Both designs are relying heavily on the anonymity provided by remailers. For very high risk traffic, even using mixmaster remailers may be risky due to the various active attacks which could be mounted by a well resourced attacker with ability to selectively deny service. An eternity service or blacknet information provider could frustrate the active attacker by having many software agents with different network connectivity and using these resources unpredictably.
(I admit that a system which can provide *A LOT* of data *VERY FAST*, and also untraceably or unstoppably, is an attractive goal. [...] The catch is that I can't see how such a system will get built, who will run the nodes, how payment will be made to pay for the nodes and work, and how traffic analysis will be defeated.)
One big opportunity we have is to subvert protocols of new services. A distributed web replacement with ecash payment for page hits I think is plausible. Web pages could migrate to meet demand. You could have a hot-potatoe effect, where high risk documents are not kept for long -- bits move faster than government agents and lawyers -- hot data could migrate every 10 minutes. However distributed web replacements are complex to design, and whether it will be possible to deploy the system widely is an open question.
And I think implementing the slower-but-no-breakthroughs approach (Blacknet or variations) has some advantages. It may be many years before we need to be in the corner of the graph that is "large amounts of data--very fast retrieval--very secure."
Most candidates for untraceable/secure storage and retrieval are NOT in this corner, yet. (Kiddie porn may be, but whistleblowing and scientific information are not.)
What about large scale software piracy. This could consume serious amounts of bandwidth. This seems to be intermediate risk in that if one observes even 30 seconds of traffic in #warez, one observes lots of commercial software trading hands. Perhaps the new draconian US software copyright law which the large software corps purchased from the politicians will move software piracy towards the higher risk end. Would the world be better off without software copyright? I tend to think so. Adam