At 2:38 AM 9/3/95, MONTY HARDER wrote:

But if the escrow agent is anonymous, we simply recurse, moving now to the question of whether anyone can trust the Anonymous Escrow Agency not to take the money and run.

Well, at one level, everything is always recursive. But seriously, the reason it differs is that the escrow agents are in the business to be escrow agents, not to take the money and run. This actually works in the real world. (Game-theoretically, they get a certain payoff in defecting, but then forego the revenue stream from future transactions....) And, of course, it is possible to structure things so that the escrow agent cannot "take the money and run," because the money is not accessible to them. This is analogous to an escrow agent in the real world holding a check from Alice to Bob until Bob completes some set of conditions. The escrow agent--call her Essie--cannot cash the check herself. She can of course renege on the deal, even if Bob does his part of the bargain, but there is little incentive for her to do so. One can ask what payoffs and costs/benefits exist for various actions by these agents, and game theory needs to take a look at these sorts of transactions, but there are plenty of reasons to believe "convergence" will occur (basically, that "effectively honest" behavior will be common).

TC> (I mention banks because, when you look at it closely, today's banks can TC> quite easily claim that a customer made a withdrawal when he didn't. That TC> they don't says more about the nature of persistent businesses than about TC> any government oversight or security features. This is a side point, but it TC> bears keeping in mind that the real world of banks and businesses, etc., is TC> not fully secure, either. And yet it mostly works pretty well. The reasons TC> for this are interesting to consider.)

A bank has $$ invested in impressive-looking buildings, (so that vanishing into the ether and setting up shop elsewhere is rather difficult) and several officers whose TrueNames are registered with the appropriate agencies, so that they may be sued if they pull this crap.

No, my point was that it is fairly easy for any bank to scam any particular customer, given the flaky nature of verification of withdrawals and other transactions (at least for smallish amounts of money). Few banks check signatures, few banks bother to demand much ID, etc. The issue is not whether you can sue Bank of America, but whether you could win. That banks don't often scam customers for small amounts of money is testimony to the fact that they've got a better payoff matrix element in being a bank than the meager payoff in scamming a few customers. I submit this not as proof, but as evidence that the type of "convergence" mentioned above mostly works. Most commerce hinges on this, not because of law enforcement. Like True Names, the threat of law enforcement is only a part of the overall equation. --Tim May ---------:---------:---------:---------:---------:---------:---------:---- Timothy C. May | Crypto Anarchy: encryption, digital money, tcmay@got.net 408-728-0152 | anonymous networks, digital pseudonyms, zero Corralitos, CA | knowledge, reputations, information markets, Higher Power: 2^756839 | black markets, collapse of governments. "National borders are just speed bumps on the information superhighway."