You see, the Cryptolocker guys are actually smart criminals. They did not, as you assume (and in doing so, greatly underestimate their cunning and intelligence), enter your supposed infinite ransom loop. Instead, they decided that keeping their word was in the best interest of their "business plan", if we can call it so. They established a weird kind of credibility, where the victims knew paying was the best option, bar none at the moment. And then they delivered.
But here is the big difference. With Cryptolocker you deal with one attacker with a certain identity, so you know they behave in a certain way because they did it in the past. In an P2P exchange you will not make the trade with the same person again (the bank details give the identity). So there is no reason to assume when you onced get blackmailed and the blackmailer acted "honest" that the next time another blackailer will do the same.
Also I am thinking on an idea to introduce a legal contract, that way (you have the identity due the dnak details) you have much protection. But I dont want to discuss that yet as it is half-baked. I will add that to the paper soon if it works out.
But thanks for your input!
Nobody can know now how it will work and if the assumtions are correct. For sure we need to improve stuff as soon we get it out and see how it works and where are the problems. That does not mean that I dont take these issues serious! Will try to find the best base setup to start.