On Fri, Feb 28, 2014 at 12:48:33AM +0100, Jorge Timón wrote: > First of all, sorry for the delayed answer. > > On 2/10/14, Peter Todd wrote: > > Got this: > [...] > Thank you, I knew this wasn't new for us but I doubted we had written > it anywhere. > As said in those mails, being only able to offer AAA for BTC and not > BTC for AAA nor AAA for BBB is enough of a limitation to justify a > hardfork IMO. As usual, you don't need a hardfork. Anyway, one-sided trade is sufficient to get a functioning marketplace up and running and test out the many other issues with this stuff prior to forking anything. > On 2/14/14, Peter Todd wrote: > > You're assuming the seller cares about fairness - why should they? They > > offered a price for an asset and someone bought it; exactly which buyer > > willing to buy at that price was able to complete the trade is > > irrelevant to them. What they do care about is being sure that at > > whatever given price they offered 100% of the buyers willing to buy at > > that price actually see the offer in a reasonable amount of time - at > > the best price the seller will get there will be only a single buyer > > after all so you need that solid proof that said buyer was actually able > > to get the offer. > > In fact, I don't think the seller will care enough about this to pay > the proof of publication fee either. Assuming sellers can either > broadcast the order on a bitmessage-like network or use your proof of > publication scheme, the later will be always be more expensive. So my > prediction is that most people will just use the simplest, fastest and > cheapest method, but I guess only time can tell. You can make the same argument against Bitcoin itself you know... A Bitmessage-like network would be trivial to front-run via a sybil attack. It's the fundemental problem with marketplaces - the data they're trying to publish has to be public. > I don't think this will be a tragedy, because like we discussed on > IRC, I don't think the primary goal of markets is price discovery, but > trade itself. > > About historic data, the actual trades are always public, and some > kind of "archivers" could collect and maintain old orders for historic > bid and asks, etc. And again, how do you know that record is honest? Fact is without proof-of-publication you just don't. > As an aside, nLockTime would be nice not to always have to > double-spend the inputs of an order to cancel it. You mean a reverse nLockTime that makes a transaction invalid after a certain amount of time - that's dangerous in a reorg unfortunately as it can make transactions permenantly invalid. -- 'peter'[:-1]@petertodd.org 0000000000000000b52709f0485161e764ac0198960885ccab019a978322cc6e