Hacker News new | ask | show | jobs
by bcoates 4572 days ago
I don't think this exactly right -- an evil node can't send a false nay because it would have to be able to forge the conflicting transaction, the signature prevents that.

Bitcoin already essentially does the 2PC that the author is asking about for unconfirmed transactions. The problem being solved is that the Byzantine Generals problem is unsolvable for anonymous actors, as a malicious participant can create a majority of evil voters, winning any dispute resolution by 'Sybil attack'.

The proof of work system allows a newly joined node to determine the current consensus even when it's disputed, without having any idea who is on the network, so long as it has at least one link to the true hashing-power consensus. It also acts as a commitment protocol -- once you've signed your winning block to the network, it's nonrepudiable even by you.

With an alternate source of identification, a pseudo-anonymous 'Infocoin' ledger should be able to function and scale just fine without all the PoW expenditure--or in other words, you must have a system of making identities expensive, and Bitcoin's is Proof of Work.