The ICO Bubble
Josh Nussbaum
4949

I’ve read the white papers on the Blockstack website, and I do not see any analysis on how the fork of consensus hashes is incentivized to converge. I am not referring to a fork in the underlying blockchain where the virtualchain consensus hashes are stored, but a disagreement between application nodes in the virtualchain as to which consensus hash is the valid total order. I have done quite a bit of research on blockchain consensus algorithms for my own project in development (Bitnet) and I am thinking that the virtualchain concept can not be guaranteed to converge to consensus without centralization of the application nodes who decide. Is there some PBFT protocol involved?

Since you claim you have invested in Blockstack, I presume you must be aware of this issue and can give me an answer? (Otherwise, you seriously should not be investing in technology you do not adequately understand.)

Edit: in other words, the design appears to be incentives incompatible, in that if there is a double-spend of sufficient value, then opposing forks in the virtualchain have a vested interest to not converge on one fork, i.e. they disagree over which of the forks from the double-spend forward is the valid one. It appears that fork*-consistency does not improve liveness and afaics never did BFT algorithms have anything to do with insuring incentives compatibility, i.e. the assumption of no more than f of 2f nodes being dishonest has nothing to do with incentives that would cause nodes to disagree about consensus. Fork consistency depends on honest clients, but double-spending is due to a dishonest client. The no-join property means that the fork consistency is impossible without ideal consistency in the presence of dishonest clients. Additionally even ideal consistency of BFT algorithms says nothing about incentives.