- Immunity to 51% attack is an incorrect claim, because the block chain hash determines which quorum, thus a chain reorganization can rewrite which quorum was authorized to M of N sign.
- The instant confirmations can not be trusted because if they are on an orphaned chain (not 51% attack but just the normal process of orphan rate or even 25 - 33% selfish mining attack), then they can be reversed. This is the same reason the prior InstantX (a more limited scope for instant confirmations than Evolution because the inputs had to precommitted to the quorum yet this committed could still be overturned by an orphaned chain, a.k.a. chain reorganization) was a technical failure.
Apologies for being clueless, and I see you said that would be your last comment on this, but perhaps you or someone else can still answer - InstantX could take a hash of a block that is old enough to be considered safe from reorg, wouldn't this solve the issue?
You compel me to reply because that is a strong idea.
That will likely open up new game theory scenarios since you know a priori which quorum will receive which inputs in which future blocks. Perhaps an adversary can try to create masternodes to target inputs to accept or deny signing.
As for the claimed 51% immunity, how can one assume if the 51% that owns the mining power doesn't also own a large chunk of the masternodes thus can wreck havoc on any minority chain which desires to respect the signatures of other masternodes for which the 51% attack is not, i.e. a war between groups of masternodes.
Thus the masternodes could be restarted to by your change to infiltrate the quorums of the minority masternodes (up to the N - M + 1 threshold, another reason to set M very close to N
), so they can remove instant transactions from the minority chain, thus making the claim of 51% immunity false.
Game theory is complex. There are likely many other scenarios of attacks.
The basic problem is using a hash to determine which masternodes are authorized (which is either subject to chain reorg or subject to game theory due to being known well in advance) combined with the fact that masternodes are not fungible but rather are entities owned and with motivation to maximize their gains.
Now having said that, masternodes earn very high payments from the block chain (I saw some chart Evan published that showed up to 50% per annum interest payments) thus I think for now the incentive is for masternodes to be cooperative. Afaics, Dash is basically designed (whether consciously or by serendipity) to funnel all the coins to the insiders over time.
But we are also discussing in the context of what would be the correct design for a coin that was not designed to be unfair and award an advantage to a few entrenched owners of the coin supply. In that case, we'd have to dump the masternodes and find a more fungible design that doesn't have these vulnerabilities. Dash may not actually have these vulnerabilities because the masternodes are making a killing by just cooperating. In a coin design where masternodes weren't leeching most of the coins away, then these game theory vulnerabilities would come into play because the masternodes would have an incentive to collude in other ways to maximize their profits. Yeah we can buy the illusion of decentralization and robust design by paying some masternodes all our coins over time. That works.
