I should clarify what I mean by this. Dash's Evolution proposes a feature where a N-of-M quorum of master nodes can instantly confirm a transaction by sending their signed confirmations to the block chain. This quorum M is chosen by using historical block chain hashes (well it was current block chain hash until I pointed out to Evan that was insecure in this Altcoin Discussion forum, and so I assume he changed his design after Illodin pointed out the improvement).
So the theory is that even if there are orphaned chains, they won't disagree (conflict) on these quorum announcements. I pointed out to Evan that if he changes the quorums on every block then there will surely be conflicts. So again I assume he altered his design to not change the quorums very frequently, thus thinking he had fixed the problem. I decided to sheepishly be silent and let him go down that direction, knowing full well that it is flawed.
Isn't it much worse than that? If I own a majority of masternodes, I can have my faulty nodes agree to confirm transaction A, dump all their votes of the their mempools and then confirm a double spend B with another merchant. None of the votes will get baked into the blockchain by that point if I'm quick enough.
Are you assuming the masternodes have a majority of the PoW? Masternodes are orthogonal to miners.
But in another way yes the Finney attack (and you don't need a majority of masternodes, just have the next block solution hidden then send your transaction to the quorum then announce your block solution with the double-spend) and pretending they didn't receive the propagation of the instant confirmation. Yes this is another problem of allowing transactions to be confirmed by instantly by quorums and also alternatively by normal spending on the PoW block chain.
It is ambiguous which is the truth, because that scenario could genuinely happen due to propagation delays and no one knows if the miner is the double-spender so penalizing the miner is a DDoS attack hole. The attacker could do this continuously (ever 1 second or whatever) with many small transactions and force all block announcements to be double spends.
There are probably other flaws along these lines. I know that without my two fundamental innovations (1. including all chains, 2. transactions can only be confirmed by quorums or the single designate), it can't be secure in any reformulation. And I still have to convince everyone (and myself via peer review) that my innovations are secure for all the scenarios.
This turns the cost of a double spend under instant X into a constant proportional to the amount of locked collateral I have, which is far worse security than regular POW, which is super linear in the number of blocks.
Could you unpack that for me? I didn't follow your math thought.
InstantX required the funds to be presigned over to masternode. I believe the idea in Evolution is all UTXO are eligible to spend through quorums without any presigning.