Quote from: monsterer on December 18, 2015, 11:27:12 AM

Quote from: TPTB_need_war on December 18, 2015, 02:27:25 AM

Realize that normally masternodes can't lie because it is deterministically determined which masternodes can lock which transactions. All the masternodes can do is sign the lock or not sign the lock.

Some masternodes could attempt to sign locks which they are not designated to sign, but the PoW block chain would never honor these locks and would not put them in the block chain. And payees can do this same verification before accepting the InstantX transaction.

Remember we are talking about 0 confirmations here, there is no POW evidence at this point, all that exists is mempool evidence. Payees might well be able to do that verification, but that's not in the protocol.

If I own a majority of masternodes, there is a greater than 50% chance that any one of my nodes will get picked by the deterministic selection process, so they can indeed lie with a high probability of success.

Lying doesn't help a masternode in terms of a transaction which is mathematically invalid, e.g. spends non-existent UTXO or spends more than the value of a UTXO.. Such lying can be provable detected by any full node observer.

However, there was an Alzheimer/grandpa moment of being too sleepy to finish what I had realized upon awakening yesterday wherein I had written you were correct about other attacks from masternodes. The statement above was just a locally (not holistically) logical statement to make on some aspect before I slept, but the original more systemically holistic thoughts I had from the morning had been lost by that time where I was too sleepy to process thoughts. When I awoke today, I remembered my logic which I didn't get to write down yesterday because the discussions got so (beneficially) sidetracked and then I got sleepy.

After reading the points I make to illodin in the prior post about inability to prove propagation ordering, the issue is precisely that there is no way to prove the ordering of announcements from masternodes. So even though masternodes can only approve InstantX transactions which they are deterministically authorized to approve, the flaw is that any conflicts that arise due to relative ordering can't be proven to be the fault of any of the involved masternodes.

A contrived example is paying to an address from an InstantX transaction and spending from that address in another InstantX transaction. It can't be proven that the second transaction was issued after the first, thus it can't be proven whether it was an invalid transaction. That particular example can't be fixed by requiring all UTXO payees of InstantX transactions to be blocked until after the next block confirmation, because again nothing can be proven from propagation about what should and shouldn't be blocked.

The challenge is to think of an example where it results in ambiguity about double-spending (which could either admit a double-spend or forking). Can you think of an example other than the Finney attack example I was discussing with illodin?