Proof of Fraud

[illdeletethis]

during a conversation about cryptocurrencies with untrustworthy names in the bitcoin.de forum i came up with a to my knowledge not yet existing method of minting coins for a cryptocurrency

PROOF OF FRAUD

what it should be is a proof that someone did a doublespend attack on the blockchain of another cryptocurrency

as i am currently unable to code anything of that kind i am putting the idea out for anyone to build on

so can anyone think of an algorithm by which you can without excessive bandwidth/blockchain memory use prove both that a doublespend happened and that it was done from an adress you control?
If you have an algorithm, does it require one of the transactions to be in an alternate orphaned block (means effectively you need a 25% attack for creating a demonstrable doublespend without there being issues such as the 0.7/0.8 split)?

[oriolpont]

I'd say that this could be possible, yes. You can never be sure which is the longest chain (e.g., in case of private mining). However, you can show, through signing, that you produced two conflicting transactions on alternative branches of the prey blockchain and that both got buried deeper than the confirmation depth consensus (6 blocks or whatever), that is, a double spending.

Of course, to generate a Proof of Fraud (PoF) block and prove it is valid, you need to include these preyed branches in it, and you require that the fraud (on the prey chain) occurs exclusively after the preyed chain portions included on the last PoF block. An important problem, though, is that if a previously unknown longer branch is discovered on the prey chain this would invalidate the PoF branch too (since it has no difficulty parameter to measure its length). This is not worse than it is for the prey chain, but it makes the PoF chain actually a slave of the prey chain.

PoF is thus a pointer to PoW, so it relies on hashpower exclusively and does not resolve any of its problems. Furthermore, it is a winner-takes-all reward system, in which you need that a single party hashes 51% for it to work at all.