This is an article by Mike Hearn against the proposed RBF (Replace by Fee) "feature", which is the ability to double-spend a transaction by sending a higher fee the second time, which would replace the first transaction by the fact of the higher fee on the second one. This would enable guaranteed double-spends to be executed. Bad idea. This post explains the full ramifications of this if it were to be implemented:
First, let's be more precise. It would enable guaranteed
0-confirmation double spends. Note that 0-confirmation double spends are already possible, but they are currently not guaranteed to work.
One benefit of replace-by-fee is that if a transaction is not included in any blocks because the amount of fee paid is too low, the amount can be increased.
When reading Mike's argument against RBF, keep these points in mind:
- Accepting transactions with 0 confirmations will always be risky, and the risk cannot be reduced without relying on a trusted third party.
- There is no number of confirmations that guarantees that a transaction cannot be double spent.
- Replace-by-fee is inevitable because it is accepted by the Bitcoin protocol and it can increase mining revenue. You can complain all you want about how it is a bad idea, but it won't make a difference.