I’ve mentioned this elsewhere, but I imagine a way to primarily considering this is weighing the reward with some kind of exponential discounting which reflects the inherent uncertainty of the mempool and of the future itself. Things further down can be replaced, buried further, trimmed, etc. This would be a fairly large departure from the current methodology of comparing diagrams, involves consequential judgment calls about weighing, and doesn’t answer anti-DoS questions, naturally.
2 Likes