i have played with go-ethereum's code in past and that sounds like something that could be implemented as an experimental consensus algo. Have you ever built some sort of model for this reward mechanism?