In reply to the OP I propose mechanisms I think preferable to manipulations of rewards curves. Given your competence, I'd appreciate you proving my proposal would not work, and is stupid. If you can't, well, you know what to do.

Thanks in advance.