It says:
A modification could be good, but there are game theoretic challenges that haven’t been fully modeled.
As much as I like playing with Game Theory, I'm unconvinced game theory will make a correct prediction for choosing a reward curve. It's a candidate for an experiment, and the sooner the better.