I have a practical suggestion. We could try to find a perfect curve with theoretical reasoning, and go back and forth for a long time...or, we could just decide to run an experiment. Suppose HF17 ships with n^2 / (1 + n) which automatically expires, and reverts back to n^2 1 month from now. This way, there is limited downside risk, and we get to actually see how the new curve performs in practice.