There is a danger with publishing hashes that you might publish opposite ones. Ideally, you should be committing to one answer or the other. High entropy predictions mitigates against this, but the effect is still there. And we can't tell if it's high entropy until it is revealed. Publishing dates mitigates against this.

I thought I recently saw an essay on the uses of hash precommitments, including this kind of problem. If that doesn't ring a bell, I guess it wasn't by you.