free paper idea inspired by sophia malamud talk i saw. in table model theories there's a natural way to define 4 levels of commitment:
—put p (or ~p) in your DC [i.e., fully committed about p]
—put {p, ~p} on T [i.e., fully uncommitted about p]
—put {p, ~p} on T and order PS so CG+p > CG + ~p [i.e., p is likely]
—put {p, ~p} on T and order PS so CG+p < CG + ~p [i.e., p is unlikely]
sophia used real numbered commitments. That's a lot of commitments!! do we need? can we get away with just these 4