Idea for guessing prioritization

[sei-eschwartz]

Right now we make guesses by making the most likely guess and accepting it if there is no contradiction. But maybe it makes more sense to guess the less likely guess and hope that we get a contradiction. If we do, that is strong evidence for the more likely guess. If we don't get a contradiction, we can save the traditional guess for later. This is in essence a way to see which guesses are implied by the current set of facts.

This is similar to the deducation/propagation step inside of SAT solvers that make assignments to variables when there is only one possible value for them based on the previous assignments.