Decimal digit computations as a testing ground for reasoning about probabilities — LessWrong