When the hypotheses make sense, the tests for them work and the next steps are based on the previous test results that's not "random stuff". I made that explicit in the first message. This problem wasn't solvable with random guesses.