when I see a an error I build a hypothesis about the error before reading into it. then I solve the hypothesized error and verify that it indeed was solved. th

when I see a an error I build a hypothesis about the error before reading into it. then I solve the hypothesized error and verify that it indeed was solved. this works in 9 out of 10 cases and is hugely effective - as long as the ls is fast. this is like branch prediction. better the faster you can verify that the branch you predicted was the right one.