PJ Media

When I have written some computer code, it isn’t enough to evaluate it in the manner that seems to be standard these days with scientists — that is, to simply verify that it behaves as I expect it to (i.e. that “observation matches theory”) — the code isn’t tested merely by observing it do what I want or expect it to do, but rather by trying deliberately to break it. And that deliberation is never a guarantee that I’ve thought of all possible ways to try to break it — my familiarity with what I expect of the code can be an obstacle.