It's not science if you only work on things you know in advanced will succeed. Just like there's no such thing as a "failed" experiment.