Design of experiments

There was something that came up at work today that made me think of traditional science. Specifically, it made me think of the scientific method.Whenever you try an experiment, you need to figure out what the success criteria are before you start the test. This is the same regardless of the fact that we're doing something in software versus something that one does with chemistry.We ran into this today. We had something that we wanted to try -- but we didn't know what would be good enough. You can run all sorts of experiments, but if at the end you simply shrug and say "well, that's good enough" you aren't setting up the experiment well enough.