You'll find lots of that in cycling. Collecting good data can be hard, time-consuming, and expensive, so much of what we know is due to trial and error.

"You manage what you measure", and too often, we've been satisfied by simple models of cycling mechanics based on what's easiest to measure. So a test jig will lead you to think that a single-speed bike with the skinniest high-pressure tires will be the most efficient and fastest. But is it really?