As researchers think about how to improve reproducibility, it's important to remember that failure is a crucial part of the scientific process.