Scientific papers are published in the goal of showing that certain techniques, given a set of protocols and materials, are deterministic. The reproducible character of such small facts is then reused to be the foundation of larger theories.
But scientists have a secret: most of their experiments are showing bad results. The vast majority never publishes those bad results, afraid of being cut down on budget because of them, and public funds are wasted by these practices.
Us software developers have the tendency to take practices from those scientists, but we should try hard not to hide bad results, and communicate about them.
In my (short) career, I failed a lot at trying new concepts and new practices, and fet ashamed at having even tried. Let’s make our community a better community, and start communicating about things that can hurt our peers.