[yummyfajitas]

I've seen A/B testing drastically reduce bikeshedding.

I worked with a PM who wanted to cut bikeshedding a while back. When some executive vociferously argued that we should do things B way, the PM just asked him to document why it's important in an email and then told him we'd do it.

A week later an email went out with the executive's strongly worded argument why B is important at the top. At the bottom is "we A/B tested, there is no statistically significant difference between A and B. Guess $executive was wrong."

After a few go-rounds of this, bikeshedding was significantly reduced.

Bikeshedding is easy and fun! But having an email go out to everyone a week later proving statistically that you were wrong is less fun.