Accepting incorrectness
Nearly every conversation I've been in about software quality treats it as a tradeoff with speed. You can move fast, or you can be correct, and your job as a lead is to pick a point on that line and defend it. I've never really believed in this, and with AI speeding things up further this is a bigger topic. After a few years of being the person accountable for both the speed and the quality of what a team ships, I've come to think that bugs in production aren't really the problem here. »