@pikhq it's funny, to me it makes the most sense at the polar opposite ends of code - things that will only be run once, so it's reasonable to not care about bugs outside the intended input, and things that are extremely thoroughly tested, like every input and output is known and the challenge is to make it complete them faster on particular hardware limits. At my most cynical I wouldn't have guessed how many people act like their work is in the first category. Like I know software broadly is looking for the "next big thing" at all times but a project like "Windows the OS" doesn't need faster sloppier commits, it needs to be a stable platform for people to use computer