We ran a chess tournament for bugs.

The question we wanted to answer: are bugs from Detail "important"? How do they compare to what code review bots catch?

One of the most important ways we benchmark ourselves is that we want the bugs we generate to be significantly more