Is the CS community more prone to game metrics? I have a feeling most in the area would do that even if there weren’t any external incentives.

I’d go for prime palindromes.

"Ruthlessly optimizing for metric X sounded great, but somehow that failed to give us outcome Y." -Anon