[krapp]

That doesn't seem likely given his ego and personality and apparent tendency to micromanage.

The problem is the premise that one can effectively decide the value of an employee's work by counting LOC or reading their output in that manner at all. It's dumb whether it's done with a REPL or a stack of paper.

[khazhoux]

You can find the lowest performers in an org pretty easily through a combination of reviewing what they’ve gotten done the last year (via self reviews, manager reviews, a recap writeup etc) and looking at their commit history and docs or other artifacts created.

High performers challenge this notion because they say “I have very difficult bug fixes that required little code.” But high performing engineers just don’t realize how little (shockingly little) output actually comes out of low performers. Like, a slow trickle of trivial code, and tasks that stretch for weeks for no good reason besides their inability to overcome obstacles

[krapp]

>You can find the lowest performers in an org pretty easily through a combination of reviewing what they’ve gotten done the last year (via self reviews, manager reviews, a recap writeup etc) and looking at their commit history and docs or other artifacts created.

Yes but that isn't what happened. By all accounts no such process took place at Twitter.