How do we evaluate people for their technical leadership?

Programmers’ chief value does not come from producing code. It comes from knowing enough about the situation to solve automation and scaling problems. Code that doesn’t solve those problems is worthless to the organization. If code that doesn’t solve those problems gets deployed somewhere, it’s worse than worthless because it incurs maintenance costs in addition to being worthless.

Perhaps even more importantly, though, they ensure that the wrong thing does not get done. And that is what makes them look worthless: their chief outcome is the absence of something. […] It’s the breaches that didn’t happen because someone double-checked what this SQL did. It’s the multi-server failure that did not occur because two separate engineers saw the warning signs and one responded to the other in a messaging channel. It’s the disaster feature that got killed because someone who understands inclusion demonstrated how it might be used to attack a marginalized constituency.