Several people objected with "what if XYZ left the company two years ago?" That points to a process opportunity in how we treat both comments and offboarding.

But first, she's now Katie. That's the name I used in my head. Katie knows this code best. https://twitter.com/hillelogram/status/1309200884535042052

If Katie is the go-to person for an area of the codebase, that means there's a lot of implicit knowledge in her head that's not written down. Fair enough.

But there's another layer of implicit knowledge: that Katie is the expert! People don't automagically know that!

Once Katie leaves, not only are you losing the implicit code expert, you may not *know* you've lose the expert. You'll burn a ton of time asking everybody else what they know.

And Katie isn't going to be an expert in just one thing. How many other things will be like this?

But if we read

# Katie knows this code best

We now immediately know who the code expert is. If she's here, great, if she left, we know we have to start understanding it from scratch.

But why wait until we read it? Handle it while she's offboarding!

Katie puts in her two weeks notice. We immediately sweep the codebase for any mentions of "Katie knows this code best" to figure out what implicit knowledge we're about to lose, so we can preemptively have Katie transfer knowledge. Turn "obsolete" comments to our benefit.

The adoption barriers, though: First, we need to make it easy to do analytics on comments. Seems not to bad, it just means adding structured metadata to comments. Maybe

# !EXPERT: Katie Hu {ref-id} {date}

To make grepping easy

Bigger barrier is this requires companies to be remotely competent at offboarding, and hoo boy Edmodo left me with AWS keys and github repo access for a year after I left, and that's nowhere near the worst horror story I've heard

Get your offboarding shit together everyone