Recently I pushed code to production without reading it.
Looked at the title. Looked at the description. Asked an AI agent to review it. The agent wrote something or other. Out of the corner of my eye I caught a green emoji — click, click — and the request flew off to prod.
It was a planned trick, to be fair. I know the codebase by heart, there are plenty of tests. A kind of Russian roulette: where you used to switch off a data center and watch what fell over, now you commit with your eyes closed. And if it falls over, you say it was a planned stress test of the AI pipeline.
What is code review even for?
Diffusing responsibility?
Because Sasha — the one I’ll send the code to for review — doesn’t read it either. He knows that code from Dima is top quality. So Sasha is simply ready to put his name under it: that if there’s an incident, he’ll show up to the incident review.
Might as well schedule that meeting in advance, no? Instead of an Approve button, an I’ll be at the incident review button.
Maybe it’s about shared responsibility?
But if you take away the collective responsibility — and really, are we to blame for the actions of a group of people? (Or are we?) — then why keep this old tradition at all, of sending someone your code for inspection?
So then: what is review for?
Learning. Studying. Because if I don’t send Sasha the code — where else is he going to look at it?
He doesn’t even have an editor open. Just a terminal.
So we send each other code for review as the only remaining way for a programmer to look at code at all. Because nowhere else will he ever see it.
So send each other code, I guess.