Code review is dead. Long live code review

blog.codacy.com

5 points·claudiacsf··2 comments

>"Low-risk, well-covered changes can flow through automated gates without waiting for ritual approval. A utility function with 95% test coverage and no security findings does not require the same scrutiny as a change to your authentication middleware."

That is exactly the problem I'm solving right now. As a software engineer in the AI era, we ship code like crazy. We already move ten times as fast. But the bottleneck is still there. In the past, the bottleneck was actually writing the code because it was the most expensive and slow part of development, but now the slowest part is actually reviewing the code. And pretty much all the solutions I've come across do AI reviews or add AI summaries or add a wall of text to each pull request, which makes the review harder and longer instead of easier and faster.

You still have to read the code anyways. even if the AI reviews the code, the output is still on you. AI is just shifting the bottleneck.

The way I'm approaching this problem is to use AI to reduce the stuff we have to review instead of increasing it. The AI is not that good at dissecting very complex code or architecting better solutions, but it's good at identifying noise or things that are safe to merge. So we could use the AI to handle all the noise and keep the important stuff that we need a human review for. GitHub is no longer a good place for reviewing PRs nowadays.

Agreed the move is AI reducing what you review, not adding to it. What's also needed is a process for classifying changes by risk so the low risk PRs get the automated gates while a human looks only at the high risk one that actually need it. And then enforcing that the same way across every repo.

Reply on news.ycombinator.com