Hmmm, interesting take. I could see this for a more mature engineering culture with experienced devs where conventions are clearly established. Though, I agree with your criticisms of code reviews being somewhat of a placebo. Pair programming and code coverage thresholds would be more effective, but many companies don't want to adopt those practices. So if you don't do those, devs can get sloppy and overconfident, which makes code review much more necessary. Not as effective, but necessary. Also, code review is often when patterns and conventions get discussed and revised, especially with remote teams, so best to address those foundational choices before merging as well. Or do more work upfront to have those discussions with an example branch, but it's a long list of choices for any project and many are not foreseeable.