I don't understand why CS conferences rely on a peer review system in which real revision is not possible. At @NeurIPSConf for example, author responses are limited to 1 page and there is no follow up; reviewers never get to see a revised manuscript.

In one case, authors could only reply to one of my comments, and said the rest would be addressed in the camera-ready version. It's not their fault. This is a ridiculous way to run peer review, and it exists only because volume and speed are prioritized over quality.

What's ironic is that it ends up wasting everyone's time, because rejected manuscripts are sent to the next conference, with new reviewers and sometimes no improvements. So a paper will get reviewed more times than necessary but without actual quality control.

I think CS conferences should operate two tracks, with a small number of papers that get real peer review process (like a journal), while the rest are workshop papers (to make sure that the conference is inclusive).