Why ML conference reviews sometimes feel like a “lottery“ [D]

About this article

I’ve been trying to make sense of all the “ML conferences are a lottery” takes, and honestly I think it’s both true and not true depending on what you mean. If a paper is clearly strong, like genuinely solid contribution, well executed, easy to understand, it usually gets in. And if it’s clearly weak, it usually gets filtered out. The weirdness people complain about mostly lives in the huge middle where papers are good but not undeniable. That’s also where scale starts to matter. There are ju...