Finding a fair review process in a sea of random acceptances

In the evolving landscape of academic publishing, the integrity of the review process is crucial for fostering innovation and quality.

3 min readMachine Learning

The review process at major conferences has become a lottery, and pretending otherwise does no one any favors. When a paper's fate hinges on one reviewer who missed the point and a meta-reviewer who lacks the spine to push back, the system stops rewarding rigor and starts rewarding luck. That is not a judgment on any single author's work; it is a structural reality that too many of us have experienced firsthand. The question is not whether the process is broken, but what you can do when your best work lands in front of someone who simply will not engage with it.

Here is the practical truth: you cannot control the randomness, but you can control how much of your career you stake on it. The commenter's frustration is valid, and the search for a venue with a slightly less chaotic process is a reasonable instinct. But the deeper issue is that no conference or journal has fully solved the problem of human bias, fatigue, or simple incompetence in peer review. Some venues have tried, with structured rebuttals or reviewer training, but none have produced a system that feels fair to the people on the receiving end. That does not mean you should stop aiming high, but it does mean you should diversify your targets and treat a single rejection as data, not as a verdict on your work.

What this means for you is straightforward: build resilience into your workflow. Submit to venues that offer transparent review criteria, and read those criteria carefully before you send anything out. If a venue does not publish its review guidelines, treat that as a red flag. When you do get a rejection that feels off base, do not internalize it. Instead, extract whatever technical feedback is useful, ignore the rest, and move on to the next submission. The meta-reviewer who goes along with a misinformed reviewer is not a gatekeeper with wisdom; they are a bottleneck with a deadline. Your job is to keep your work moving past them, not to convince them they were wrong.

The takeaway is not that you should abandon the pursuit of quality or stop caring about where your work lands. It is that you should stop treating acceptance as the only measure of value. Build a portfolio of work that stands on its own, and let the venues sort themselves out. When you do find a venue that consistently gives thoughtful, engaged feedback, support it with your best work and your reviews. That is how you nudge the system toward fairness, one submission at a time. The randomness will not disappear, but your exposure to it can be managed.

From Machine Learning

Major conference acceptance has become pretty much random and review quality is constantly dropping.

​There is always that one reviewer who understood nothing but still rejects the paper because you didn't cite "X" or compare with "Y", and the meta-reviewer usually just goes along with it. In your opinion, is there a conference or journal with a solid review process that is even slightly less random than the others?

Read the original at Machine Learning