The conference submission cap isn't a fairness problem, it's a symptom of a system that hasn't caught up with how research actually happens today. The researcher who posted this story is stuck in a familiar trap: a large lab, a borderline NeurIPS rejection, a resubmission to ICLR that landed worse reviews than the original, and now one of their precious 20 submission slots feels wasted. They're asking whether ICML will impose similar limits and where else they can turn. Our view is clear: caps that treat every author as an isolated individual ignore the reality that modern machine learning is a team sport, and the current approach is punishing junior researchers for their supervisor's success.
This tension between individual quotas and collaborative scale isn't new, but it's getting sharper. We've seen how Exploring the Productivity Surge in China's Machine Learning Research has reshaped the field's output, and we've also covered how Understanding NeurIPS Recording Rules for Presenters adds another layer of logistical friction. The pattern is the same: well-intentioned rules designed for a smaller, slower field now strain against teams that can generate a dozen viable projects per cycle. The cap at ICLR, 20 submissions per author, sounds generous until you realize that a single large group's output can consume most of a junior member's quota, especially when resubmissions from earlier conferences eat into the same pool.
What the original poster describes is the human cost of this mismatch. They aren't asking for special treatment; they're asking whether a system built to prevent spam now traps legitimate work in a loop of random reviews and shrinking opportunities. The randomness they mention isn't an outlier, it's a known feature of peer review that becomes more painful when every slot carries opportunity cost. For someone far from the top of a large group, the message is clear: your work competes not just on merit, but on timing and luck, and the cap makes bad luck more expensive.
So what should ICML do? It could follow ICLR's lead and adopt similar limits, or it could try something different, perhaps a per-lab or per-group cap that reflects actual research capacity, or a resubmission budget that doesn't compete with new work. The specific answer matters less than the principle: any rule that forces a junior researcher to choose between iterating on an old submission and submitting something new is a rule that needs rethinking. Until then, the practical takeaway for anyone in this position is to treat the submission cap as a strategic constraint, not a moral judgment. Prioritize projects with the strongest signal from earlier rounds, and consider venues outside the big three, workshops, journals with rolling deadlines, or specialized conferences, where the review process may be less crowded and more predictable. The system isn't going to fix itself this year, so your best move is to navigate it with clear eyes and a realistic plan.