There's a quiet irony in a researcher asking for simple clarity and getting silence in return. The post about the Epistemic Intelligence in Machine Learning workshop at NeurIPS captures a familiar frustration: a paper's fate can hinge on a page limit, yet the organizers left it invisible. The author emailed twice, checked the site, and even considered the social etiquette of nudging a stranger. That's not impatience; that's diligence meeting an avoidable gap. For anyone who has ever stared at a submission portal wondering if their work is too long or too short, this is the unglamorous reality of academic publishing. It's also a reminder that clear communication isn't a luxury in technical fields, it's a feature of respect. We've written before about how Talking to My AI Clone Taught Me to Question the Tech exposes the gaps between intention and execution in AI tools; this page-limit puzzle is a smaller, more mundane version of that same gap.
The practical question here isn't just "what's the limit?" It's about how we signal competence in an environment where ambiguity is treated as a minor inconvenience. The previous workshop at ICML used six pages; the main conference allows nine. Those are very different targets, and guessing wrong could mean a desk rejection or a rushed rewrite. The instinct to assume the stricter limit makes sense, but it's still a guess. And a guess is a poor foundation for a submission that took weeks to prepare. This is where we'd tell a reader: if you find yourself in this situation, don't wait for a reply that may never come. Look at the workshop's camera-ready guidelines from prior years, check the openreview page for accepted papers, or simply format to the shorter limit and note your flexibility in the abstract. It's not ideal, but it's actionable. The folks behind Unlock LLM Training: A Practical Guide to Distributed Algorithms would likely agree that clarity in process is as important as clarity in code.
What stands out is not the missing number, but the absence of a response. Twice emailed, twice ignored, the researcher is left to crowdsource an answer from strangers. That's a failure of the organizers to model the very epistemic humility their workshop claims to study. If you're running a workshop on machine learning, you're asking people to trust your framework for evaluating ideas. The least you can do is specify the page count. For the author, the lesson is simple: treat missing information as a signal, not an oversight. Either the organizers are disorganized, or they assume everyone knows the unwritten rules. Neither reflects well on the experience. The takeaway we'd quote: "When a workshop can't specify its own constraints, it's fair to question how carefully it will evaluate yours." That's not cynicism; it's a practical filter for where to invest your effort. And for anyone else in this spot, the concrete move is to check the previous workshop's accepted papers, note their lengths, and submit with a brief cover note stating your formatting choice. Then move on. The paper matters more than the margin.