The moment a conference starts rationing its own venue like a limited-edition sneaker drop, something structural has broken. When NeurIPS passes for Sydney sold out in minutes, and paper authors learned that selecting a preferred location does not guarantee presenting there, the community got a blunt look at how multi-venue logistics can quietly undermine the very principles of fairness and accessibility that peer review is supposed to uphold. The user who posted this is not complaining about a minor inconvenience; they are pointing at a systemic tension: if a "main" venue exists, then every other site becomes a secondary option by default, and that hierarchy shapes who gets heard, who networks with whom, and whose work gets the spotlight. This is not just about travel budgets. It is about whether a conference that shapes the AI field can afford to look like it is sorting participants into first-class and coach.
The practical stakes are immediate for anyone submitting a paper or planning to attend. If you are based in Europe, you are looking at a real possibility of being assigned to a satellite location while the "cool stuff" happens in Sydney. That is not a neutral logistical choice; it is a statement about which researchers are considered central and which are peripheral. The community has seen what happens when review processes lose credibility, as discussed in NeurIPS Acceptance Raises Questions About AI Review Justifications, and a venue lottery that feels arbitrary will breed the same kind of distrust. We would tell any author who asks: do not assume your preference form is a promise. Plan for the possibility that your designated location is not where you want to be, and push the organizers to publish the allocation criteria. If they cannot justify the assignment process transparently, that is a signal about how much they value the people who make the conference possible.
There is a broader lesson here about how we design technical communities, not just conferences. When we build systems that claim to be accessible but then concentrate resources in one physical place, we recreate the same barriers that AI tools are supposed to remove. The related discussion on Exploring Real-World Computer Vision: Deployments, Edge Models, and Current Challenges shows how practitioners care about deployment realities, not just theoretical elegance. The same instinct applies here: a conference that cannot deploy its own participants fairly is going to have a harder time convincing people that its research priorities are aligned with the community's needs. We would ask the organizers directly: why have a "main" venue at all? If the goal is inclusion, spread the keynotes, the posters, and the social events evenly across all sites. If the goal is to maximize the Sydney experience, then say so plainly, and stop pretending that the satellite locations are equivalent.
The open question to watch is whether the community will accept this as a one-off planning failure or treat it as a moment to demand a different model. We are not looking for a revolution; we are looking for accountability. A simple, concrete step would be to publish the demand for each location alongside the allocation formula, so authors know their actual odds before they book anything. Until then, the message to anyone navigating this system is straightforward: treat the venue selection as a soft preference, build your presentation around the possibility of remote participation, and hold the organizers to their stated commitment to fairness. The future of AI research should not depend on which time zone you happened to be born in.