Two years of research in continual motion generation, rejected without explanation. That is not an edge case. It is a signal that the peer-review process, already strained, has started to fail in ways that directly harm researchers who are trying to do something genuinely new. When a venue like TMLR desk-rejects work without offering a reason, it erodes trust in the very system that is supposed to validate progress. Our take is blunt: this is unacceptable, and the community should demand better.
The researcher in this story is not alone. Every practitioner who has pushed boundaries in A practical guide to diffusion models that balances deep math with clear insight knows that novelty often comes with friction. But friction is one thing; silence is another. A desk rejection with no rationale tells the author nothing about whether their approach was flawed, their framing unclear, or their contribution simply misaligned with the reviewer pool. Worse, it wastes the very resource that research depends on: time. Two years of motion generation work, designing architectures, running ablations, iterating on failure modes, cannot be reduced to a one-click rejection. The lack of transparency sends a message that the process values speed over fairness.
What makes this particularly frustrating is that the field of continual motion generation is exactly the kind of technically demanding area where clear, accessible guidance is most needed. Compare this to the conversation around How small AI models closed the gap on a test built for humans, where the community rallied around reproducible benchmarks and open discussion. That is how progress should work: through dialogue, not through a closed door. The current system incentivizes safe, incremental work over risky, long-term projects. If you spend two years on a hard problem and get a form rejection with no feedback, the rational response is to stop trying hard problems. That is a catastrophic loss for the field.
We are not arguing that every paper deserves acceptance. But every paper deserves a reason. The specific consequence here is that researchers working on motion generation, a domain with real applications in robotics, animation, and simulation, will now think twice before submitting to TMLR. They will hedge their bets, fragment their contributions into smaller pieces, and avoid the kind of sustained effort that produces breakthroughs. The open question is whether the community will push for a standard where desk rejections include a brief, specific justification. Without that, the review process becomes a black box that punishes ambition. And that is a future none of us should accept.