The novelty critique is one of the most exhausting and revealing gatekeeping mechanisms in computer vision research. A researcher on Reddit asks how to frame contributions when thousands of papers are published annually, wondering how much genuine novelty is left to explore. Our take is blunt: the obsession with novelty is often a crutch for reviewers and a distraction for researchers, and it is time to reframe what we value.
This researcher's frustration is shared by many, and it reflects a systemic tension that we see play out across the field. When compute is limited, choosing between stronger results and a strong submission becomes a painful trade-off, as documented in our related piece When compute is limited, choosing between stronger results and a strong submission. Researchers often optimize for reviewer expectations rather than scientific depth. The novelty critique is the same problem in a different disguise: it pressures authors to manufacture a "first" rather than build something solid and useful. Meanwhile, the broader community celebrates breakthroughs like those highlighted in From Code to Conference: One Developer's AI Research Earns a Spot at NeurIPS, where individual developers find paths to top venues through genuine contributions. The contrast is instructive: novelty should not be a barrier to entry, but a measure of clarity in communication.
So how should this researcher, and anyone facing similar reviews, respond? First, stop chasing the myth of absolute novelty. In a field publishing tens of thousands of papers per year, the likelihood of a truly unprecedented idea is vanishingly small. What matters is not whether your method is entirely new, but whether your contribution is clearly articulated and meaningfully contextualized. Reviewers look for novelty, but they also look for rigor, reproducibility, and practical impact. Frame your work by explicitly stating what existing approaches miss, what your method adds, and why that addition matters for real-world use cases. Distinguish incremental progress from insufficient novelty by showing that your incremental step enables something previously impossible or impractical, not just a marginal gain on a benchmark.
The practical takeaway is direct: stop letting reviewers define what counts as novel. Define it yourself, in your own terms, with specific evidence. If your contribution is a clever combination of known techniques, say so and explain why the combination is non-obvious and effective. If it is a small improvement in efficiency, show how that efficiency unlocks a new scale of application. The Reddit researcher is asking for framing advice, and the best advice is to treat novelty as a communication problem, not a discovery problem. Write your paper to answer the question: "What would someone building on this work be able to do that they could not do before?" That is the novelty that matters, and it is available to every careful researcher.
One specific detail to watch: the researcher mentions that this critique appears repeatedly across NeurIPS, ICLR, and CVPR. That pattern suggests a systemic failure in reviewer training, not a flaw in the research. Until conferences address that, the burden falls on authors to be clearer, more precise, and more confident in their framing. The future of the field depends on valuing progress over perfection, and that starts with how we write about our own work.