preprint

Navigating Preprint Citations for Your Camera-Ready Paper

Getting a paper accepted is a win, but the preprint problem is real.

4 min readMachine Learning

The moment a paper moves from preprint to camera-ready, the relationship between your own work and the work it inspired becomes a delicate dance. One researcher recently asked a sharp question: how do you handle the related work section when later papers cite your preprint, extend your methodology, and now sit between your original submission and its final published form? It feels awkward to cite yourself. It feels worse to ignore the conversation. But the real challenge is that your novelty, once a single flame, now has company.

This is not just an academic etiquette puzzle. It is a practical problem with real consequences for how your work is perceived and how the field builds on it. The instinct to avoid self-citation is understandable, but it is also misguided. The camera-ready version is not a fresh artifact. It is the formalized version of a piece that has already entered the discourse. Pretending the subsequent works do not exist does not protect your novelty. It just makes your related work section look out of touch. The better move is to treat those citations as evidence of impact. When you write that others have built on your preprint, you are not undermining your contribution. You are showing that your contribution mattered. That is a form of authority no amount of clever phrasing can manufacture.

The deeper issue here is about how we think about provenance in research. Your preprint is not a draft to be forgotten. It is a public timestamp of your thinking. The camera-ready version is the place to reconcile that thinking with what happened next. You can frame the subsequent works honestly: acknowledge they extended the methodology, then clarify what your version adds that they did not cover. That is not defensive. That is precise. In a field where Clean Data Starts With Catching AI Slop Before It Skews Your Model shows how easily context gets lost when data sources are muddled, the same principle applies to citations. You are clearing the signal. And when you look at how Exploring Real-World Computer Vision: Deployments, Edge Models, and Current Challenges navigates the gap between research output and practical deployment, you see the same lesson: what matters is not whether you were first, but whether your framing helps others build correctly.

The practical takeaway is straightforward. Cite your preprint. Cite the works that cited it. Then use one or two sentences to draw the line between what you originally contributed and what the follow-ups changed. That is not self-undermining. That is scholarly maturity. If you are worried about novelty, remember that novelty in research is not a single event. It is a trajectory. And the camera-ready version is your chance to narrate that trajectory accurately. The alternative, ignoring the conversation, only creates the impression that you either do not know the literature or do not respect it. Neither serves you. So take the awkwardness as a sign you are doing something right: you are engaging with the messy, iterative reality of how knowledge actually moves. The question is not whether to cite, but how to frame it. Start there.

From Machine Learning

I had a paper accepted to a conference. This paper was originally published as a preprint. Subsequent works citing our preprint focused on the same topic and reused/extended our methodology. I am now preparing the camera-ready version of that preprint and I'm wondering how I should deal with this for the Related work section. It seems odd to me to cite my own preprint for the camera-ready version of the paper (and I am not even sure if this is allowed), but at the same time, I don't want to undermine the novelty of my original work (nor undermine the…

Read the original at Machine Learning