The question of how much work in progress a workshop submission can be, well, it's one that almost every researcher has faced at some point. The person asking this is sitting on a partial implementation, a set of initial results, and a clear mathematical plan for the full version. They are unsure if that's enough. The short answer is yes, but with a catch. Workshops exist precisely for this kind of intermediate stage. They are not the same as a full conference or journal submission, where the expectation is a complete, self-contained contribution. A workshop is where you test ideas, get feedback, and align with a community that is thinking about similar problems. The key is to be transparent about what you have and what you are planning to do.
That said, the real value here is not the partial algorithm itself, it's the clarity of the roadmap. The person has already identified that the basic version solving problem A is not novel on its own. That is a strong signal of self-awareness. What makes the submission compelling is the gap between the mini version and the full one, and the mathematical reasoning that bridges that gap. That is the story. A workshop reviewer would want to see that you understand why the full algorithm works, even if you haven't implemented it yet. They want to see that the plan is credible, that the math holds up, and that the initial experiments, even if limited, point in the right direction. This is similar to how Exploring Real-World Computer Vision: Deployments, Edge Models, and Current Challenges highlights the gap between theoretical model performance and the practical constraints of deployment. In that case, the value is in the engineering, not just the architecture. Here, the value is in the proof of concept, not just the final result.
What we would tell this researcher is to lean into the plan. Frame the submission around the full algorithm, not the mini version. Use the initial results as evidence that the core principle is sound, and then spend the majority of the paper explaining the mathematics behind the full solution. Show that you have thought through the challenges of scaling from A to A and B. Address potential pitfalls. This is not a weakness; it is the nature of research. Many published papers are incremental steps toward a larger goal. The Explore the Forrester Function: Beyond Mathematics, a Tool for Machine Learning article is a good example of how a single mathematical concept can be positioned as a practical tool, even when the full implications are not yet fully realized. The Forrester function is not new, but the framing of it as a machine learning tool gives it relevance. Similarly, your full algorithm is the contribution, not the mini version.
The practical takeaway is this: submit the work in progress, but be explicit about what is done and what is not. Do not oversell the basic results. Instead, sell the potential of the full approach. A workshop is the right place to get feedback on whether the math is convincing and whether the problem is worth solving. And if the reviewers say the plan is not enough, that feedback is just as valuable, because it tells you where to focus your next round of work. The question is not whether the submission is "enough," but whether the idea is clear enough to be evaluated. That is the bar. We would also point out that ICLR Submissions Exposed: Addressing Data Privacy Concerns in AI Research shows how quickly the research community reacts to perceived flaws in the submission process itself. That is a reminder that clarity and honesty in your submission are not just good practice, they are essential to maintaining trust. So, be straightforward about the stage of your work. The community will appreciate it, and you will get the feedback you need to move forward.