There is a quiet anxiety that runs through every conference submission, and it rarely has anything to do with the science. It is the fear of a desk rejection over a formatting rule that feels arbitrary, even when the instructions are printed clearly. The question posted about ICLR 2027 table fonts captures this perfectly. A researcher has used `\footnotesize` for main-text tables and `\resizebox` for two wide ones, all while following the official LaTeX style. They have not touched margins, page dimensions, or global font settings. Yet they are worried, and honestly, they should be.
The official instructions say not to change font sizes except perhaps in the references. But here is where the confusion begins: the style file appears to define `\small` and `\footnotesize` at the same size. So the researcher is not actually shrinking anything. They are using the intended commands, and the result is a table that fits. The `\resizebox` is a different matter. That command scales the table to the line width, which can distort the font in ways that are visible to a careful reviewer. The question is not whether it is legible, but whether it violates the spirit of the formatting guidelines. We have seen this tension play out in other contexts, such as ICLR Submissions Exposed: Addressing Data Privacy Concerns in AI Research, where the pressure to comply with submission standards collides with the reality of preparing complex work under tight deadlines.
The practical answer, from our standpoint, is that `\footnotesize` inside a table is acceptable. The style file defines it, the instruction not to change font sizes refers to global changes, and the table remains readable. The `\resizebox` is riskier, not because it is forbidden outright, but because it can produce inconsistent scaling across tables, which draws attention. A desk rejection is unlikely for this alone, but why invite the scrutiny? The real issue here is that the researcher is asking for permission to follow the instructions literally, and the instructions are ambiguous. That ambiguity is the problem. We would tell them to test the output, print it, and look at it. If the table looks like it belongs, it probably does.
What this points to is a deeper truth about how we evaluate research. The formatting should never be the barrier to good ideas, but it is often the first filter. For readers who are preparing their own submissions, the takeaway is straightforward: use the commands the style file provides, avoid `\resizebox` unless there is no other option, and always check the final PDF. If you are unsure, ask the program chairs directly. That is what they are there for. The fact that a researcher has to ask this question on a forum, rather than finding a clear answer in the official documentation, says more about the state of conference formatting than it does about their preparation. We would add one more piece of advice: do not let this consume your energy. The work matters more than the font size. But if you want a concrete rule to follow, here it is: `\footnotesize` is fine, `\resizebox` is a gamble, and the person who checks your paper has likely done the same thing themselves.