Position: Not All Explanations for Deep Learning Phenomena Are Equally Valuable

Alan Jeffares, Mihaela van der Schaar

International Conference on Machine Learning 2025 · Oral

In this thought-provoking position paper presented at ICML 2025, Alan Jeffares, a PhD student with over three and a half years of experience researching deep learning phenomena such as **double descent** and **grokking**, challenges the prevailing research paradigm within the deep learning community. Co-authored with Mihaela van der Schaar, the paper, titled "Not All Explanations for Deep Learning Phenomena Are Equally Valuable," seeks to formalize Jeffares' growing confusion regarding the community's approach to these topics. He questions whether researchers are uncovering fundamental insights or merely solving "artificial puzzles" that hold limited practical relevance.

AI review

A competent and honest position paper that names a real dysfunction in the deep learning phenomena literature — the tendency to accumulate ad-hoc explanations for synthetic edge cases without connecting them to generalizable theory. The central distinction between broad explanatory theories and narrow ad-hoc hypotheses is useful, and the recommendations (pre-registration, negative results, standardized benchmarks) are reasonable. However, the contribution stays at the level of diagnosis rather than cure: the paper identifies the problem clearly but does not provide a formal framework for…