Empirical Research in Machine Learning Ended Math’s Monopoly

A theorem-first paper and an ablation-heavy systems paper can now describe the same model class and leave with very different outcomes, because modern ML did not become less rigorous - it changed how rigor earns legitimacy. The first may offer a clean result under narrow assumptions. The second may show weaker elegance but stronger evidence: better baselines, harder evals, robustness checks, and logs from deployment-like tests. In modern ML, that second paper is often the one that gets accepted, funded, implemented, and copied.