abt

01 : Posts

  1. Pre-registration changed how I read my own results

    Notes on the small ritual that made our BlackboxNLP submission more defensible — including the one pre-registered hypothesis we cleanly failed, and why that failure ended up being the most interesting result in the paper.

Want to argue with one of these? — atonmoy27@wabash.edu. RSS feed coming whenever I get around to it.