Chapter Four Reads the Failures Science Leaves Behind

Field notes

The project treats doctoral dissertations as a record of negative results, extracting failed methods, rejected designs, baseline defeats, and unfinished work that may not appear in published papers. Its sample claim compares active-learning experiments across four institutions, while the homepage reports 22,537 dissertations read end to end from a larger 69,718-document corpus. The service offers cited memos within a stated 48-hour window and says each memo includes findings it could not locate.