Paper review experiment

Editable parent selection

Does letting an agent choose which ancestor to improve produce a more accurate paper reviewer?

Preregistration · Study paper

Reviewer outcomes

Example outputs