If a study registration exists, but nobody ever reads it, was there a point in it being registered in the first place?

Delivering on The Promise of Study Registration

Across medicine, psychology, economics, and other fields, researchers agree — in the abstract — that study registration is often beneficial. Write down what you plan to do before you do it, then readers can tell what was planned in advance versus what was decided after seeing the data. That transparency is supposed to reduce bias and help us calibrate our confidence in results. In practice, something else happens: study registrations are often left unread and are rarely compared to the final published work (not even by the authors themselves).

Why is this? Comparing a registration to a finished paper is slow, difficult, and boring. Peer reviewers are already overburdened; adding "now cross-check the registration line-by-line" is a hard sell, as knowing what to compare, finding the relevant text in each document, and synthesizing discrepancies between documents are time-consuming tasks.

A sufficiently motivated reader might take on the endeavor, but this nonetheless comes at an opportunity cost (e.g., that time could instead be used for activities more rewarded by current academic incentives, such as working on one’s own research). In other words: study registrations make discrepancies detectable in principle, but discrepancies are often not detected in practice.