The reason peer review rewards caution

Two manuscripts arrive with the same referee. One is careful, incremental and certain to be cited eleven times. The other is odd, possibly wrong and possibly important. The referee has an hour, no payment and an asymmetric exposure, and the odd one goes back for major revisions.

What everyone sees

Conservatism in review is read as intellectual temperament — a matter of taste in what counts as sound work. Temperament varies between individuals; the tilt does not. It appears across fields, panels and career stages, which points at the structure of the task rather than at the disposition of the people performing it.

What is actually happening

Cole, Cole and Simon resubmitted funded and unfunded proposals to fresh reviewer panels and found agreement close to what chance would produce for a large share of decisions, indicating that outcomes depend heavily on which reviewers were drawn rather than on a stable property of the proposal. Nosek, Spies and Motyl set out the incentive structure around that noise: the costs of the two possible errors fall in different places. A wrongly rejected paper leaves no trace attributable to anyone, since it is either published elsewhere or never seen. A wrongly accepted paper that fails is a durable, searchable object attached to a journal and, informally, to the people who let it through. Facing noisy signals and asymmetric costs, the risk-minimising response is to demand more evidence.

Why it stays hidden

The asymmetry hides because rejection is framed as rigour, and rigour is a virtue rather than a position. Nothing in a referee report says that the risk of being associated with an error exceeded the risk of missing something. The reasons given are always about the work, and they are usually accurate reasons that would also have applied to accepted papers.