Peer review reduces the variance in quality of published research. Since the value of research is right-skewed, it therefore follows that peer review will simultaneously raise E[value of any given paper | do(PR)] and greatly reduce E[value of all published research | do(PR)].