Leopold Stotch at OTB is indignant over Princeton’s new policy of capping the number of A’s at 35% of the class. I’m a little new to academia, but I’m not sure the policy is as objectionable as he seems to think.
Administrators are limited in their ability to judge the performance of professors. The problem—and the reason many administrators will default to grade distributions as a measure of grade inflation—is that educational outputs aren’t easily observable, whereas grades are easily observable. To make educational outputs observable, administrators have to overcome a great deal of uncertainty and cost. The attempts include taking student surveys—which, not surprisingly, are strongly correlated (positively) with grades—and observing professors in class. This last item is quite expensive and may just result in the professor being on his best behavior when the auditors are present.
One of the reasons that research universities use publication to make tenure decisions is that publication is easily observable, as is the quality of the publication (an ‘A’ journal, ‘B’ journal, and so forth). In any case, this is a topic that will be with us a while, if not forever.