A senior philosopher at another top program (not Michigan) e-mailed an interesting question about attrition; she writes:
[Y]our most recent post raises a question that might be worth discussing, namely whether all grad school attrition is a bad thing, and how to tell good from bad if not. I think US News etc. have done a huge disservice to academe by suggesting that the higher the completion rate the better at the undergrad level, and I wouldn't like to see the same assumption become unquestioned at the graduate level. Some people are better off doing something else, especially when there are not going to be enough jobs for all qualified candidates, and it's often in everyone's interest for them to find that out before completion -- though the how and when is always tricky, and most departments don't handle it well. And of course from an applicant's point of view a high attrition rate is always reasonably taken as a bad sign.
I am inclined to agree with my correspondent. Readers?