Which is a reputational hit on departments against their will.

It really is. I have a recent Nature paper where, as the result of the email problem, a reader could easily think I'm actually more connected to a different university. I expect the administration, and publicity people, are not happy about it. But IT won't change their policies, and I'm not going to put an ephemeral email address on a permanent publication.