Making the assumption that Mayer slash Yahoo wouldn't implement a policy like this on a whim, I bet this comes out of a bigger plan. If one had a mandate to do X, studied how to do that, and found X would be more likely to succeed with a team working purely onsite, then you just do a cost benefit analysis and make a decision. You'll find out later if you're decision was a good one, so hopefully on the front end you've got good data, judgement, and moxie to effectively make bold decisions!