In an interview around the announcement of the plan, we got this exchange: Tech interviewer: Do you, was there any sort of control for, sort of, selection bias

In an interview around the announcement of the plan, we got this exchange: Tech interviewer: Do you, was there any sort of control for, sort of, selection bias. Maybe that people less likely to graduate are more likely to select Senior House? For example, maybe like, students who ranked Senior House first but didn’t get in, how they did? Chancellor Barnhart: Yeah, oh, that’s actually a really good way to think about that. So, that’s something we’ll look at. This was after the initial decision to bar freshmen came through. In the analogy, I think it's a mistake to compare to spinning a server up and down - that's a low-consequence, low-cost decision. Maybe compare to decommissioning a physical server and buying a new one. I think before throwing away the high-latency server, it'd be worthwhile to make very sure that the problem wasn't about some process running on the server that would simply carry over to the new machine. I'm worried that MIT has decided to solve a problem by shuffling 'hardware' without any clear evidence on whether that's the layer with the problem. Every university has populations with higher and lower graduation rates, and we know that Senior House attracted a lot of groups who might have been at high risk (e.g. people with financial difficulties or substance abuse issues). If those people just end up in other dorms with no more support, this might become a costly change (financially and culturally) with no real benefits.