Whenever these cases come up, we tend to look at it with a status-quo bias - assuming that of course this data exists, and it's a question of whether it should be "free" (whatever your interpretation of that term). But in reality, things like YBB+ can impact the existence of the data in the first place - in this particular case, the evaluations existing in as comprehensive a form as they do depends on Yale's explicit sponsorship (having professors hand out the eval forms, etc.)
I'm not sure what the policy implication is - there's no easy rule here. But it's worthwhile to keep in mind that "information wants to be free, Yale can't decide how it's used" (an attitude that's a bit of a straw-man, but not too far from what some have argued here) ignores the reality that the next time the Yale faculty debate whether or not to collect and collate evaluations (or whether to have a numeric score component), they may well decide not to. (Which, of course, is a principle that applies to other cases of data collection - sometimes it's best not to collect it, just because you don't know how it'll be [mis-]used...)