Many universities pay a yearly subscription to journals so that their researchers (usually identified by IP address) can transparently access scientific papers from these journals. However, more and more papers are made available online in open access nowadays, so some universities have been tempted to cancel some of their subscriptions.
While some universities took the plunge, others have been hesitant. I can understand this to some extent: the university doesn't want to cancel, and then discover that their researchers are up in arms because they cannot access some papers, and the university has to come back to the table of negotiations with the publisher and subscribe again while being in a bad bargaining position. (This is important, because the subscription prices are often heavily negotiated and opaque.)
I was thinking that a tempting solution for a university could be to do a kind of experiment: while they are subscribed, temporarily "disable" their subscription for some set duration, and see whether their researchers are having any problems or not. This could be technically achieved, e.g., by routing their Internet traffic via some IP that is not subscribed, or something of the sort. And one could imagine it could give valuable insight to the university, without requiring any renegotiation with the publishers.
Do we know of any universities that made experiments of this kind? and of what the outcome was?