@ Contract is broken
Well, that would be possible, of course. Sort of like arXiv or the like, sure (there's nothing like that, significant, in my area: biological sciences). But libraries and universities don't usually do typesetting, proofing, editing, peer reviewing (which, for good or evil, is going to continue being the system in publishing), distributing (how are the guys in China going to read your paper, or you theirs? At least a website has to be set up and managed), etc. My boss is one who always keeps saying that we should get away from journals and just put research results on web pages. Too big of a cultural shift, but might happen with time.
And, I forgot to mention in my first post, even if the public institutions do not subscribe for the content, you will still pay twice for it if you need it: first with your taxes that went to fund the work and pay for publication fees, and then when you pay again to privately access the content.