This conflict goes back a long time. In the early 1990's, with online journals just getting started, MIT was able to insist on principles like "anyone physically in the library has full access, even if they are not otherwise affiliated with MIT". A couple of years later, power shifted, and Elsevier could "our terms, take it or leave it". Then three decades, a human generation, of Elsevier rent-seeking, and so many people working towards Open Access, unbundling, google scholar, arXiv, sci-hub, and so on. Societal change can be so very slow, nonmonotonic, and profoundly discouraging. And yet here we are, making progress.
For anyone unfamiliar with "author [...] required to relinquish copyright [...] generous reuse rights", journals would require authors to completely sign over copyright, so authors' subsequent other-than-fair-use usage of fragments would be a copyright violation. Rarely enforced, but legally you'd have to obtain permission. While some institutions sign contracts easily, and struggle with the fallout later, MIT legal culture has been pain-up-front careful with contracts. Which is sometimes painful. But IIRC, we're today using X Windows instead of CMU's Andrew, because MIT could say "sure!" while CMU was "sure, err,... we'll get back to you... some mess to clean up first".