Editors of $prestigous_journal are the most prestigious professors in their field themselves, no?
So let's say $prestigious_journal is one of the journals that the university students/faculty get access to from Elsevier. University pays Elsevier, university gets access. Fine.
Now, suppose the editors of $prestigious_journal care about open access. But given the current arrangement with Elsevier they cannot become an open access journal tomorrow.
Is that the problem?
If so, the obvious next chess move I see is this:
Editors of $prestigious_journal resign from $prestigious_journal today. Tomorrow, they start $prestigious_journal1 (modulo some new name-choosing scheme) and post a website or whatever.
Granted I'm biased-- is the problem that the editors don't know how to publish content on a website?
If so, then the obvious chess move is to use whatever turnkey open source journal publishing system exists.
What is it, and why hasn't every prestigious journal already changed its name and started publishing there?
The argument cannot be, "Well, no one would know where to find the new journal." They are the most prestigious editors-- their field will know immediately where they went. Everyone else will follow from there.
The only sticking point I can see is that $prestigious_editors can't take their previously published work with them to redistribute over their new website. But they still get career credit for their previous publications, and they/students/faculty can still access that through Elsevier. And all their new publications can be accessed and redistributed under a liberal license by all, for a publishing fee approaching zero dollars.
We have a rich history of big software projects making much more disruptive changes in a single day. E.g., Mambo became Joomla. The fact that prestigious journal editors haven't done likewise in droves leads me to believe they are saying one thing about open access in public and saying something much more important by doing very little to change the status quo.