The code was published under a public license, it has been available long enough to garner thousands of commits from over 192 "contributors" (albeit some of whom work for MetaMask), and it has over 1.5k forks. Also, there is no contract.

Merely viewing the license/repository does not constitute acceptance of said license (it's not even a clickwrap contract.) While what you can do with the code without accepting said license is quite limited, by publishing publicly to github in the first place they've presumably agreed to Github's ToS, including section D5 ( https://docs.github.com/en/github/site-policy/github-terms-o... ), which at least grants github users the right to view and fork the repository.

Which, presumably, grants users the right to view the repository in it's older states - where I'd argue the old license still clearly applies to the old code. That said, they could always flatten their git history tomorrow or yank the repository outright, which would leave you relying on pre-existing forks.

I'm also less sure about what you can do with old MIT forks of the software if you implicitly accept the new license terms by, say, using the new software. Does "The Program" refer to that exact version? All versions going forward? All versions past future present and in alternate histories?

I don't see why. If some code is made available under an MIT license and then the copyright owner switches to a new license, nothing stops others from continuing to use the original code under the original license.

Of course to do this you would have to know the code exists, but you could learn about its existence after the license change from someone other than the copyright owner.