The problem isn't that you're somehow branded a 'rebel' if you refuse to publish with Elsevier journals. It's that, when you're applying somewhere, they'll rank you by SUM(paper_i * impact_factor(journal_i)). That score will suffer, and nobody will care about your reasons.

You're also underestimating the independence of researches at a university. They're not employees in the usual sense. In fact, the best method to stop them from publishing at some journal Y would be for the administration to tell them to publish in Y.

Also, in some counties (the UK is a prime example) the rules governing "impact" is largely dictated and/or influenced by central government. So not only do you need to change just your local institution but a whole country's institutions at a go; a multi-year political fiefdom protecting shambles at best.

That "impact" scoring system is utterly useless. I hope it doesn't survive the next non-Tory government.

See [https://news.ycombinator.com/item?id=15007958]. Their impact factor is now 2.450 vs 1.848. Of course I am sure it's way harder to do this for far more conservative and larger industries. At the same time it may be just the case of starting small.

that easy? "You should publish only in Elsevier journals!"

If done collectively that has worked before. In 2001, almost the entire editorial board of the Springer journal Machine Learning resigned [1] in order to lend their support to a new open-access journal, Journal of Machine Learning Research, which quickly supplanted the former as the top journal in the field. It helped that this included a lot of senior people in the field (Stuart Russell, Geoffrey Hinton, Leslie Kaelbling, etc.) who would be the ones judging ML hiring and tenure cases at many universities.

[1] Resignation letter: http://www.jmlr.org/statement.html

That only works if the editorial board "flips", though - unfortunately, universities mostly aren't able to start journals with reputable editorial boards.

I'm currently leaning towards finding a way to convince large numbers of boards to flip as the most viable path to proper open access.

Those academics would still have less-prestigious publications, which is a net-loss for both them and their university.

A solution might be dual publishing, but I believe some of the bigger journals require some form of exclusivity.

You mean a new journal for every scientific field? That's helluva lot of journals. Then, you need to attract people to send the papers there. There are countless local journals that mainly attract local authors now, and those are a joke when they are not outright junk. To publish even a half decent journal, you need to attract researchers from all corners of the world to send you their best work. How?

Isn't there already a journal for every scientific field, or are you speaking of only a part of the industries?

Many papers have multiple authors from different universities/institutions. It would be difficult if just one was boycotting the top journals in the field.