Many agencies do that, e.g. ERC requires it. However it hasn't changed anything, because these publishers charge (a lot) for open access, and ERC just ends up subsidizing them even more.

Then perhaps those agencies could create their own journals which are open access and then publish everything through those? That'd be an easy way, without having to pay for the open-access?

The problem with open access is that typically you, as an author, get a choice:

- assign copyright to the publisher, have your work paywalled and pay for the access to it via exorbitant subscription fees the university pays

- ask for having it available as open access - and then the publisher charges you thousands of USD/Euro ($4000 is common) for the privilege as a condition of having it published.

Very few university departments can afford to pay such fees so you can guess where the papers end up.

And if you mean open access journals that don't demand these fees - well, that's of little use to an average researcher when their career depends on peer reviewed publications in reputable journals. Publishing it wherever, in a journal nobody heard about with who knows who doing the reviews has zero value and you won't get any credit for that.

That's definitely part of the problem, and would inhibit laws since the publishers know they'd get the free money. Perhaps the government could also regulate the publishers and say they can't charge any more for open access than they can for subscription fees?

Or, perhaps the different agencies can create their own journals for their funded projects and let the research appear in them? Basically, create new reputable journals precisely for open access? Or make it where a paper has to be submitted to a "prestigious" one and an open access one a year or so later (i.e. author keeps copyright, period), which would hopefully build up the prestige of them?

I'm not sure the solution, just trying to think of ideas.

Perhaps we can craft a section to include in research papers citing research showing peer review is currently insufficient for driving science and is in fact contributing to reproducibility crises? How research gets published and the pressures impacting any singular study seems like important metadata to include in reports.

almost all journals let you upload your preprint in arxiv or biorxiv etc.