> You can publish unedited, dubiously archived, difficult-to-cite, poorly advertised articles without anonymous peer review for the cost of hosting, right now.
Yes I'm not saying arXiv has no costs, I'm saying mechanical_fish pretty much described arXiv. As you pointed out, arXiv still does not run on unicorn farts and rainbow tears (although a budget of $420k is a tad lower than PLoS's 2010 budget of $12m)
There are a couple thousand accredited universities in the US alone. Thousands of philanthropists who donate to academic causes. Hundreds of corporations looking for good PR.
If the academic community were to rally around open access, and then fail to find (even a lot more than) $500k/year, I'd have to say good riddance. That would take a very special kind of incompetence.
The problem with that grant-based model is that you're making the publisher dependent of that precise source of income (rather than of its own activity), and the publisher will have to gear its fund-raisin activity towards getting more grants and donations rather than getting publications.
I'd say the latter is far more desirable than the former.
"getting publications"? What does that even mean? You think academia has to be convinced to pump out papers?
If the standard is an open-access site similar to arXiv, that's where the papers would go. No effort needs to be spent there, and there would be plenty of motivation for people and institutions to meet its modest funding requirements without a massive drive.
The primary costs are support/technical staff and equipment, that would not change.
That would be arXiv, would it not?