Almost all publishers and journals have adopted a "hybrid" model, where you can publish in a paywall journal but pay a fee to have your particular article made available OA. I think if you want to publish in Cell and get your article OA all you have to do is pay $5k.

At Sage our OA fees for hybrid journals range a lot, but are typically either $1,500 or $3,000 (source: http://www.uk.sagepub.com/repository/binaries/pdf/SAGE-Choic...). The more prestigious the journal the more the fee is typically. Also, high fees are common in hard sciences (where everything is grant funded, like from Gates, NIH, etc) but the fees are typically much lower in social sciences and humanities where there aren't typically publication fees built into research grants. I don't think we (Sage) publish anything that can command the $5k like Cell and Nature, which is basically the price point where OA fees top out.

So the authors will still want the prestige of the top-tier journals (at least while hiring decisions are still made with that metric), and one byproduct of mandates like these is more APCs flowing to the big publishers.

That's what I thought at first, but the language from the original post seemed ambiguous—that it could only be published in Open Access Journals, and I also couldn't find OA costs for Cell or Science. I did find this page from Elsevier, though, and it looks like it ranges from $500 to $5,000: http://www.elsevier.com/about/open-access/open-access-option...

Thanks for the first-hand knowledge about Sage. Presumably, though, at a one-time price of $1,500–$5,000, publishers e.g. Elsevier weren't making much from keeping them closed-access in the first place, were they?

Not sure if the trend of more APCs going from grant funders to publishers is a good one, but probably better that the grant funders pay than those who can't afford access.

Not sure what you mean by "weren't making much from keeping them closed-access". Paywall publishers earn the bulk of their revenue from charging academic libraries subscriptions to the journal content. This subscription revenue can only exist if the content is not freely available to all. So there's definitely a big pile of cash that keeping articles closed access brings in. The worry about having OA articles isn't that you might miss out on the $30 pay per view charge that you or me run into when we hit the paywall, that's almost inconsequential from a revenue standpoint. It's the worry that they won't be able to charge a library a yearly subscription fee.

So, it's a ripoff? These numbers sound completely made-up.

Well, there's some price point at which if everyone paid the OA fee (effectively making your entire journal open access) then the revenue would about equal the revenue you get from traditional paywall subscriptions for that journal, and the publisher wouldn't care one way or the other.

In reality it's almost impossible these days to figure out what the paywall revenue from a single journal title is anymore, since everything is bundled in huge packages. But yeah, more or less the fees are based on what the publishers think authors will pay based on the prestige offered by the journal. There are people trying to change the typical price point of OA publishing (PeerJ's $99/author fee comes to mind), but there's A LOT of intertia in academia.

Not in the medium term. The paywalled articles' days are counted already -- in physics, putting the paper into arxiv is the standard practice, and it doesn't hurt anybody.

It might just mean that they publish in the same places and just check a box that says "our funding agency mandates open access" when filling out the copyright license/transfer paperwork.