On a large scale, initiatives like PLOS One (http://www.plosone.org/) are a great example of this. On a smaller scale, journals like Sociological Science (http://www.sociologicalscience.com/) & (http://www.gsb.stanford.edu/stanford-gsb-experience/news-his...) are also being successful managing the entire toolchain themselves.
While things seem dire now, I'm confident they'll get better. Scholars with status are increasingly throwing their weight behind Open Access initiatives. Tim Gowers, that droque mentioned in this thread, is evidence of this.
Disclaimer: I co-founded a startup in the space (http://www.scholasticahq.com) and Sociological Science uses our platform for managing their peer review process.
I too am confident that things will continue to improve.
Not in the dictionary, Urban Dictionary, Google just tries to translate it from French. What is a droque?
Regardless of how difficult it would be get academics to change how they do things, it's also a problem that we can't agree on an acceptable and sustainable Open Access model.
The 'green' option sets an embargo (usually 6 months), during which time universities (etc.) must pay to view the article. After that, however, the journal article becomes freely available.
The 'gold' option asks academics to pay a small fee (£couple of hundred) to have their article published--what's known as Article Processing Charges.
The problem with the 'green' option is that in disciplines like science, medicine and technology, the first six months after publication probably encapsulate 90% of the article's value--after which point it has been replaced by something else--which means people would probably continue to pay for these things anyway.
The problem with 'gold' is that (a) you start publishing stuff based on who has the ability to pay, rather than academic merit, and (b) it would make academic publishing the only industry in the world where the supplier is paying the purchaser/buyer.
 A joke about Oxford Uni goes like this:
Q. How many Oxford academics does it take to change a lightbulb?
I can't figure out how it would be possible that shifting from physical paper to online hosting could be more expensive
Developing and hosting something like JSTOR in the early to mid 90's was no where near as cheap and easy as it might be today (and let's face it even today it isn't completely trivial). Doubly so if you had no staff with the relevant skills and had to do everything via consultants.
Honestly, I don't know the whole story. I've only been around here for a few years.
Especially if you have no domain knowledge in running a paywalled website, you could end up getting charged an insane amount of money by some consultants to set it up for you.
Hopefully some reputable open journals will distinguish themselves from the predatory journals that take advantage of unsuspecting junior researchers (or worse, that take advantage of unsuspecting people who don't know better).
Announce a policy: "We will be reducing our purchase of proprietary journals by 10% per year over the next decade. We will be shifting holdings to journals published under free and open terms."
If that message comes from enough premier libraries, the market will shift.
Yes, I could get an alumni card at my own alma mater, for a substantial fee, and check things out there. But then I'd have to take a plane across the country to have access since like many people I did not stay in the town where I went to school. Also, they also don't have parking. Oh there's free parking on Sunday. When the library is closed, so that doesn't help much.
It's also frustrating because most times that I bring this situation up, people will appear in comments and say I am wrong or lying and claim that public libraries all have full journal access. It's not true. I've asked. My public libraries, in the county I live in, and all adjacent counties, do not have journal access. I've also called all of the colleges and know that the state university is the nearest location where I can get access. Despite this there's usually people that will show up and claim that's all wrong and everyone has access. No, we don't.
Also a lot of the research tucked away was directly or indirectly supported with public funds. In some cases there is a requirement it be openly published, but not always. It seems unreasonable to have to support research and then not have open access to it.
Granted, not all fields utilize such a preprint server, but in some fields, the public's access to research results is not as dire as your comment states. There is, of course, still significant room for improvement, even in fields that use preprint servers.
That being said academia these days is full of shit anyhow. So quality of the publications reflects the quality of the institution.
This is entirely possible but it doesn't address the need for journal subscriptions to access pre-existing articles. As I understand it, unless copyright law changes, if a researcher wants to read online a 1970s paper from Nature, their library has to subscribe to Nature forever (approximately), even if everything they produce from today onwards will be uploaded to github.
It seems the only time I'm ever on Springer or Wiley is to read some conference proceedings from 1992.
There's far less staff needed to curate the journals.
And the millions saved in time with researchers getting them on line.
Libraries are much better off. They just could be even more so.