Speaking as a semi-academic at a UK university, I do see why sheer inertia keeps the current business model of scientific publishers working, but it's such a rip-off that it must change in the long run. Here's the procedure by which a scientific article gets into a university library.
1) A scientist or group of scientists do some research. They are employed by a university, possibly with additional money from government funding bodies (EPSRC/BBSRC, etc. in UK, NSF/DoE/DoD,... in the US). They write up their results in an article.
2) They submit this article to a scientific journal. The main criterion in choosing a journal is how widely it is read in the community of researchers who might be interested. Actually, they submit it to a scientific editor. This is an academic who does this job because it adds to his standing in the academic community, not because he might be paid for it.
3) This editor, after sifting out any totally hopeless submissions, sends the article to 2-4 other academics in the same field for peer review.
4) The reviewers read the article and write a report, recommending to reject it, accept it, or demand changes. There is little advantage for them in doing this (for one thing, they remain anonymous), but it's part of what's expected from an academic. Abuses do happen at this point, but in my experience they are rare.
5) Based on these reports, the scientific editor accepts or rejects the article. If accepted, the (non-scientific, even though they usually have a degree in the field) editors of the journal create a publication-ready layout. This is the first step in the process that's paid by the publisher. Until a few years ago, this involved combining separately submitted text, figures and tables in a readable form. Nowadays journals require submission in the final publication format, so it's barely any work.
6) This publication-ready "proof" is sent back to the authors, who check for final typos and misspellings. These are sent back to the journal. For free, of course.
7) The article appears online. Web hosting needs to be paid by the publisher.
8) The paper appears in the next issue of the journal, which is sent out to subscribers (university libraries and very flush companies). This is a waste of dead trees. I'm doing literature search on a daily basis, but haven't looked at a bound copy of a scientific journal in years. But we have access to the online archive because we pay for the paper subscription...
- What the publisher pays for: Cursory editing. Web hosting. Paper, printing and postage.
- What the publisher charges: tens of thousands of pounds/dollars per year and institution. Having a de-facto monopoly on an important journal, they bundle it with less popular publications to hide the price.
The alternative: Open access journals. Drawback: It doesn't work.
The model: the author pays for being published. It's a few thousand pounds per article, which an academic without a grant just doesn't have. If you do apply for a grant, there is provision for it, but at the same time you are supposed to keep the cost down. And if you do get the money, you are tempted to spend it on something more useful and publish with Elsevier.
The solution:Beats me. Either a wholesale switch to open access (with changes in funding to allow/enforce this), or a slow drift to slightly more reasonable publishers.