Crisis in scholarly publishing?

Michael Eisen writes:

Universities should have stopped paying for subscriptions, forcing publishers to adopt alternative economic models. And they should have started to reshape the criteria for hiring, promotion and tenure, so that current and aspiring faculty did not feel compelled to publish in journals that were bankrupting the system. But they did neither, choosing instead to let the problem fester. And even as cries from the library community intensify, our universities continue to shovel billions of dollars a year to publishers while they repeatedly fail to take the simple steps that could fix the problem overnight.

The source of the problem:

Virtually all of the problems in scholarly publishing stem from the simple act, repeated millions of times a year, of a scholar signing over copyright in their work to the journal in which their work is to appear. When they do this they hand publishers a weapon that enables them to extract almost unlimited amounts of money from libraries at the same research institutions that produced the work in the first place.

Sounds familiar?

Expenditures on scholarly journals at American research libraries quadrupled from 1986 to 2005, increasing at over three times the rate of inflation. This despite a massive reduction in costs due to a major shift towards electronic dissemination. These rates of growth continue nearly unabated, even in a terrible economy.

The main point:

The biggest obstacle to the rise of open access journals was (and to a large extent still is) the major role that journal titles play in how universities evaluate candidates for jobs and promotions. In most academic disciplines, careers are built by publishing papers in prestigious journals – those that are the most selective, and therefore have the most cache. Scholars rising through the ranks of graduate school, the job market, assistant professorships and tenure face a nearly contant barrage of messages telling them that they have to publish in the best journals if they want to succeed at the next step. Never mind that it is far less true than people believe. That people believe it is all that matters.

See more here. There are many good ideas in Eisen's post.  

There is a path dependence here. Most of the work is done by university professors. They do the research, the writing, the editing, the peer-reviwing. But they seem to be unable to do the publishing - they can't solve the collective action "problem" of publishing. Would it be better if universities become the main publishers of journals again, as it used to be in the past?