I've written a few times already about why I think (university) libraries are in a perfect position to take over scholarly communication from corporate publishers. Among other things I've mentioned that they already publish many of our theses, ancient texts and host journal article repositories (although it is of course clear that at the moment "repositories are failures – most have at best a few per cent of their output and academics either ignore them or regard them as a distraction or impediment. No-one uses them to discover scientific information because they are disorganised, have no useful search tools, and do not interoperate" this is easily fixable). Any such transition could be easily funded by the billions saved anually if subscriptions were cut. Given current publisher profit margins, the projected annual savings, once the transition is complete, would be about 4 billion (EUR/USD) every year - that's not exactly pocket change.

In two comments on the recent posts, both David Crotty and Neil Stewert pointed out that, in addition to the expertise I've already listed, (some) libraries already have all the tools, knowledge and training for proper academic publishing. The two examples the commenters mentioned (I'm now sure there are many more) were SAS Open Journals and Highwire Press. Thus, all it takes to transition from a system in which we outsource publishing to commercial publishers and lose approx. 40% of the money we pay to their shareholders to one where we get to keep these 40% to innovate and develop a modern scholarly communication system, is to cut subscriptions to free funds for disseminating  the know-how and the technology, as well as for extending the available infrastructure to make the different libraries interoperable according to a global standard (of course by hiring a bunch of smart experts, preferably from corporate publishers).

The more I learn, the more realistic it sounds to me that I actually might live to see a scholarly communication system that's not totally FUBAR.