Subsidise, subsidise - but don't let the real issues evade your eyes (free after Tom Lehrer).
Recently, early November 2006, a quick thread was spun on the AMSCI open access list about subsidising journals (though the title of the thread is 'What Can and Should Be Mandated').
I've looked in Wikipedia for a definition of subsidy: "...generally a monetary grant given by a government to lower the price faced by producers or consumers of a good, generally because it is considered to be in the public interest". Would this also cover situations in which governments are the ones that provide the funds for consumers - because it is in the public interest - to buy the goods or services in the first place? I think it does, and that this provision of funds in the first place is a form of subsidy.
Ergo, all journals are subsidised. Although, a lot of microsubsidies could perhaps be seen as constituting a 'market'. The money for journals comes, to a very large degree, from governments. If not directly, then indirectly, via the circuitous route of research and learning institutions with their librarian gatekeepers and content collectors. The subsidy is just misdirected. And ill-suited to what the intention is: serving the public interest.
Why is that so? Subsidising the consumer, in order to be able to buy scholarly journals, is accepting the premise that a journal's value is primarily in its content. It isn't. While that perhaps used to be so, it isn't any longer as a result of the internet. The content can be found elsewhere, and for free. In a preprint repository, for instance, or even in a postprint repository. The internet has made one of the classical functions of publishing - dissemination - exceedingly easy to do by anybody else as well, and particularly by the author. Journals are no longer needed for dissemination per se.
The function of a journal was always much wider, even when all its economic value was bound up with just dissemination. Journals organise the formal acceptance and embedding of the literature in the record. They keep the 'minutes of science' as I mentioned in an earlier post.
This is a very important function of journals, as I argued in the same earlier post. And that's the function that needs to be 'subsidised'. If the same money that's now sloshing around in the L-sphere (licence sphere) were used to enable peer-reviewed articles to be added to the free and open Noösphere (knowledge sphere) by paying for the service of formal publishing rather than for access, we would have so much more 'bang for the buck': open access.
Here's an illustration why formal publication is important: for the UK Research Assessment Exercise (RAE), the Higher Education Funding Council for England (HEFCE) has negotiated a licence with the Publishers' Licensing Society (PLS) to gain access to the formally published literature (only the "authoritative final version" will do; so they're not after the content, but after its label of authoritativeness, established by being published in formal, peer-reviewed journals), "only for the purpose of conducting the RAE." The licence is free of charge. Somewhat perverse, in my view: the system pretends to pay for access to the content per se, yet wants to have access to the real value of formally published literature for free. If HEFCE supported open access at source - the article processing charge model - then it would have free, open access to the material without any need for a licence and so would everyone else.