Paul Krugman had an excellent blog post (here) the other day exploring the nature of academic exchange in the internet age, which is both pithy and interesting. In it, he describes the deplorable delays and increasing irrelevance of publishing in academic journals. Already 30 years ago, Krugman points out, economists and other social scientists relied mainly on working papers from reputable outlets (such as the National Bureau of Economic Research), with "journals serving as tombstones" for the finished products. And that was before the Social Science Research Network almost completely democratized the dissemination of academic working papers.
Academic journals are still with us, but, as Krugman argues, they no longer serve the traditional purpose of disseminating scholarly ideas. I even subscribe to a few (mostly economics) journals, but it is an increasingly rare event to come across a published article in my areas of research which I have not already read (or skimmed) in some working-paper version.
So what purpose(s) do (costly) journals serve? I can think of two. First, journals persist as (fallible) proxies by which academic departments assess an individual scholar's reputation and worthiness for promotion and tenure. Second, because working papers can and do change as authors respond to comments and criticisms, citation becomes a confusing and potentially misleading business. Journals, as "tombstones," identify the final remains and resting place of articles, providing reliable focal points for citations.
Are these two purposes enough to ensure the continued persistence of academic journals? Well, probably not in expensive paper form. Based on anecdotal evidence, electronic publication definitely seem on the upswing, and we might predict that, based on cost considerations alone, all journals ultimately will be published only in electronic form; print versions are likely to disappear entirely (saving lots of trees in the process). I am aware that similar arguments have been made about books, but it seems to me the extinction of printed academic journals is more likely than the extinction of either scholarly or non-scholarly books. For one thing, a lot of people (myself included) like to hold books in our hands. I don't know anyone who feels so strongly about the feel (and smell) of journals. I could well be overestimating the relative value of physical books, compared to physical journals; but I'm pretty sure, at least, that I'm not understating the value of printed journals.
The remaining question is whether democratization of the scholarly mission, represented by on-line publishing, including self-publishing, is a good or bad thing on balance. As a non-elite academic myself (who has a relatively tough time getting his work into the top peer-reviewed journals as well as top-20 student-edited law reviews), I tend to favor easier dissemination of a wider range of scholarly research. But within limits. As the pool of academic research widens and deepens, it requires more and more time and effort for readers to separate the wheat from the chaff (assessed by their own lights). Many (if not all) of us will continue to desire proxies, including gatekeepers, to minimize our search costs for quality scholarship in our areas of research.