Underlying Shieber's argument, and that of a related blog post is the idea that subscriptions pay for peer review and other services such as copy-editing that add value to the author's work. The question that both beg is "how effective is the subscription at paying for these services?" Shieber points to hyperinflation in journal subscriptions. Clearly, the costs of providing peer review and the related servicces have not been the cause of this hyperinflation. Elsevier makes a billion dollars a year in profit from journal publishing. None of that profit pays for peer review; it is accounted for as a cost before profit is computed. ACS pays its executives enormous salaries, but the executives don't actually review the papers or copy-edit them.
If there is to be a level playing field, the question should be how cost-effective the various publishing models are at adding value to the author's work. At a time of constrained budgets, the more cost-effective model should have a great advantage.
Here is an example of the calculation we need to do, taking Elsevier and PLoS as examples simply because I have the numbers handy. Elsevier publishes less than 400K peer-reviewed articles a year. Elsevier's revenue from journal publishing is about $2400M, so the average paper costs the academic community at least $6000 to process. Thus the average Elsevier paper costs the community more than twice as much as the most expensive PLoS paper (The processing fees for PLoS journals vary from $2900/article down to $1350).
This measure, effectively revenue per paper, is interesting but not really useful. It assumes that each paper is equally valuable, which it clearly isn't. We need to normalize the cost by the quality of the papers whose quality we are paying to enhance. Eigenfactor is described as "A measure of the overall value provided by all of the articles published in a given journal in a year", so it would be a suitable measure for this purpose. An initial attempt at this normalization would be to divide each publisher's aggregate cost by the product of the number of papers and the average eigenfactor of the publisher's journals. This would be called "Quality Adjusted Processing Cost" (QAPC).
Alas, when I tried to get the data to normalize the costs I ran into two problems:
- I couldn't find a way to get the data from the Eigenfactor web site as a spreadsheet, and from Question 42 on this page it appears that this isn't a mistake.
- The most recent data in Eigenfactor is for 2008, and only contains PLoS Biology.