Publishing in open access journals is the way of the future. Indeed it's difficult to see a future in which academic publishing looks the way it does now. I'm writing a book for Palgrave at the moment, but I wonder whether - if I write another book - I won't prepare and format the manuscript myself, and then publish it directly as a pdf through scribd or as kindleware. That way I get to own not only the whole product, but also the whole process, and the timetable that governs it. But not only that, the product itself ceases to be fixed. The manuscript can be amended, on the go, in the light of new thoughts and findings, and authoring processes can be shared in new ways using google wave or wikis like those provided by pbworks. How cool is that?

Another advantage of open access journals is that one can have a sense of the frequency and density of readership - simple data about numbers of accesses are provided by BioMedCentral.com for articles published in BMC journals. I knew Normalization Process Theory was worth pursuing when the first paper was accessed more than 1000 times in the month after publication in the on-line journal BMC Health Services Research. Taken together, the NPT papers on BMC have now been accessed more than 15,000 times through the BioMedCentral portal alone. 

Readership data is interesting because of what it tells you about trends, but it's hard to make more sense of this. I'm on the editorial boards of Sociology of Health and Illness and Nursing Inquiry, and publishers' data for those journals distinguishes between most cited and most read. This kind of data is available on the web for many journals. I find that some of my papers that have hardly been cited have been read in large numbers, suggesting that they are more useful to students than researchers. Which is great. It would be interesting, wouldn't it, if readership, not citation data, was used to assess research quality?