What I am missing in both tables is an overview of what journals are actually offering for these costs. For example, PLoS One* has no copyediting process, and also seems to skip the final proofreading step common at most other journals (galleys are not sent to the author for final approval).
As a result, some authors decry the quality of their published work. For example, John Logsdon [WebCite] writes
There was no opportunity given for making corrections to proofs. I have already identified an issue with one of the tables that would have been corrected in proof had there been an opportunity.On the same page, Banoo Malik offers an insightful comment:
I contacted the production staff and they mistakenly gave it the citation year of "2007" not 2008. Thus, my first primary-authored paper will likely NEVER appear in anyone's eTOCs,
Obviously you get what you pay for. Cheap production costs yielded some production oversights. I can appreciate how many hours and effort a copy editor and production staff spend on these seemingly small details now since I followed up on a few points myself. [WebCite]
On the same page, Andrew Staroscik writes:
I am unhappy with parts of the BMC process. I agree that the pros outweighed the cons but I also had an issue with proofing. The BMC process did not have a copy editing step! There are typos in the published version of our paper that I and the other authors did not catch until it was tool late.
Is having a competent copy editor look over the proofs really a particularly costly step in the process? [WebCite]
The answer to the last question - at least from my perspective as publisher of JMIR - is a clear "yes". JMIR charges a $1500 Article Processing Fee, but spends most of this budget - hundreds of dollars - for every article to clean it up before publication, to standardize the reporting, and to improve language issues. Bringing an article into the final form, including checking all the references, crosslinking them to DOIs, PMIDs, PMCIDs etc. is - even though a semi-automatic process - a very timeconsuming and expensive undertaking.
This is what authors often don't realize when looking at APF comparison tables such as those published above: There are considerable differences between journals on how much time and costs they spend on the actual production process, which justify different article processing fees.
The notion that the "ONLY major cost in running a journal is the time spent by the editorial team and reviewers" [WebCite] is - at least for how I run my journal - not accurate.
In fact, JMIR is spending most of the article processing charges on external contractors - copyeditors, XML typesetters etc. There are several copyediting and proofreading steps in the production process, which not all journals seem to employ. I invite authors to critically assess what they get for the money instead of just looking at the article processing fee. JMIR staff and external contractors spend on average 10-20 hours to bring a manuscript into its final form.
And open access journals not charging any Article Processing Fees are almost guaranteed to skip these steps. Many won't even have XML versions of the articles (i.e. no submission to PubMed Central), because creating them is very expensive.
So I would encourage academic authors considering to submit articles to an open access journal to not only ask how much the article processing fee is, but also what they get for their money.
Correction notice: Edited 25/11/2008. I am taking back (and have removed) my original remark on "superficial" [WebCite] peer-review. More correct and value-neutral would have been to say that some journals have a leaner ("hassle free) [WebCite] peer-review process than others (for example, editors may decide to accept an article without sending it out to external reviewers).