Varmus Screws the Pooch

Harold Varmus is one of the most high profile advocates of open access to biomedical research. As one of the cofounders of the Public Library of Science (PLoS), he has played an important role in making published results freely available to all. And he's a Nobel Laureate, which ain't too shaby either.

Varmus was interviewed by Ira Flatow for NPR's Science Friday program about the NIH's new policy requiring that research publications presenting results funded by the NIH be deposited in PubMed Central (the NIH's free online archive of biomedical journal articles) within a year of publication. Before I get into the problems with Varmus' interview, I'd like to highlight one important point Varmus made: the current NIH policy is a comprimise between the ideal open access solution and what pay-access publishers want. In the ideal situation, all papers published about publicly funded research would be made freely available at the moment of publication. There would be no moving wall before they can be freely obtained. But that would ruin the business model of the pay-access journals, who make a fair bit of their profit from both individual and institutional (i.e., university) subscriptions. If a big chunk of the articles in those journals were freely available, the journals would stand to lose many of their subscribers.

Despite making that point, Varmus struggled to point out the big differences between open access and pay-access journals. He was clear that the differences lie in the business models, but he did not present the details of those differences with the clarity required when speaking to the general public. First, he made it seem as if open access journals are the only ones that require a payment from the author in order to publish in those journals. Second, he failed to clearly point out that open access and peer review are orthogonal issues, despite a perfect opportunity to do so.

Varmus made the important point that open access and pay-access journal differ primarily in their business models, not in their approach toward publishing scientific results. However, he made it seem as if open access journals are the only ones who require authors to pay to publish. This is not the case, as pay-access journals also require authors to pay publication fees. Using a few journals from my field as examples, Genome Research charges $40 per page, Molecular Biology and Evolution charges $50 per page, Evolution charges $55 per page, and Genetics charges $65 per page. Those page charges may differ depending on whether you are a member of a society associated with the journal, and they often also charge additional fees for publishing color figures. Generally, a ten page article with one color figure will cost approximately $1000.

PLoS also charges authors to publish in its journals. For example, PLoS Biology charges $2750, PLoS Genetics charges $2100, and PLoS ONE charges $1250. These charges are fixed and do not depend on the length or contents of the article. Nucleic Acids Research, an open access journal published by Oxford University Press, charges $2670 to authors. Additionally, pay access journals often give authors an option to publish their paper with free access, for an additional cost. For Genome Research this cost is $2000.

As you can see, the publication charges for open access journals are often slightly more than those for pay-access journals. But it's not as if one business model requires author payments and the other does not. Additionally, all journals I have encountered allow editors to waive publication costs if the author(s) cannot afford them. Some are more draconian in their presentation of those rules -- see the page charge policy of MBE -- but they all at least make that option available. This is true for both pay access and open access journals. Therefore, the primary difference in the business models between pay access and open access journals is not in the presence of author charges. There is a difference in the magnitude and application of those charges, but the primary difference is in whether the journal requires readers to pay to access the content.

What about Varmus' defense of the peer review in open access journals? In this case, a caller phoned in to ask whether open access journals have the same peer review standards as pay-access journals (quite reminiscent of the whole PRISM affair). Varmus correctly pointed out that open access journals require the same scientific rigour as pay-access journals. However, he failed to hammer home the point that open access and peer review are completely orthogonal issues. That is, whether a journal offers open access to its content is independent of the nature of peer review the articles are put through. Conflating open access with peer review is a propaganda strategy of the anti-open access lobby, PRISM. I think Varmus wasted a great opportunity to make this explicitly clear.

In addition to discussing the NIH open access policy, open access business models, and peer review, Varmus also talked about publishing negative results. He pointed out the need for journals to make negative results available. This was somehow connected to Varmus plugging the new interactive features available in some PLoS journals -- specifically, the ability of readers to comment on articles (a feature this is noticeably under-used), which I presume would allow readers to mention how their own negative results relate to the article.

The entire interview is available on the Science Friday website. Varmus did a good job, but he could have done much better. He needs practice in hammering home the primary talking points. Some of his message got muddled in his lack of a punchy delivery. I worry that the points may have been missed by those people not familiar with the details.

More like this

So are you saying that Dr Varmus may need some coaching on framing his message(s)???

Well, I tried to make it through the entire post without mentioning "framing", but, yes, it's an issue of framing. The difference, here, is that the framing is not of a scientific issue the public fails to grasp (i.e., evolution, global climate change, etc), but a policy issue. In the case of changing policy, proper presentation of your talking points is key.

"Generally, a ten page article with one color figure will cost approximately $1000."

I've seen statements like this several times, but never any real data to back it up. Yes, you provided some anecdotes, but that's not exactly systematic. I've never been charged to publish a paper. That's anecdotal as well, but the combination of your anecdote and mine suggests that some more systematic evaluation is necessary before we can start speaking of what "generally" is or isn't done in scientific publishing.

MRW, based on the prices I quoted, ten pages (at $50 per page) costs $500, plus $500 for one color figure (taken directly from one of the journal webpages), which comes to a total of $1000. That's a back of the envelope calculation, but it is based on real data (the actual prices journals charge to publish in them). If you submit an article that is long enough or has multiple color figures, you could get charged more to publish in a pay-access journal than an open access journal.

All true, but that has nothing to do with it being anecdotal. I'm not surer how you get from the policies of four journals to "generally".

I'm not saying it's necessarily an incorrect generalization, just that it's one that I haven't seen it supported by anything more than anecdotes of a few (in this case, four) journals. Starting from a different handful of journals could lead to a very different generalization. For example, as I mentioned, none of the journals I have published in charge to publish.

Here are the publication charges for the first few pay-access journals that I have bookmarked in my web-browser:

A J Human Genet: "Members of the American Society of Human Genetics will not be charged page charges or color charges. Nonmember authors will be charged $450 for the first color figure and $200 for each additional color figure. In lieu of a per page fee, non-members will be charged a single fee of $825 for Articles and $525 for Reports."

Am Nat: approx. $50 per page + color figure charges

Bioinformatics: $190 per page for going over page limit. Optional fee to make article open access.

Cell: Color figure charges.

Chromosoma: No page charges or color charges.

Chromosome research: No page charges or color charges.

Current Biology: Color image charges.

Development: No page charges or color charges.

EMBO: $200 per page for going over page limit, plus color image charges. Open access available for $2,540.

I'd do more, but I don't have an infinite amount of free time. You are correct, some journals do not charge page-fees. The Springer and Cell-press journal appear to fall into that camp, although it would be worth examining more of those journals. But page-fees are common, if not the norm, in pay-access journals. And to make a pay-access journal publish something as open access seems to cost more than if the article had just been published in an open access journal to begin with.

Also, in geoscience journals at least, color images tend to cost a great deal more than grayscale images. I've had lots of practice making small, information-packed, readable grayscale maps. AAAGGGHHH!

MRW: a little out of date now, but the Kaufman-Wills study The Facts about Open Access showed that ""...more than half of DOAJ journals did not charge author-side fees of any type, whereas more than 75% of ALPSP, AAMC, and HW subset journals did charge author-side fees."

DOAJ = Directory of OA Journals
ALPSP = Association of Learned Professional and Scholarly Publishers (non-OA)
AAMC = Association of American Medical Colleges (non-OA)
HW = HighWire (non-OA)

See here for a bit more detail ('scuse the self-link). In fact, the sorts of questions you are asking are very valuable, but very difficult to answer. One has to either design and run large surveys (and have some clout, or get no responses!), or laboriously transfer the relevant information from clumsy journal/publisher websites into some kind of useable format. Given that there are about 25,000 scholarly journals now, even a representative sampling is quite the undertaking.

Bill,

Thanks, that's useful.

Mike