Socialism in science, or why Open Access may ultimately fail

“Open Access is the best thing since sliced bread” I hear a lot of scientists say. The blogosphere overflows with enthusiastic support for OA. Hey, even I expressed some guarded optimism in some of my previous posts. This unconditional love for the NewestAndHawtest publishing model is very common, but is it really justified? More importantly still, will those same ardent OA supporters put their money where their mouths are? I am not so sure.

OK. So here’s the argument: Science needs to be made known through publishing. In order for science to be published, we need a team of people who will do the job of supervising the review process, editing, storing, and distributing bite-sized pieces of science called papers. These people need to be paid. Traditional closed access (CA) journals pay their employees from revenues they generate by selling their journal or the right to access it online. OA journals accomplish the same through selling space in their journal to people who want to publish in them. The main question is “Can OA journals publish high impact science, and also break even financially?“. On John Wilbanks’ blog on nature network there was a discussion a while ago on the profitability of OA journals, and I particularly enjoyed Martin Fenner‘s comment. He said that there he only sees four ways how high-impact OA journals could be making a profit (comments are mine):

Additional sources of income, e.g. ads on the website. The problem here is that the whole idea of OA is that anyone can download the paper and make any use of it they want. But that means that anyone can just post the paper on their website, and, by adding value, say automatic indexing of keywords iHop style, have people come to their website rather than the journal website to download it. And there goes the advertising money.

Much higher author fees (e.g. $5000-7500) or an author fee apply to every submitted rather than every accepted paper. From the point of view of authors, both solutions are fundamentally the same. Since a high-impact journal will have a high rejection rate, you will need to submit several times to such a journal to get accepted, so it doesn’t really matter whether you pay a few grand for the final accepted result or whether you pay a few hundred each time you submit. As long as you are being realistic about what will fly and what will not, you will be spending about the same amount per paper. But the trouble is, authors will not go for that. Imagine you have a piece of really hot, high impact data that you want to put on display. Are you going to go to a CA journal that will gladly accept your paper without making you pay a dime, or are you going to flesh out $5000 to have it published in an equally prestigious OA journal? I mean, c’mon. Let’s be realistic here for a moment. Of course you will pick CA. If you will excuse a little parable, papers in CA journals are like art exhibitions, while papers in OA journals are more like paid ads – if the stuff people have is worthy of an exhibit, they will very likely not pay for an ad. Also, the incentive on the side of the journal is all upside-down in the OA model. CA journals want to publish the best science possible, because that’s what their clients, the readers, will pay for. OA journals want to publish as much as possible, irrespective of the impact, because that’s what generates income for them. The client is different here – it’s the authors, not the readers.

Subsidies by other journals. That’s the PLoS model. They have a few high impact journals, and they have PLoS ONE, where anyone can publish any work irrespective of impact, as long as it contains broadly understood “good science”, and as long as they are willing to pay for it. Their high-impact journals generate losses, and PLoS ONE is supposed to generate enough profit to make up for that. That this is not yet the case was widely discussed in the blogosphere after an extremely inflammatory article in Nature. Never mind that, I don’t necessarily doubt that PLoS will break even in the upcoming years, but there is another major flaw in their model – there are only so many people willing to pay to publish low impact data. That doesn’t mean that PLoS will run out of customers, but it does mean that any other publishing group who wants to repeat PLoS’ (still uncertain) success will. Also, any for-profit OA group based on the PLoS model will be tempted to expand low impact bulk publishing and shrink high impact loss-generating journals.

Government/charity donors (I added this one). That’s how a lot of OA community efforts run right now (Science Commons, for example). That is not the way to go, because at some point the donors will want to see some tangible effects of their donations, and there aren’t likely to be many. Besides, any organization based on government subsidization is likely to become less efficient and end up sucking more and more of the taxpayers money.

And so I hope you see by now that there are fatal flaws to all possible solutions of the high-impact OA publishing perpetual money problem. There are people who point out that from a macroeconomic perspective CA and OA are really equivalent, because it is all taxpayers’ money, whether it goes for paying publishing fees or buying subscriptions for CA journals. But it’s microeconomic effects where OA breaks down. It’s a question of individual scientists deciding where to publish and individual editors deciding whether to accept a paper and whether to charge for it and how much.

From what I have said so far you might think I have become an opponent of open access. That I have turned to the dark side, so to speak, and become a minion in the hands of the evil Nature empire. That is, most emphatically, not the case.

First, I think that OA is a great paradigm for a lot of areas other than scientific publishing. I am really looking forward to the expansion of OA data and biological materials repositories, such as GEO for microarray data and Developmental Studies Hybridoma Bank for antibodies. I hope that the whole Semantic Web hype will turn out to be well founded and that it will prove practical.

Second, I don’t deny that OA has its place in scientific publishing, but for the reasons I have outlined above, that place will probably lie in the low-impact end of the spectrum. We may see PLoS-like hybrids emerge, but it is extremely unlikely that they will take the publishing world by a landslide. Also, I acknowledge that the attempts to introduce the OA model in publishing are noble and carry a lot of potential benefits if by some miracle they succeeded. However, a lot of these benefits can be achieved much more easily without turning the whole world of publishing upside-down. For example if we want to give more people access to the newest scientific papers, let’s subsidize electronic subscriptions of high-impact CA journals to public libraries. If we want to hyperlink papers between one another, let’s expand current CrossRef schemes to not only include citation data, but also semantic realtionships between papers. And so on, and so forth… If only rather than dreaming of OA revolution all these intellectuals started thinking of how to do without it, we would do just fine.

And so to finish up I would like to make another parable. OA publishing is in a lot of ways like socialism. They are both intellectually and morally appealing at first glance, but they have some well hidden fundamental flaws, which make them unsustainable in the long run, unless enforced. Is OA publishing going to share the fate of the communist regime? Is it going to stay in vestigal form at the low-end margins of the scientific publishing world? Or maybe I am not getting it? Maybe OA is the best thing since sliced bread after all? Let me know what you think.

Here’s some more links to OA publishing discussion:

• a nice list of pros and cons on Informationsplattform Open Access
• on Bench Marks blog
• on Open Access Now
• on Medscape (pretty old)
• on Nonoscience

There are many, many more websites and blogs, mostly supporting OA, but the arguments are becoming repetitive after a while, so I will break up here, and leave the floor open to discussion.

Update: Please see Part 2, Part 3, and Part 4 of the series. Happy reading!