Researchers, post your work online

I just submitted an invited review of the evidence of reciprocity in vampire bat food sharing. This allowed me to get out a bunch of data from unpublished studies, most of which were negative results from groups of vampire bats that did not share food with each other.

Publishing negative results is important, especially as the difficulty of publishing becomes trivial. This leads me to think about scientific publishing and the ethical imperative of open access. This has all been discussed elsewhere and elsewhere and elsewhere by many people, but here’s my two cents and my attempt to get the word out…

Scientific publishing is in my opinion a strange affair that is still based on archaic ways of thinking founded on outdated technologies. The good part of the current system works like this: Scientists convince funding agencies to give them money so they can do research. They want their research to be disseminated as widely as possible, so they write up their results, and turn them over to publishers free of charge. The publishers send the work to an editor. Scientists also give these editors a list of potential referees, other experts who would be able to critically peer-review the work. Each scientist also typically reviews the work of others (again free of charge) and can often act as an editor too. The authors then revise their work and the reviewers look it over again (free of charge), until the editor decides it’s worth being published. The paper can also be rejected (free of charge, haha).

In a more sane world, scientific publishing would work like much like wikipedia. Scientists would post all their work online for all the world to see and evaluate. Peer-review would be an open and continuous process. Text, citations, and datasets, both small ones and huge public ones like Genbank, would be networked and could always be updated. Each scientists would make a trackable contribution to this huge open and public library of scientific knowledge. The goal of publishers would be to help make content visible or to provide commentary on it, like a magazine. This is what some publishers like the Public Library of Science (PLoS) are trying to do.

But instead, this is where things get weird.  Most of the time, this is what actually happens. After all the actual work is completed on a free and voluntary basis by scientists, publishing companies step in to adjust the layout and font of the paper to their own liking, and then claim copyright over the article. If you are lucky, they do some proofreading. Although their main job is to disseminate information as widely as possible, they do the exact fricking opposite. They lock their now-copyrighted article behind a pay-wall on their website, and then charge a huge fee for downloading it. And publishers typically don’t even do any marketing or disseminating of the research. They do however “brand” it, meaning that your article becomes a “Nature” paper or a “Biology Letters” paper or a “Canadian Journal of Zoology” paper.

If you think you can properly evaluate the validity or importance of these papers before you even read them or know what they are about, then branding has done it’s job. Branding is when companies make money off people’s tendencies to form stereotypes based on limited bad information. For example, because of the “mere-exposure effect” –well known to advertizers– fast food logos displayed in monkey cages can shape the monkeys’ later preference for the same foods packaged with different brand logos (Laurie Santos, pers. comm.), and similarly, scientists will think papers in a journal are better simply because they have been exposed to the journal more. This judge-a-book-by-its-cover philosophy is not the finest example of critical thinking, but it’s the reason why universities rank research output based on the journal and why we all fight to get our papers into the best journals. And it’s how publishers can make huge profits off scientists without paying them.

I have no idea why science puts up with this system. It screws everyone but the publishing companies. It hurts the authors who want their work to be freely available and not hidden behind a pay wall. It hurts the public who pays for the research twice: first they (often) fund it as taxpayers and then they have to pay to access it… even though the author *wants* to give it away for free! It hurts universities that have to pay utterly ridiculous prices for access to scientific information, up to $40,000 per year for a single popular journal. Yes, $40K for a magazine. It hurts students because these costs then increase tuition prices. It hurts the entire scientific enterprise, because the publishing process becomes so longwinded, inefficient, and focused on journal impact, that scientists don’t bother to do things like publish negative results or datasets.

Many academics forget that paywalls exist, because we usually get access through our own universities. We forget that there are also many non-academics who would like to read scientific papers. Inevitably though, we come across a paywall in trying to get access to some article. Many people also think that we acaemids at universities don’t pay for journal subscriptions (because the school does), but we do indirectly. The average price of a journal subscription for a science journal was $1,500 in 2009 and rising *per title*. How much is the total cost? Well, Harvard spends $3,750,000 on journal subscriptions each year from a few “big providers” and that’s only 20% of their subscription costs. Similar pressures are likely at work at my own university, which has been forced to cut thousands of journal subscriptions just to operate within budget. How could this not trickle down to faculty and graduate students?

Most damaging of all in my opinion is that publishing companies have over time helped re-frame the purpose of scientific publishing from [making new information freely available] to [advancing one’s career through branding]. I have even heard of journal panels considering the idea of asking the authors to cite more articles from their own journal to help increase the journal’s impact factor (the brand). This is of course completely unethical, but I’m sure it simply didn’t occur to these otherwise moral people, because incentive structures shape ethical behavior more than people being good or bad. The current system simply does not reward good scientific practice.

So what’s the other side of the story? How do the publishers see it? Here’s a copied-and-pasted discussion I had last year with someone more connected to the publishing company world (emphasis mine):

Me: Providing maximal access to content is not some kind of special feature of publishing. In my mind, the whole point of publishing scientific results is to make them available to others. Since you clearly disagree, I would like to know what you consider to be the “service” of publishing.

Rick: Gerry, if your only concern is providing maximal access to content, then why are you bothering to publish with your papers formally at all, whether with PLoS or anyone else? You can save yourself a lot of trouble by simply posting your research reports on a blog or other publicly-available website.

Me: Hi Rick. Short answer is peer-review. But there’s a longer answer that addresses what I think is your point. Most scientists, myself included, have both (1) a “selfish” motive to maintain or advance their career and compete in the job market, and (2) a socially “altruistic” motive to solve puzzles, and create and share new information. These motivations typically work together of course. But sometimes they are at odds. For example, scientists dont always publish negative results, since the net career benefits are often small or negative (motivation 1), but null results are still very important for science (motivation 2). Also, scientists keep their newest results secret to avoid being “scooped” (again, good for scientists, bad for science). Scientists are also implicitly encouraged by the current system to “sell” (and perhaps over-sell?) the significance of their findings and confidence of their results. It seems to me that policy-makers and publishers should act to reduce these sorts of motivational conflicts. The peer-review process of PLoS ONE is a good example (the science is reviewed first, then the paper is published, then the paper-specific impact can be reviewed and tracked continuously). This process emphasizes the science over “spin” and allows findings to be peer-reviewed continuously. If you imagine a completely non-careerist and altruistic scientist (with motivation 2 only), you realize they might publish *only* in PloS ONE. The success of “public goods experiments” like wikipedia and PloS ONE are an inspiring testimony to people’s inherent need to think, share, and contribute ideas. Getting back to your question, I do have a blog, and I do get annoyed when authors don’t make their papers freely available on their website. In this day and age, they should. Obviously, my POV is as a reader and writer, not as a publisher. Perhaps that’s why we aren’t making sense to each other.

Rick: But peer review does nothing to maximize access to content. In fact, it does only the opposite — it’s a barrier between the author and the reader, one that causes both delay and (in many cases) rejection, making dissemination less likely, not more. Why do scientists put up with this barrier, if their whole goal is to provide maximal access to content? Or, to phrase the question your way, what is the “‘service’ of publishing?”

In reality, I think you’ve answered your own question. You’re correct that scientists are interested in much more than just disseminating information. They are also (like most of us) interested in advancing their careers, and the only way they can do so is if their products are validated by trusted third parties. That process of validation is one of the primary services provided to authors by publishers (whether for profit or nonprofit, whether toll-access or Open Access). In other words, “providing maximal access to content” is not “the whole point” of scholarly publishing; it’s one purpose among several, third-party assessment and validation being one of the others. What you describe as PLoS’s validation process–peer review followed by publication, after which it becomes possible to monitor impact–is the same process followed by toll-access publishers. The service provided by PLoS is essentially the same as the service provided by Elsevier or Wiley. The big difference between them lies in who pays for the service.

———-

But what is third-party assessment? Publishers cannot seriously take credit for the scientific part of peer-review since that is done by other scientists for free. So “third-party assessment” must mean determining for the reader what is trendy and important, or determining which editors knows what’s important. So there you go. That’s what publishers do for us. They tell us what to read and give us “validation”. In exchange, all we have to sacrifice is public access to information.

The funny thing is that branding and validation is no longer a service required by journals, because it can take place on an individual-by-individual basis thanks to Google Scholar citations profiles, which measure the impact for each article and for each individual scientist. Journal impact factors are already obsolete when assessing and hiring faculty. As for the validation and branding of articles before we even read them, we should ask ourselves: is that even a good idea??

What’s the take home message? First off, scientists could completely fix this problem if they cared enough about it. Along with 13,750 others, I joined a boycott refusing to publish or referee for Elsevier journals when they tried to make mandatory open access of publicly funded science illegal (afterall, why should the government be able to provide public access to research funded by taxpayers??). This did actually had an impact.

One huge solution is very simple: All researchers should at least make all of their papers freely available on the internet. Ideally, this would include both preprints and postprints, as well as publishable datasets. Also, if you have old work that’s only in a book or on paper, scan it, and put it online! Are we trying to keep this stuff a secret??