Wednesday, December 19, 2012

Open access publishing

I am still not entirely sure what I should think of open access publishing of scientific research. Don't get me wrong, I am not at all fond of the traditional system which, in case a non-scientist reads this, works like this:
  1. Jane and Joe Taxpayer fund a researcher's salary and their research costs.
  2. The researcher submits a manuscript presenting their results to a journal, where it is peer reviewed by qualified colleagues. This quality control is done for free, as a service to the scientific community, with the understanding that other colleagues will in turn referee the referees' own manuscripts sometime.
  3. If the manuscript is accepted by the journal, it is handed over to a commercial publisher. That publisher sends it to some underpaid chaps usually in India who typeset it into the journal format.
  4. The final paper is then printed and placed behind a paywall on the internet. If your (taxpayer funded) university library wants to provide access to the published results of the (taxpayer funded) research, it has to fork over a hefty fee to the commercial publisher. If a (taxpayer funded) research institute wants its staff to be able to access the results of the (taxpayer funded) study online, it has to fork over a hefty fee to the commercial publisher. Note also that Jane and Joe Taxpayer cannot read the research papers they funded unless they happen to be staff members at an institution that is forking over aforementioned fees.
  5. At the end of the year, the shareholders of the publishing company buy themselves a second yacht or another villa at the coast. Seriously, this is not about paying a reasonable price for good quality, this business is so obscenely profitable that the term "market failure" springs to mind.

The problems are obvious, and the last few years have seen a strong movement in science trying to make scientific publishing more open and to pry it from the hands of a small number of publishing houses. This movement reaches from a grassroots campaign to boycott the publishing company Elsevier to the government of the United Kingdom, and accordingly diverse are the reforms that are suggested:

Green Open Access means that the journal allows the author to make a pre-typesetting version of the paper publicly available, for example on the author's personal website, as long as it contains link and reference to the commercially published version. Gold Open Access means that the final paper as set into the journal's format is publicly available, generally after the author has paid a considerable publication fee. The governments of some countries are considering or, in some cases, have enacted laws to make open access publication mandatory for publicly funded research.

Some researchers go even further and want to revolutionize the entire setup of journals with editorial board, peer review and painstakingly formatted publication. Gripped by enthusiasm about social media, crowdsourcing and citizen science, they sometimes envision blog-like scientific publishing: a scientist would simply publish their results online, and then colleagues could peer review and discuss the "paper" as in the comment thread under a blog post, and maybe even give it "likes" or up- and downvotes.

The thing is, I see potentially severe problems with all these except perhaps the first and least radical - Green Open Access is simple and straightforward, but of course only possible if the publishers are willing to allow it. But the others?

Gold Open Access basically only shifts the costs around. You still need to pay a lot of money to get quality typesetting and profitability for the publishing houses, but instead of making the libraries and readers pay you make the scientists pay. While now the issue is that cash-strapped libraries cannot afford to subscribe to the journals they need and that colleagues from poor countries or underfunded institutions cannot access articles that they need on the internet*, the open access utopia would see scientists from poor countries and underfunded institutions unable to publish in any decent journal because they could not afford the fees. Some non-profit open access journals may waive the fees if the author can make plausible that they are unable to afford them, but it cannot be denied that the whole process will constitute another barrier to entry for some colleagues, either practically or at a minimum psychologically.

But what I am most worried about are incentives. In the old model, a journal publisher is paid per published volume of each journal that a library subscribes to - regardless of the number of articles in it as long as they are enough and of sufficient quality to make the journal interesting and relevant in the field. On top of that, they are paid per article that some individual finds relevant enough to buy access to on the internet. Notabene: both these processes provide incentives to accept only manuscripts of high quality. With open access, however, a journal is paid by the number of papers it accepts, and that means there is an incentive to accept as many as possible. Now I don't mean to say that I know any details about their standards of quality - I have no inside knowledge - but it has been widely discussed that the non-commercial open access journal PLoS ONE, for example, has an acceptance rate of around 70%. Although I have so far not served on any editorial board myself, I know from colleagues that other good journals these days have acceptance rates of 30% or less. Interesting observation, isn't it?

Finally, the really revolutionary considerations, like transforming science publishing into some kind of blog-like format. As mentioned, you can see where some people get these ideas if you look at Wikis and social media, but think for a minute about what it would do to quality control. Now, the assumption is that you have probably done decent work if you manage to get your manuscript past two to four peer reviewers and at least one managing editor and one editor in chief. Does not always work, of course, but their role as gatekeepers greatly reduces the amount of crap that you would otherwise have to sieve through to find the good papers. It is perhaps even more important for non-scientists who are hardly able to judge the methodology quality of, say, a paper on climate change or a drug trial but who assume that a decent journal will have vetted the paper before accepting it. The entire process also guarantees that the author will receive comments and suggestions for improvement that they have to act on or rebut before their manuscript is published.

How would that work if we all just uploaded our papers into a big database where people can do the peer review in a comment stream or with up- or downvotes? For starters, quite a lot of nonsense could end up in that database, as is indeed happening with the existing real life example ArXiv that has long been used by physicists and mathematicians. That will make it harder to find the good articles. A solution would be to check out only the publications from well established and renowned colleagues (who will also be so well known and influential that they attract lots of upvotes/likes/positive reviews), but that will create a huge barrier to entry for competent newcomers at the beginning of their career.

Second, many a lazy colleague would consider the publication process to be finished when they have dumped their paper in the database. You may envision people in the comment streams to make intriguing suggestions for improvements and to provide some crowdsourced feedback, but where would the author take the motivation to actually correct anything, be it a misleading sentence in the discussion section or even just a formatting error? The manuscript is published, it can be cited, time is precious, on to the next project!

In summary, I don't really know what to do. The obvious solution that occurs to me would be to keep the old system but to have an antitrust agency look into the commercial publishers or, better yet, to simply turn them all into publicly funded utilities or non-profit organizations, but that is not going to happen. Something will need to change. The point I want to make is this: Let us all think very carefully about the incentives that will exist in an alternative system.


*) Note, however, that it is standard practice to simply request a reprint or PDF from one of the authors if one does not have access from one's institution. No commercial publisher forbids that, and the authors are nearly always happy to provide you with it because they hope you will cite their paper and because they rely on the same goodwill from others. This considered, the current system does not really restrict access to papers for scientists; the real problems are lack of access for the public and high costs for libraries.

No comments:

Post a Comment