“Citing and Reading Behaviors in High Energy Physics,” A. Gentil-Beccot, S. Mele and T. Brooks (2010)

Invention and research, particularly its speed, is a critical question for economists. I wonder how many of us economists know just how strange journal practices are in our field. The years-long review process is perhaps known as an anomaly; 3 months from submission to publication is the norm in many hard sciences. But the really unique thing in economics is the near universal “green” open access. That is, economists almost always publish working papers on the web as they go, and they almost always put ungated versions of their final published papers on the web as well. This is strange indeed. Many fields have a custom of rejecting any research that has been “published elsewhere,” with a working paper on a personal website counting as such. Further, very few fields have ungated access to their research: all forms of open access make up only about 20% of published scientific research. Many governments including the US are trying to change this. Since 2008, the NIH requires grantees to place an ungated copy of their research in a central repository within a year of publication. Unfortunately, a group of congressmen made up of the usual cabal of copyright maximalists is trying to reverse this rule.

Such political posturing calls for evidence to counter it, then. Does open access lead to “better science”? More and quicker citations? Faster rates of invention by non-academics? Should there be any year-long embargo period in the NIH or in potential NSF rules? Is there an “access problem” at all, or are non-academics totally uninterested in academic research? In general, these questions are totally unanswered in the literature. Nearly everything on the topic thus far has focused on download numbers and citation counts for individual articles that were randomly or pseudo-randomly declared OA. Download and non-academic mentions of the OA articles were higher, but citation counts didn’t change much. I find it much more interesting (and I am doing exactly this in a current research paper) to consider an ecosystem of open access, or fields like economics and high energy physics where ungated research is nearly universal, and where non-academics know they can read frontier research without a hassle.

The present article considers high energy physics. Particle physics is concentrated in a relatively small number of labs, and since the 1970s, article preprints and working papers have circulated among these labs. A database from that era began tracking these working papers and citations to them. This database eventually moved online (indeed, it was the first database on the internet) and physicists now nearly universally post preprints of their research on ArXiv. This near-universality has been the case for over a decade now.

Articles submitted to ArXiv and then published have been cited 4 to 5 times as often as articles which are only published. This isn’t an apples to apples comparison due to self-selection: I may wish to put my best work online right away but not bother with lower quality work. As I noted above, randomized open access hasn’t really shown any citation advantage to the OA article. Much more interesting here is the timing: articles submitted to ArXiv as working papers garner nearly 20% of their citations before they are even formally published (for articles not submitted, the total by the time is, of course, 0%). This is the best evidence I know of for the “speeding up of science” due to open access and related policies. The work of Benjamin Jones suggests that speeding up the ability of researchers to reach the frontier in their field is critical for maintaining economic growth in coming decades.

One other interesting data point from this article. A bibliographic search tool in physics called SPIRES links to both the publisher website version and the ArXiv version when both are available. The publisher version is just a pdf of the final journal article, while the ArXiv version often lacks nice formatting. Nonetheless, 82% of clicks on SPIRES when both versions are available go to the informal ArXiv version. What’s going on here? The authors speculate that since ArXiv is known to require only one click, and is ensured to never have any of the minor access problems (log in here, do this there) that gated copies may have even for institutional subscribers, ArXiv has grown to be the de facto choice. This strikes me as compatible with other research on innovation – Heidi Williams’, for one – which suggest that minor hurdles for scientists can dramatically affect the direction of science.

I will have much more to say about the economics of open access in coming months.

http://arxiv.org/vc/arxiv/papers/0906/0906.5418v1.pdf (Fittingly, I link to the ArXiv version here. A post today on orgtheory mentioned that a major breakthrough in medicine has just been published in an open access PLoS journal; that is, open access journals are totally mainstream now, and are respectable venues for even top-notch research.)


One thought on ““Citing and Reading Behaviors in High Energy Physics,” A. Gentil-Beccot, S. Mele and T. Brooks (2010)

  1. LuigiP says:

    Cool post buddy but I still prefer your monkey pics, at least weakly 🙂 Anyway:

    “This is strange indeed.”

    maybe it’s weird at a first glance but it doesn’t strike me as a particularly counterintuitive results once you think about individual incentives.

    (1) if the publication process is a lot faster, there is no need for a time stamp before publication itself.

    (2) Our “intrusive” refereeing practices make it essential for people to gather all the comments they can before submitting the paper to a journal, in the hope they can skip a round or 2.

    (3) In the hard sciences the fields are way more narrowly defined than in Econ and there is more weight on replication exercises and the like. Being scooped is a real danger: Put an innovative but somehow flawed study on the web and someone will publish a better one in the blink of an eye, reaping most of the benefits. Compare that to Econ where research efforts sprawl in often counterintuitive directions (sumo wrestlers, anyone?) and replication exercises and comments are rarely published in good journals.

    I didn’t write down a model but I think you could easily do…

    “This strikes me as compatible with other research on innovation – Heidi Williams’, for one – which suggest that minor hurdles for scientists can dramatically affect the direction of science.”

    This is a very interesting point, let’s gather data 🙂

Comments are closed.

%d bloggers like this: