An archive of first resort

The year 1991 was a momentous and turbulent one. A US-led military alliance drove the army of Iraqi dictator Saddam Hussein from Kuwait, the Soviet Union disintegrated into 15 separate countries, and the Yugoslav Wars, which would displace 4 million people and last until the end of the decade, began when Slovenia then Croatia declared independence.

The year was also memorable in popular culture. Terminator 2: Judgment Day was the biggest-grossing movie of 1991. At the Academy Awards, the most successful movie of that year was The Silence of the Lambs, which garnered Oscars for best picture, best actor, best actress, best director, and best adapted screenplay. Many, maybe even most, critics of popular music hailed Nirvana’s Nevermind as the best album of 1991.

And on 14 August 1991, Paul Ginsparg, a particle theorist based at Los Alamos National Laboratory, launched a new electronic service for sharing preprints.

Known first as the LANL preprint archive, the service was quickly adopted by nuclear and particle physicists. Other kinds of physicists followed their lead. In 1993, two years after its foundation, the LANL preprint archive received about 7000 papers. Twenty years later, after its move to Cornell University and a change of name to arXiv, the rate topped 90 000 papers a year.

The success and popularity of arXiv led to speculation that it would render traditional journals superfluous. My first inkling of that possibility came in 2005 when I wrote a news story for Physics Today about topological quantum computing. The paper that begat the field, Alexei Kitaev’s “Fault-tolerant quantum computation by anyons,” was posted on arXiv in 1997. But it wasn’t submitted to a journal, Annals of Physics, until 2002.

In 2008 I wrote about Hideo Hosono’s discovery of superconductivity in an iron-based compound and the explosion of research that ensued. As theorists developed their ideas about the new materials and as experimenters created new members of the iron-based family, they posted their papers to arXiv to establish priority. But the researchers cared just as much about submitting the papers to peer-reviewed journals. ArXiv, it seemed, was not competing with Physical Review Letters, Nature, and other journals, but complementing them.

Thanks to Vincent Larivière of the University of Montreal and his colleagues, we have a comprehensive and quantitative assessment of how the posting of preprints on arXiv affects the publication of the corresponding papers in peer-reviewed journals. Their paper, “arXiv e-prints and the journal of record: An analysis of roles and relationships,” was posted on arXiv on 13 June 2013 and published in January 2014 in the Journal of the Association for Information Science and Technology.

The scope of Larivière and his colleagues’ investigation was immense. They set a computer program loose on all 744 583 preprints published in arXiv up to 22 March 2012 and had the program compare the preprints to all the journal papers indexed in Thomson Reuters’ Web of Science . By looking at author lists and abstracts, the program could identify preprints that later became papers.

Larivière and company’s paper has 11 figures, each depicting a major finding. Despite differences in the adoption of arXiv among various disciplines, the picture that emerges from the study is consistent. Physicists and other scientists use arXiv as it was originally conceived and in the same way that their predecessors used paper preprints: to establish priority and to promptly alert their peers.