Academic researchers are known for keeping their work close before they publish it in order to avoid getting scooped. At the same time, research thrives on the open sharing of ideas and information.
It’s tough to balance those competing pressures, but 23 years ago a computer-minded physicist named Paul Ginsparg created a unique platform that met both needs. At the end of December it hit a big milestone: the one-millionth research paper was uploaded to the site.
The website, arXiv.org (pronounced “archive”—the “X” is for the Greek letter “Chi”) is a repository of “pre-prints,” drafts of papers that researchers circulate while they’re waiting for their work to be published in a journal, which can take months or more. Pre-prints give everyone faster access to results and also allow researchers to stake a quicker claim to their ideas.
In the 1980s, when Ginsparg began to conceive of arXiv.org while at Harvard, pre-prints were a valuable commodity that often only the highest-profile and most plugged-in researchers had access to. This tilted the race to produce new research in their favor.
“When we make a discovery we like to think it’s because we had some greater intuition or worked harder,” Ginsparg says, “not because we had some advanced access to information.”
Ginsparg, who’s now at Cornell, launched what would come to be known as arXiv.org in August 1991 while on staff at the Los Alamos National Laboratory. He says the service was quickly popular with low-level graduate students and with researchers in developing countries, who lacked institutional library resources. “It had this fantastic leveling effect,” he says.
Initially the platform was used by researchers in high-energy physics, Ginsparg’s subfield. Over time the practice of posting pre-prints to arXiv.org spread, and now the site is approaching 10,000 submissions a month (in addition to 10 million downloads). About half the submissions come from physics. The site is also popular with mathematicians and computer scientists, and is beginning to catch on in biology and finance.
At one point Ginsparg imagined arXiv.org would replace the traditional journal system. A couple decades on, he’s not so sure. “I’m on record as saying the current situation of having the journal and the arXiv providing these two parallel feeds can’t persist,” he says. “Of course I was wrong.”
Today, Ginsparg acknowledges that a main benefit of the journal system is that it gives people a guide to which research is worth paying attention to: if it’s published in a respected journal, you know it doesn’t contain, what he calls, “flagrant errors.”
Theoretically, arXiv.org could provide that same quality-control service, using a combination of the crowd-sourced methods found in places like Wikipedia and algorithms like Google’s PageRank. But Ginsparg worries those systems would be too easy to game.
“We all have this fear of the Internet being this anything goes venue,” he says. “It would be this Wild West thing unless you had adults inserted into the equation at various key places providing the right supervision.”
Inserting adults into the equation boosts quality, but also slows things down. When Ginsparg looks ahead, he imagines that for arXiv.org, the challenge will be finding ways to make it even easier for academics to share research. He compares the situation to social media platforms like Facebook and YouTube, which boost what he calls “spontaneous participation” every time they streamline the way content is shared.
Twenty-four years ago, sharing a pre-print meant making photocopies and mailing them to select colleagues and institutions. Today, it takes a practiced user a matter of seconds to upload a paper to arXiv.org that anyone can see.
“What we’re trying to do now is eliminate even that five or ten seconds,” Ginsparg says.
Correction: An earlier version of this piece misstated the arXiv.org’s monthly submissions. It is approaching 10,000 submissions a month.