Tuesday, August 04, 2009

Science: open for business

A lot of the day-to-day of science is being shared all over the place, if you know where to find it. But historically, collaborations and idea-swapping took place at scientific conferences (albeit more likely in the breaks between sessions rather than at the talks themselves), and the latest research appeared about six months after it happened, printed out on real paper with real ink. Take this first issue of Physical Review Letters from July 1958:

Over fifty years later, it makes you smile to read this introduction, which includes gripes about the limitations introduced by switching from Monotype printing to the photocopying of typewritten pages. ("Alphabets other than Latin and Greek should be avoided," the editors plead.) Physical Review Letters was a journal spin-off of the "letters" section of the already venerable, half-century-old Physical Review. Outdated as their efforts may seem, the editors were accelerating the dissemination of scientific knowledge, both by taking advantage of this newfangled photocopying thing and by shepherding important papers through the review process and into the hands of scientists.
I learned all this from Alan Chodos, associate executive public information officer of the American Physical Society and talking treasure trove of all things PRL.
"In those days, that's how you found out what other people were doing," he told me. "Now you've already read it months ago on the web."
Physical Review Letters shortened the publication process from months to weeks. But by today's standards, as Chodos points out, and even in its current electronic form, PRL can't compete for speed with the place where scientists share their work months ahead of publication: the online preprint arXiv. And that's because every paper that appears in PRL is vetted by experts; the arXiv is the firehose scientific research, unrefereed and, for that reason, near-instantaneous.
ArXiv started in 1991 as a stew of the latest theoretical high-energy physics work (think the Higgs boson and beyond), but soon broadened to include all the major fields under the physics umbrella, including computational physics, condensed matter, and nuclear physics. Now it includes more exotic realms, such as quantitative finance and quantitative biology. But while the arXiv gets the research out fast, it does so with some caveats. Along with the influx of papers are two dozen "replacements." Because papers appear on the arXiv while still in progress, authors might rewrite a paragraph after posting; or they may report new, game-changing results, or even pull the paper altogether.
What does the physics arXiv mean for the rest of us? Can science journalists use it to get the scoop on a hot piece of research? Not really, it seems. The arXiv is open science, but you have to speak the language, and most science journalists don't. That's where middle-men like the arXiv blog come in. Under the moniker Kentucky FC, the arXiv blogger sips from the firehose, identifying potentially newsworthy papers that sometimes get picked up by science journalists. While originally an independent effort, the blog, like those recruited for the stables of Discover or SEED, garnered notice, and as of March 15 now appears exclusively in MIT's Technology Review.
But there are downsides to the firehose method of spreading around information. While Kentucky acts as an intelligent filter, the arXiv itself gives no indication of a paper's coherence or integrity, much less importance. In fact, one of the founders, Paul Ginsparg, has written several papers (appropriately enough, they're on the arXiv, if you want to read them) about how the arXiv has affected scientific citation, a trackable measure of how ideas are circulate and influence. He made the frightening observation that the arXiv papers cited most aren't necessarily the most important or interesting; rather, they're the ones that happen to make it to the top of a given day's list. This makes intuitive sense to me. As scientists scroll down the page, information overload kicks in, but the first paper they see, when they're looking with fresh eyes, sticks.
This phenomenon reminded me of something I read on Sabine Hossenfelder's blog, BackReaction. Hossenfelder, a theoretical physicists at the exciting Perimeter Institute, raises an eyebrow at the fast-paced circulation of ideas that the internet has made so easy:
In fact, I think we already share way too much too premature information. The reason is that scientists too are only human. If we hear some colleagues talk who are genuinely excited about a topic, chances are we'll get interested. If we have an idea in an early stage and bounce it off a lot of people, it will lose its edges because we'll try to make it fit. If we hear something repeatedly, we are likely to think it's of some relevance. If we know the opinions of other people, in particular people with a higher social status or more experience, we'll try to fit in. That's what humans do. That's why crowds make dumb decisions.

If you're nevertheless interested in exploring the (sometimes dauntingly erudite) world of open science, here are a few related links:
- Open Science project, dedicated to writing and releasing free and Open Source scientific software.
- Public Library of Science, the open-access scientific journal.
-Journal of Visualized Experiments, a "video journal" of experimental methods in biology and medicine. I can't decide if this is awesome or terrifying.
-The Future of Scientific Publishing, an essay by 20-years-and-counting PRL editor Jack Sandweiss

No comments:

Post a Comment