In his wonderful book The Kindly Inquisitors, Jonathan Rauch notes that our Enlightenment civilization stands on three pillars: democracy, which is how we determine who gets to wield legitimate coercive force; capitalism, which is how we determine who gets what; and what Rauch calls liberal science, which is how we determine what is true.
In Rauch's conception, liberal science embodies the principle that the "checking of each by each through public criticism is the only legitimate way to decide who is right." Liberal science encompasses everything from the most biased activist pamphlet to peer-reviewed scientific journals. The peer review of scientific articles is our most stringent formal process for determining whether or not a claim is true. The fundamental concept of peer review is an assessment of technical or scientific merit by individuals with sufficient technical competence who do not have conflicts of interest.
Formal scientific peer review was begun in the eighteenth century by the Royal Societies of London, Edinburgh, and Paris, and emerged nearly at the same time as our modern democratic and economic institutions. Prior to the rise of modern science, scholars, alchemists, and metalsmiths kept their formulas and discoveries secret—and thus beyond criticism or testing. Scientific and technological progress, needless to say, was slow.
So how does peer review work? Typically, what happens is that a researcher sends in a manuscript to a journal describing her recent work. The editor of the journal makes an initial judgment of whether the work is of interest or not. If it is, the editor sends out the manuscript for a review by two or three anonymous experts in the area under consideration. The reviewers may advise that the manuscript be rejected, ask for improvements and corrections before acceptance, or recommend that it be accepted as is. The reviewers are looking to see if the findings and arguments presented in the manuscript are plausible given what they know of the field. Peer review is designed to catch errors, not fraud. Consequently, scientific fraud is often caught only after peer review publication.
Consider the case of Jan-Hendrik Schon, a physicist at Bell Labs who published studies purporting to show that he had created molecular scale transistors in Nature and Science between 1998 and 2001. The fact that Schon had faked data came to light after other researchers reported that they could not replicate his findings. Others noticed that he had published an identical graph in two different articles—the only problem was the graph was supposed to illustrate two different phenomena. Schon was dismissed in 2002.
More recently, MIT fired immunologist Luk Van Parijs in October, 2005 for fabricating data in his studies of short interfering RNA in mice. Members of his lab reported him for scientific misconduct when they couldn't find or identify some of his data. Suspicions about Van Parijs' work was raised when other researchers noted an unwarranted similarity between graphs purporting to show immunological responses from different strains of lab mice.
Which brings us to the latest scientific scandal: the fabrication of data by Korean stem cell researcher Hwang Woo Suk. In May, Science published a breakthrough paper by Hwang in which he claimed to have created 11 patient-specific stem cell lines. As amazing as this achievement would have been in itself, Hwang also claimed to have been able to do it using relatively few human eggs per stem cell line. This advance in efficiency seemed to be opening the door to therapeutic cloning as a way to produce transplants perfectly matched genetically to individual patients. Hwang got caught when inconsistencies showed up in the data he submitted. First, the photos purporting to show 11 different stem cell colonies looked way too similar. Also, graphs of DNA fingerprint tests that were supposed to show that the stem cell colonies were genetic matches for their donors were too perfect.
Of course peer review, like democracy and economic institutions such as corporations, is not perfect. The process by which scientific publications are reviewed can doubtless be improved. One proposal is that scientists place pre-prints of their articles online, where their colleagues and competitors can critique them before they are submitted to journals for publication. Others have even argued that peer review doesn't actually improve scientific reports and that it could be eliminated. Peer review is only one, though important, part of the larger system of liberal science that eventually enables us to determine what is true. The gold standard for what is true is not peer review, but experimental replication and the extension of scientific results. The last 200 years of technological progress is powerful evidence that scientific fraud is rare, and the Enlightenment institutions of free speech and mutual criticism that form the basis of liberal science remain strong.
Ronald Bailey is Reason's science correspondent. His book Liberation Biology: The Scientific and Moral Case for the Biotech Revolution is now available from Prometheus Books.