Pride and Prejudice

Competition and feedback are the solutions to a gatekeeper's blind spot.


A FEW YEARS AGO, three Canadian scientists wrote an article in New Scientist magazine criticizing peer review. The article did a good job of revealing the limits of gatekeepers -- but not in the way it intended.

The theme of the piece was that anonymous peer review survives not because it serves its purposes better than any alternative system but because, once established, it has continued through inertia. It is, the authors said, like the standard QWERTY keyboard, whose "characters . . . were deliberately set to be inconvenient, thus ensuring slower typing speeds" to avoid jamming the keys. "All recent attempts to create a mass market for more efficient keyboards -- the Dvorak keyboard, for example, on which typists can achieve touch-typing speeds about 40 per cent faster than on QWERTY keyboards -- were stonewalled."

The problem with this example, which the authors took from Stephen Jay Gould's Bully for Brontosaurus, is that it is utter bunk. The key-jamming story is simply a myth. And the claims of the Dvorak keyboards' enormous superiority are self-interested exaggeration. The major "study" on which they are based was done under the direction of none other than August Dvorak, the keyboard's inventor.

As an example of irrational technological "lock in," QWERTY has been discredited since 1990. That's when economists Stan Liebowitz and Stephen E. Margolis published the results of their careful detective work in the Journal of Law and Economics. Yet the QWERTY legend still passed the scrutiny of the New Scientist editors and of Gould's editors at W.W. Norton. It even continues to turn up in refereed economics literature.

Last June I argued in this space that Marvell's law -- life is limited -- makes "the dreaded 'gatekeepers'" essential. No single individual can possibly sort through all the world's information, and technofixes don't promise to solve that problem anytime soon. We need editors. But human editors have blind spots. They may be ignorant or they may ignore stories that don't fit the scripts they want to follow, perpetuating stories that fit. The QWERTY legend survives because it is a simple tale that says inertia overwhelms technical excellence. It captures that excellence in a single quality dimension -- typing speed -- with no messy tradeoffs.

Its recent popularity has nothing to do with typewriters and everything to do with Microsoft. By telling "the fable of the keys," economic theorists and antitrust lawyers can ignore the complexities of how Windows became the dominant operating system, a story that includes many dimensions of quality and many missteps by competitors. They can argue that decentralized decision making leads to bad standards and that someone (like them) ought to be in charge.

Marvell's law, then, presents us with an inescapable dilemma: We must trust other people to help us sort through information. But gatekeepers have prejudices that affect the information they present as true.

They also have friends, professional networks, and geographic orientations that influence their filters. Slate's notion of which magazines are important enough to review is clearly formed in New York and Washington. Wired reflects Bay Area sources centered on the Global Business Network. Stories covered in the Atlanta Constitution can end up on Atlanta-based CNN even if no one else carries them.

SINCE I'M IN THE sorting business myself, I'm well aware that Marvell's law makes no exceptions for editors. Because life is limited, we gatekeepers follow "mattering maps" that tell us who or what is important, reliable, or interesting -- and sometimes those maps lead us astray.

The only solution (and it is a partial one) lies in competition and feedback. More sources of information push Marvell's law to the limit. But they also make it more likely that important criticism will leak out, if only because competing gatekeepers, with different prejudices and different scripts, think it deserves to be heard. In a world of many voices, nobody has a monopoly on the stories that get told. Such diversity means more falsehoods, perhaps, but it also means a much better chance at truth.