QWERTY in the Long Run
| Dick Langlois |
The new issue of Industrial and Corporate Change has an article by Andreas Reinstaller and Werner Hölzl called “Big Causes and Small Events: QWERTY and the Mechanization of Office Work.” Although it’s an interesting paper in many respects, I think it fails in its avowed aim to defend Paul David against the attack of Liebowitz and Margolis. Mostly, they don’t get L&M right (and explicitly get them wrong in footnote 1). The issue is whether the QWERTY keyboard is an example of what L&M call “third-degree” path dependency, that, is path dependency leading to an outcome that is both regrettable ex post and would somehow have been remediable ex ante. The criterion of “remediable” to R&H seems to be whether contemporaries “knew about” superior alternatives. That’s not quite right, of course: the real issue is whether any alternative institutional structure could have done a better job of choosing a standard under the conditions of knowledge at the time. Their only example is the existence of a French “Ideal” keyboard layout (which some people “knew about”) that was swept aside by the tidal wave of the American QWERTY standard (and became AZERTY in France). But they have no evidence about how much better this keyboard was — or if it was better at all. In footnote 1 they cite Donald Norman’s interesting book on design to the effect that the Dvorak keyboard is 10 per cent faster than QWERTY. But (A) Norman’s point in the book is how insignificant this difference is and (B) that doesn’t demonstrate third-degree path dependency, since no one “knew about” the Dvorak keyboard until Dvorak invented it (an extremely laborious process, according to Norman).
Again, I don’t want to be too hard on R&H: I think there’s a lot that’s interesting in the paper, especially the discussion of the mechanization of office work. What really struck me in this context, however, is how irrelevant, or at least dated, the QWERTY saga is. And I say this not for the usual reason: that computers now allow us to have any keyboard layout we like. Rather, what struck me is that the production of documents has long since become demechanized, making even more-than-nominal differences in typing speed irrelevant. Since we now all (or almost all) compose right on the computer, and never send our documents out to the typing pool, manuscript production has become a craft again. What is slowing us down is how quickly we think of something to say, not how fast we can type. And I doubt that, fifties nostalgia notwithstanding, we are unlikely to see the return of the typing pool anytime soon. So, from a historical perspective, QWERTY will have been technically inefficient (though not therefore economically inefficient) only for that brief historical period between the invention of Dvorak and the coming of the personal computer.
The same issue of ICC also has a paper by Ashish Arora and coauthors that’s worth a look.