Professional Documents
Culture Documents
Clio and The Economics of QWERTY
Clio and The Economics of QWERTY
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The
American Economic Review.
http://www.jstor.org
By PAUL A. DAVID*
been a conspiracy among the members of the succession was a particularly serious defect.
typewriter oligopoly to suppress an invention When a typebar stuck at or near the print-
which they feared would so increase type- ing point, every succeeding stroke merely
writer efficiency as ultimately to curtail the hammered the same impression onto the
demand for their products? Or perhaps we paper, resulting in a string of repeated letters
should turn instead to the other popular that would be discovered only when the typist
"Devil Theory," and ask if political regu- bothered to raise the carriage to inspect what
lation and interference with the workings of had been printed.
a "free market" has been the cause of ineffi- Urged onward by the bullying optimism of
cient keyboard regimentation? Maybe it's all James Densmore, the promoter-venture capi-
to be blamed on the public school system, talist whom he had taken into the partner-
like everything else that's awry? ship in 1867, Sholes struggled for the next six
You can already sense that these will not years to perfect "the machine." From the
be the most promising lines along which to inventor's trial-and-error rearrangements of
search for an economic understanding of the original model's alphabetical key order-
QWERTY's present dominance. The agents ing, in an effort to reduce the frequency of
engaged in production and purchase deci- typebar clashes, there emerged a four-row,
sions in today's keyboard market are not the upper case keyboard approaching the mod-
prisoners of custom, conspiracy, or state con- ern QWERTY standard. In March 1873,
trol. But while they are, as we now say, Densmore succeeded in placing the manufac-
perfectly "free to choose," their behavior, turing rights for the substantially trans-
nevertheless, is held fast in the grip of events formed Sholes-Glidden "Type Writer" with
long forgotten and shaped by circumstances E. Remington and Sons, the famous arms
in which neither they nor their interests makers. Within the next few months
figured. Like the great men of whom Tolstoy QWERTY's evolution was virtually com-
wrote in War and Peace, "(e) very action of pleted by Remington's mechanics. Their
theirs, that seems to them an act of their own many modifications included some fine-
free will, is in an historical sense not free at tuning of the keyboard design in the course
all, but in bondage to the whole course of of which the "R" wound up in the place
previous history..." (Bk. IX, ch. 1). previously allotted to the period mark "."
This is a short story, however. So it begins Thus were assembled into one row all the
only little more than a century ago, with the letters which a salesman would need to im-
fifty-second man to invent the typewriter. press customers, by rapidly pecking out the
Christopher Latham Sholes was a Milwau- brand name: TYPE WRITER
kee, Wisconsin printer by trade, and a me- Despite this sales gimmick, the early com-
chanical tinkerer by inclination. Helped by mercial fortunes of the machine, with which
his friends, Carlos Glidden and Samuel W. chance had linked QWERTY's destiny re-
Soule, he had built a primitive writing ma- mained terrifyingly precarious. The eco-
chine for which a patent application was nomic downturn of the 1870's was not the
filed in October 1867. Many defects in the best of times in which to launch a novel
working of Sholes' "Type Writer" stood in piece of office equipment costing $125, and
the way of its immediate commercial intro- by 1878, when Remington brought out its
duction. Because the printing point was Improved Model Two (equipped with car-
located underneath the paper carriage, it was riage shift key), the whole enterprise was
quite invisible to the operator. "Non-visibil- teetering on the edge of bankruptcy. Conse-
ity" remained an unfortunate feature of this quently, even though sales began to pick up
and other up-stroke machines long after the pace with the lifting of the depression and
flat paper carriage of the original design had annual typewriter production climbed to
been supplanted by arrangements closely re- 1200 units in 1881, the market position which
sembling the modern continuous roller- QWERTY had acquired during the course
platen. Consequently, the tendency of the of its early career was far from deeply
typebars to clash and jam if struck in rapid entrenched; the entire stock of QWERTY-
elsewhere. (Notice that it was the wartime in a second ball of the same color being
U.S. Navy, not your typical employer, that returned to the urn; the probabilities that
undertook the experiment of retraining typ- balls of specified colors will be added are
ists on the Dvorak keyboard.) Nevertheless therefore increasing (linear) functions of the
the purchase by a potential employer of a proportions in which the respective colors
QWERTY keyboard conveyed a positive are represented within the urn. A recent the-
pecuniary externality to compatibly trained orem due to W. Brian Arthur et al. (1983;
touch typists. To the degree to which this 1985) allows us to say that when a gener-
increased the likelihood that subsequent alized form of such a process (characterized
typists would choose to learn QWERTY, in by unbounded increasing returns) is ex-
preference to another method for which the tended indefinitely, the proportional share of
stock of compatible hardware would not be one of the colors will, with probability one,
so large, the overall user costs of a typewrit- converge to unity.
ing system based upon QWERTY (or any There may be many eligible candidates for
specific keyboard) would tend to decrease as supremacy, and from an ex ante vantage
it gained in acceptance relative to other sys- point we cannot say with corresponding cer-
tems. Essentially symmetrical conditions ob- tainty which among the contending colors
tained in the market for instruction in touch -or rival keyboard arrangements-will be
typing. the one to gain eventual dominance. That
These decreasing cost conditions-or sys- part of the story is likely to be governed by
tem scale economies-had a number of con- "historical accidents," which is to say, by the
sequences, among which undoubtedly the particular sequencing of choices made close
most important was the tendency for the to the beginning of the process. It is there
process of intersystem competition to lead that essentially random, transient factors are
towards de facto standardization through the most likely to exert great leverage, as has
predominance of a single keyboard design. been shown neatly by Arthur's (1983) model
For analytical purposes, the matter can be of the dynamics of technological competition
simplified in the following way: suppose that under increasing returns. Intuition suggests
buyers of typewriters uniformly were without that if choices were made in a forward-look-
inherent preferences concerning keyboards, ing way, rather than myopically on the basis
and cared only about how the stock of touch of comparisons among the currently prevail-
typists was distributed among alternative ing costs of different systems, the final out-
specific keyboard styles. Suppose typists, on come could be influenced strongly by expec-
the other hand, were heterogeneous in their tations. A particular system could triumph
preferences for learning QWERTY-based over rivals merely because the purchasers of
"touch," as opposed to other methods, but the software (and/or the hardware) expected
attentive also to the way the stock of ma- that it would do so. This intuition seems to
chines was distributed according to keyboard be supported by recent formal analyses by
styles. Then imagine the members of this het- Michael Katz and Carl Shapiro (1983), and
erogenous population deciding in random or- Ward Hanson (1984), of markets where
der what kind of typing training to acquire. purchasers of rival products benefit from ex-
It may be seen that, with unbounded de- ternalities conditional upon the size of the
creasing costs of selection, each stochastic compatible system or "network" with which
decision in favor of QWERTY would raise they thereby become joined. Although the
the probability (but not guarantee) that the initial lead acquired by QWERTY through
next selector would favor QWERTY. From its association with the Remington was
the viewpoint of the formal theory of sto- quantitatively very slender, when magnified
chastic processes, what we are looking at by expectations it may well have been quite
now is equivalent to a generalized "Polya sufficient to guarantee that the industry even-
urn scheme." In a simple scheme of that tually would lock in to a de facto QWERTY
kind, an urn containing balls of various col- standard.
ors is sampled with replacement, and every The occurrence of this "lock in" as early
drawing of a ball of a specified color results as the mid-1890's does appear to have owed
something also to the high costs of software (1915), on the problem of Britain's under-
"conversion" and the resulting quasi-irre- sized railway wagons and "the penalties of
versibility of investments in specific touch- taking the lead" (see pp. 126-27); they may
typing skills. Thus, as far as keyboard con- be painfully familiar to students who have
version costs were concerned, an important been obliged to assimilate the details of de-
asymmetry had appeared between the soft- servedly less-renowned scribblings (see my
ware and the hardware components of the 1971, 1975 studies) about the obstacles which
evolving system: the costs of typewriter ridge-and-furrow placed in the path of British
software conversion were going up, whereas farm mechanization, and the influence of
the costs of typewriter hardware conversion remote events in nineteenth-century U.S. fac-
were coming down. While the novel, non- tor price history upon the subsequently
typebar technologies developed during the emerging bias towards Hicks' labor-saving
1880's were freeing the keyboard from tech- improvements in the production technology
nical bondage to QWERTY, typewriter of certain branches of manufacturing.
makers were by the same token freed from I believe there are many more QWERTY
fixed-cost bondage to any particular key- worlds lying out there in the past, on the
board arrangement. Non-QWERTY type- very edges of the modern economic analyst's
writer manufacturers seeking to expand tidy universe; worlds we do not yet fully
market share could cheaply switch to achieve perceive or understand, but whose influence,
compatibility with the already existing stock like that of dark stars, extends nonetheless to
of QWERTY-programmed typists, who could shape the visible orbits of our contemporary
not. This, then, was a situation in which the economic affairs. Most of the time I feel sure
precise details of timing in the developmen- that the absorbing delights and quiet terrors
tal sequence had made it privately profitable of exploring QWERTY worlds will suffice to
in the short run to adapt machines to the draw adventurous economists into the sys-
habits of men (or to women, as was increas- tematic study of essentially historical dy-
ingly the case) rather than the other way namic processes, and so will seduce them
around. And things have been that way ever into the ways of economic history, and a
since. better grasp of their subject.
In place of a moral, I want to leave you Arthur, W. Brian, " On Competing Technol-
with a message of faith and qualified hope. ogies and Historical Small Events: The
The story of QWERTY is a rather intriguing Dynamics of Choice Under Increasing Re-
one for economists. Despite the presence of turns," Technological Innovation Program
the sort of externalities that standard static Workshop Paper, Department of Econom-
analysis tells us would interfere with the ics, Stanford University, November 1983.
achievement of the socially optimal degree of Arthur,W. Brian,Ermoliev,YuriM. and Kaniov-
system compatibility, competition in the ab- ski, YuriM., " On Generalized Urn Schemes
sence of perfect futures markets drove the of the Polya Kind," Kibernetika, No. 1,
industry prematurely into standardization on 1983, 19, 49-56 (translated from the Rus-
the wrong system -where decentralized deci- sian in Cybernetics, 1983, 19, 61-71).
sion making subsequently has sufficed to hold I , and _ , "Strong Laws
it. Outcomes of this kind are not so exotic. for a Class of Path-Dependent Urn Pro-
For such things to happen seems only too cesses," in Proceedings of the International
possible in the presence of strong' technical Conference on Stochastic Optimization,
interrelatedness, scale economies, and irre- Kiev, Munich: Springer-Verlag, 1985.
versibilities due to learning and habituation. David, Paul A., "The Landscape and the Ma-
They come as no surprise to readers pre- chine: Technical Interrelatedness, Land
pared by Thorstein Veblen's classic passages Tenure and the Mechanization of the Corn
in Germany and the Industrial Revolution Harvest in Victorian Britain," in D. N.