QWERTY – Kay’s analysis of a non sub-optimal standard

QWERTY and the search for optimality

By: Neil M Kay (University of Strathclyde Business School, Department of Economics)

Abstract: This paper shows how one of the developers of QWERTY continued to use the trade secret that underlay its development to seek further efficiency improvements after its introduction. It provides further evidence that this was the principle used to design QWERTY in the first place and adds further weight to arguments that QWERTY itself was a consequence of creative design and an integral part of a highly efficient system rather than an accident of history. This further serves to raise questions over QWERTY’s forced servitude as “paradigm case” of inferior standard in the path dependence literature. The paper also shows how complementarities in forms of intellectual property rights protection played integral roles in the development of QWERTY and the search for improvements on it, and also helped effectively conceal the source of the efficiency advantages that QWERTY helped deliver

URL: http://econpapers.repec.org/paper/strwpaper/1324.htm

Circulated by NEP-HIS on: 2013-11-22

Review by Anthony Gandy


The paper reviewed here is another instalment in Neil Kay’s systematic exploration of the history of the QWERTY typewriter (see also Kay 2013a and 2013b). Through his work Kay questions the QWERTY typewriter as the primary example of how inferior design can become establish and create a path of dependency which will always be sub-optimal. The case Kay makes both in this paper and in earlier ones is that the design was in fact near-optimal for the state of knowledge at the time. In this paper he produces further knowledge that the designer made later efforts to refine the design using principles which show the designers’ advanced understanding of the concepts needed to produce an efficient design given the outcomes required (later definitions of efficiency were based on a different concepts and on different technology capabilities).

Christopher Latham Sholes (1819-1890)

Christopher Latham Sholes (1819-1890)

Kay studies the work of designer Christopher Latham Sholes and how he refined his own design of typewriter designs. In 1872 he and his associates created the familiar QWERTY layout which is so familiar in many Western and English language countries. The design was bought by Remington in 1873 in a deal arranged by Sholes’ business partner James Densmore who also retained a financial stake in the QWERTY keyboard after Remington bought it. Kay shows the deliberate design concepts which lay at the heart of the QWERTY keyboard which then refutes some earlier works (such as David 1985) which claimed the design process for QWERTY and the eventual lock-in created was generated by random events and that the design was effectively accidental and a sub-optimal standard. The path dependency concept based on network externalities (typist unwilling to face conversion to another standard even if it were more efficient) maybe fair, but the design was not at that time accidental or sub-optimal. Kay shows by inference that Sholes work was more deliberate than may have been understood, primarily because Sholes was protecting his trade secret which was his understanding of how to avoid frequent letter pairs from occurring and jamming the early typewriter technology. In this paper he explores two patents issued to Sholes after his death and finds them as evidence that Sholes continued to improve his basic understanding, but continued to protect his knowledge.

Remington No 7 – an improved version of the Sholes design

Remington No 7 – an improved version of the Sholes design. Source: http://machinesoflovinggrace.com/large/remstandard703.jpg

Like Kay’s earlier work this paper is highly detailed, looking at the available evidence, and analyzing the likely approaches used. It considers how, after Sholes sold his rights to the QWERTY typewriter, he continued to use his principles to develop an improved “perfect” typewriter, and how this was evidence that he knew more about the underlying principles than others have given credit to.
The real challenge at the time was typebars jamming. Using a keyboard layout of ABCDE, it was known that too often there would be pairs of letters used which would jam the typewriter. The goal was to reduce these pairs to reduce jamming risk.
Kay shows (2013a) that when typing the book Life of the Mississippi by Mark Twain, the Sholes Qwerty typewriter would have created 146 events where letters next to each other in the “typebasket” would have occurred and jammed. Others have argued (David 1985) that later designs of typewriter such, as the 1936 design by August Dvorak, were more efficient because they were faster. However, Kay shows that in the Life of the Mississippi test, the Dvorak design would create 2358 conflicting pairs. Kay points out that the comparison needs to be in era context. While later development was based on ten figure typing and efficiency focused on speed out output, Sholes’ era was one where two fingers of each hand was seen as optimal (and the reviewer here is certainly in this camp!!!) and the real challenge was one of the mechanical jamming or letter pairs.

This paper extends Kay’s research to the period after the Remington company purchased Sholes’ design and when he then went on to try to develop the “perfect” typewrite. Frankly, these are detailed and intricate discussions. Kay studies two patents issued after Sholes’ death to show he was working further with infrequency principles for his improved QV typewriter, the ultimate version of which would have reduced the incidents of jamming pairs by 97.7% compared to the QWERTY design. It is the evidence presented by the “jigsaw” of the two patents supporting the QV typewriter which is the core the Kay paper, combining them to show Sholes had an advanced knowledge of the processes required to reduce conflict between letters. Kay believes Sholes split the patents into two separate ones so as to protect his trade secret as to how to prevent conflict through infrequency principles, a concept which in itself could not be protected under copyright.

Brief discussion:

While the paper is maybe too detailed for most, and on occasion difficult to read without prior understanding of the issues, it serves as an interesting discussion paper on both how intellectual property was protect, even when the underlying principles could not be copyrighted and a lesson not to infer the efficacy of a technological solutions in the light of later technological capability. While the case for path dependency has been well made over the years and clearly make a great deal of sense, paths which have been founded on technology that is at its origin less than optimal are rare, maybe this is the reason the QWERTY keyboard has been so popular as an example as the thesis that its development was less than deliberate suggests it may have always been less than optimal. Kay’s forensic disputes this very. Kay generates evidence of a deliberate process by Sholes and his associates and while clearly the design could be refined (as Sholes tried) or replaced (as advocates of the Dvorak would argue for), it was nevertheless close to the optimal design at the time of development.

For those interested in what has been a very long running debate about QWERTY, optimality and lock-in, a more in depth review of Kay’s work and the prior debate has been produced by Jean-Philippe Vergne (2013) in the journal Research Policy which certainly seems to have dedicated its existence to discussion the QWERTY typewriter!!!

Additional References

Kay’s other QWERTY Papers

Kay, N. M. (2013a) “Rerun the tape of history and QWERTY always wins” Research Policy, 42:1175-85.

Kay, N. M. (2013b) “Lock in, path dependence, and the internationalization of QWERTY. Strathclyde Discussion Papers in Economics, no 13-10, http://www.strath.ac.uk/media/departments/economics/researchdiscussionpapers/2013/13-10FINAL.pdf (accessed 14 April 2014).

Jean-Philippe Vergne’s excellent review can be found in Research Policy:

Vergne, J. P. (2013) “QWERTY Is Dead, Long Live Path Dependence!” Research Policy, 42: 1191-1194. See also http://www.academia.edu/3495369/QWERTY_Is_Dead_Long_Live_Path_Dependence_

5 thoughts on “QWERTY – Kay’s analysis of a non sub-optimal standard

  1. Joshua Rosenblooom

    It is not clear to me in what respects Kay’s argument refutes or significantly alters the account provided by Paul David in his 1986 work on this subject. As David notes, Sholes initial motivation was to resolve the problem of frequent jamming of the type bar and he worked for a long time to find a keyboard arrangement that would do so. In this sense the QWERTY arrangement was optimal given the mechanical constraints of the first typewriters.
    However, the optimality of this arrangement was reduced by redesign of the type bar mechanism in the 1880s, which made the QWERTY arrangement unnecessary. The point of the lock – in argument is that once the technical constraint was removed, further optimization was not pursued. Not that the initial design was not purposive or seeking to optimize the device given the existing constraints.

  2. Neil Kay

    I am grateful to Anthony Gandy for a thoughtful and detailed coverage of my Working Paper and also the invitation of the editor to respond to Joshua Rosenbloom’s comments on it.
    First Joshua says
    “As David notes, Sholes initial motivation was to resolve the problem of frequent jamming of the type bar and he worked for a long time to find a keyboard arrangement that would do so. In this sense the QWERTY arrangement was optimal given the mechanical constraints of the first typewriters”.
    But that is not what David (1985) actually said. David instead claimed Sholes made “trial-and-error rearrangements” in “an effort to reduce the frequency of typebar clashes” and that QWERTY was an example of “historical accidents” or “happenings dominated by chance elements rather than systematic forces”.
    So David never claimed QWERTY was optimal in any sense (he only suggested that Sholes tried to reduce jamming problems). Nor did he suggest that QWERTY was the outcome of rational careful design (or was “purposive” in Joshua’s words), on the contrary David saw it as a product of trial and error, chance, and was itself a historical accident.
    But instead QWERTY was optimal (or near-optimal)on the grounds that really mattered at the time and was the product of creative and rational thinking. These points were made in my 2013 article which was originally submitted to Research Policy as a normal peer-reviewed article but which the editors generously decided to publish with accompanying commentaries on it by Brian Arthur, Steve Margolis and Jean Philippe Vergne and a reply from me, see here:
    The working paper which Anthony Gandy reviews here showed further evidence of the rational careful (and non-random) design using what could be described as the Sholes’ infrequency principle and which had underlain QWERTY, as does as a separate Working Paper on the internationalization of QWERTY, see here
    “Lock in, path dependence and the internationalization of QWERTY” University of Strathclyde Economics Department Working Paper, available at http://www.strath.ac.uk/media/departments/economics/researchdiscussionpapers/2013/13-10FINAL.pdf
    Joshua also says “However, the optimality of this arrangement was reduced by redesign of the type bar mechanism in the 1880s, which made the QWERTY arrangement unnecessary. The point of the lock – in argument is that once the technical constraint was removed, further optimization was not pursued. Not that the initial design was not purposive or seeking to optimize the device given the existing constraints”
    Joshua is presumably referring to the introduction of front-strike or front-stroke technology closer to 1900, not the 1880’s as he suggests. However, apart from that he misses the point of the article by Paul David that he cites here, David goes on to say. “competition in the absence of perfect futures markets drove the industry prematurely into standardization on the wrong system”, implying that Dvorak was the right system. But there would have been no advantage to be made from theoretical gains in typing speed (and user/format compatibility) if the device just kept on jamming (because of problems in format/device compatibility).
    The issue as to whether Dvorak was somehow superior in efficiency terms to QWERTY later in the Twentieth Century can also be looked at as a separate but related issue. These claims have been exhaustively analyzed and rejected by Liebowitz and Margolis (1990), and in another Working Paper i argued also that there was no sound basis for the claims made on behalf of Dvorak, the arguments here were to some extent different from, but broadly consistent with, Liebowitz and Margolis, see
    “The QWERTY Problem” (presented DRUID Barcelona, 2013, available online at http://druid8.sit.aau.dk/acc_papers/rluar3kk4i5t4gelr7j36gx9hma5.pdf
    Finally I would like to add a personal note. As a self-declared left-leaning economist who generally believes that markets often fail and need careful regulation to make them work effectively, the results and my own conclusions in this area surprised me. Maybe markets actually work better, and engineers are in fact cleverer, than we generally give them credit for.
    Neil Kay
    Edinburgh Business School, Heriot-Watt University and Institute for Business Innovation, Berkeley

    David, P. A. (1985) Clio and the Economics of QWERTY, The American Economic Review, 75, No. 2, (Papers and Proceedings), pp. 332-337.
    Liebowitz, S. J., and Margolis, S. E. (1990) The fable of the keys; Journal of Law and Economics, 33, 1-25.

    1. Joshua Rosenblooom

      In his response to my comments on his original blog post Neil makes two points. Neither of which strikes me as directly relevant.

      Neil’s first point concerns Shole’s motivation in producing the QWERTY keyboard arrangement. This response conflates two distinct considerations: (a) Shole’s motivation, and (b) his method. The fact that he proceeded by trial and error is largely irrelevant to the solution he was searching for. How it got to his goal may not have been the shortest path, but it was a viable search method.

      I don’t think it is a stretch or a misinterpretation of David’s original article to infer that Sholes search (however indirect) was motivated by a desire to solve the problem of jamming of the type bars that plagued early typewriters. To the extent that this was initially the binding constraint on typing speed, then resolving this problem is reasonably seen as a form of (constrained) optimization.

      Neil’s second objection raises the historical evidence that Leibowitz and Margolis have presented to demonstrate that QWERTY was in fact not markedly inferior to the Dvorak keyboard layout. Certainly if one accepts their evidence, then much of David’s argument loses its force. I don’t disagree, and I suspect that the QWERTY keyboard case is probably not the strongest argument for the phenomenon of path-dependence available.

      So I am still left with the question I posed at the outset, which is what exactly is the novel component of Neil’s analysis. In part my confusion may arise from the way Grandy presents the summary of Neil’s work. Having gone back to Neil’s working paper I can see that he has conducted an interesting analysis of typewriter design that augments and extends the argument that QWERTY was actually a very good (perhaps even optimal) keyboard layout, which would reinforce Leibowitz and Margolis. And he has considered how this played out in international diffusion, which is an interesting extension of the story of which I was not aware.

      These are valuable additions to our understanding of the process, but they strike me as more incremental than revolutionary. And I would contend that the question of Shole’s motivation as well as his methods in achieving his goal is largely irrelevant to this discussion.

  3. Neil Kay

    I think Joshua makes some good points – and I thank him for the generous comments on my work.

    We agree on a lot especially his conclusion that “I suspect that the QWERTY keyboard case is probably not the strongest argument for the phenomenon of path-dependence available”.

    But that still leaves what I could summarise as Joshua’s “so what?” question – what is the significance of all this?

    If it has significance it lies not in what I claim but in what others claim. David in his 1985 paper that started all this claimed that QWERTY was the “wrong” system and that by implication Dvorak would have been the better system.

    Since then QWERTY has been treated as the paradigm case of an inefficient standard. Brian Arthur in his commentary on my 2013 paper noted “ …QWERTY, as a standard—or better as an example of what the market has served us up in the long evolution of one particular technology—has become in economics a focal point, a rallying point for a larger issue: whether the market can lock us into an inferior standard. And this itself is part of a still larger issue: whether the free markets of capitalist economies can drive us into inferior outcomes.” Arthur (2013).

    And the Nobel Prize winning economist Paul Krugman in his introductory economics text with Robin Wells defined in their glossary the “QWERTY problem: an inferior industry standard that has prevailed possibly because of historical accident” (p. G-12)”. Krugman and Wells’ (2006) advise their beginning economics students; “Government can play a useful role both in helping an industry establish a standard and helping it avoid getting trapped in an inferior standard known as the QWERTY problem” (p.536) and also “in principle government intervention might be useful in moving an industry to a superior standard” (p.534) .

    But if – as I think both Joshua and I have been inching towards agreeing – there is no solid empirical evidence that Dvorak was the “superior” system, then that really questions QWERTY’s assigned status as an inferior system. If Dvorak was not superior to QWERTY, what system was? And if QWERTY is not a good example of “the QWERTY problem”, what is? And if we cannot find convincing examples of “the QWERTY problem”, what implications does this have for the potential roles of government in setting or influencing technical standards?

    So I think that if there is significance in the analysis here it lies not so much for what I claim but in terms of the empirical evidence produced that can raise legitimate questions about what others have claimed – and have based their policy prescriptions around in some cases. I have to admit that some of this makes me feel uncomfortable – for example, on the issue of government intervention, if there is one conclusion that would seem reasonable about the financial markets debacle of recent years, it is that lack of appropriate government intervention and regulation can be a bad thing, at least in that context. But I still think that the questions raised or implied by my papers here are good questions, and sometimes perhaps a good question can be better and more significant than a bad answer.

    Arthur, W. B. (2013) Comment on Neil Kay’s paper: Rerun the Tape of History and QWERTY Always Wins. Research Policy, 42, 1186-87.
    David, P. A. (1985) Clio and the economics of QWERTY; American Economic Review, 75, 332-37.
    Krugman P., and Wells R. (2006) Economics; New York, Worth Publishers.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.