QWERTY forever: path dependency
« previous post |
The QWERTY Keyboard Will Never Die. Where Did the 150-Year-Old Design Come From?
The invention’s true origin story has long been the subject of debate. Some argue it was created to prevent typewriter jams, while others insist it’s linked to the telegraph
Jimmy Stamp; Updated by Ellen Wexler (Updated: February 25, 2025 | Originally Published: May 3, 2013) includes embedded 3:35 video and several interesting historical photographs
Those who have learned to touch type most likely have wondered about the illogical, unalphabetical arrangement of the letters on the keyboard. But we have learned to live with it, and some of us have become highly proficient at it, while others spend their whole lives hunting and pecking for the desired letters.
A few years after the iPhone’s debut, an innovative new keyboard system started making headlines. Known as KALQ, the split-screen design was created specifically for thumb-typing on smartphones and tablets. It was billed as a more efficient alternative to the ubiquitous QWERTY keyboard, named for the first six letters in the top row of keys.
How did this strange order of the letters — QWERTY — come about?
In the 1860s, a politician, printer, newspaper man and amateur inventor in Milwaukee by the name of Christopher Latham Sholes spent his free time developing various machines to make his businesses more efficient. One such invention was an early typewriter, which he and several of his colleagues patented in 1868. Their keyboard resembled a piano and was built with an alphabetical arrangement of about two dozen keys. The team surely assumed it would be the most efficient arrangement. After all, anyone who used the keyboard would know immediately where to find each letter. Hunting would be reduced; pecking would be increased. Why change things? This is where the origin of QWERTY gets a little foggy.
One popular theory states that Sholes had to redesign the keyboard in response to the mechanical failings of early typewriters, which were slightly different from the models most often seen in thrift stores and flea markets. The type bars connecting the keys and letter plates hung beneath the paper. If a user quickly typed a succession of letters whose type bars were near each other, the delicate machinery would jam. As the story goes, Sholes redesigned the arrangement to separate the most common sequences of letters, like “th” or “he.”
In theory, then, the QWERTY system should maximize the separation of many common letter pairings. However, the “e” and “r” keys are right next to each other, even though “er” is the fourth most common letter pairing in the English language. One of Sholes’ early prototypes addressed this problem—the “r” key is swapped with the period key—though that design appears to have been scrapped. If it had been put into production, this article would have been about the QWE.TY keyboard.
Thus, the theory that the QWERTY system was designed to reduce mechanical error is not conclusive.
In a 2011 paper, Koichi Yasuoka and Motoko Yasuoka, researchers at Kyoto University, tracked the evolution of the typewriter keyboard alongside a record of its early professional users. They concluded that the mechanics of the typewriter didn’t influence the keyboard design. Instead, the QWERTY system emerged in response to one group of early users: telegraph operators who needed to quickly transcribe messages. These operators found the alphabetical arrangement to be confusing and inefficient for translating Morse code.* The researchers write:
[American Morse] code represents Z as ‘… .’ which is often confused with the digram SE, more frequently used than Z. Sometimes Morse receivers in United States cannot determine whether Z or SE is applicable, especially in the first letter(s) of a word, before they receive following letters. Thus S ought to be placed near by both Z and E on the keyboard for Morse receivers to type them quickly.
*I feel duty bound to emphasize how important Samuel F. B. Morse (1791-1872) was in the history of American science and art. Not only was he an outstanding portrait painter, he was also the co-inventor of a single-wire telegraph system and co-developer of Morse code, which I had to learn as a boy scout.
The researchers suggested that the typewriter keyboard evolved over several years as a direct result of input provided by telegraph operators. In other words, the typist came before the keyboard. They also cite the Morse history to argue against the theory that Sholes wanted to stop his machine from jamming by rearranging the keys with the specific intent to slow down typists:
The speed of Morse receiver should be equal to the Morse sender, of course. If Sholes really arranged the keyboard to slow down the operator, the operator became unable to catch up the Morse sender. We don’t believe that Sholes had such a nonsense intention during his development of [the] typewriter.
Regardless of why he developed it, Sholes himself wasn’t fully convinced that QWERTY was the best system. Although he sold his designs to Remington early on, he continued to invent improvements and alternatives to the typewriter for the rest of his life, including several keyboard layouts that he determined to be more efficient. He filed a patent application for a XPMCH keyboard in 1889, a year before he died.
/https://tf-cmsv2-smithsonianmag-media.s3.amazonaws.com/filer/17/1a/171af8aa-1178-477a-b4a9-e5c1bb174148/sholes-patent-1889.jpg)
Sholes filed this patent in 1889, a year before his death. Google Patents
VHM: Note that all the vowels are in the middle row and easily accessible to the right hand. And look where Q and Z are. I think that Sholes was on to something.
But the biggest rival to challenge QWERTY was the Dvorak Simplified Keyboard, developed by August Dvorak in the 1930s. The Dvorak increased the number of words that can be typed using the “home” row of keys where your fingers naturally rest. These are the letters you most often type when you’re hitting keys at random: asjdfkal. sdfjkl. asdfjklasdfjk. Some research showed that Dvorak users reported improved speed and accuracy—though much of that testing was funded by Dvorak—and other research has since suggested that Dvorak isn’t more efficient. But efficiency hardly matters, as it was already too late for a new system to gain a foothold by the 1930s. While Dvorak certainly had its champions, it never gained enough of a following to overthrow King QWERTY.
When the first generation of computer keyboards emerged, there was no longer any mechanical reason to use the system—computers didn’t get jammed. But, of course, there was the minor fact that millions of people had already learned to type on QWERTY keyboards. It had become truly ubiquitous in countries that used the Latin alphabet.
When a design depends on a previous innovation too entrenched in the cultural zeitgeist to change, it’s known as a path dependency. This is also why the KALQ proposal is so interesting. It attempts to break from the tyranny of Sholes, whose QWERTY system makes even less sense on the virtual keyboards of tablets and smartphones than it does on computer keyboards.
Is the KALQ system any different? In some ways, the answer is obviously yes. It was designed around a very specific, very modern behavior—typing with thumbs. Like the 19th-century telegraph operators, the users determined the structure of the KALQ keyboard. But it could still be argued that the KALQ system, or any similar system that may be developed in the future, is also a product of path dependency. Because no matter how the letters are arranged, the basic notion of individually separated letters distributed across a grid dates back to Sholes and his colleagues in Milwaukee. If you gave an iPad to someone who had never used a keyboard and told them to develop a writing system, chances are they would eventually invent something more intuitive. Perhaps a gesture-based system that borrows from shorthand? Or some sort of swipe-to-type interface?
Of all the miraculous things about typing with the Latin alphabet, here are two that boggle my mind:
1. people can type rapidly on those little glass plates with just two thumbs, even those with fat, pudgy fingers
2. people like my son can seemingly effortlessly and nonchalantly swype type, their finger gliding over the surface of that little glass plate.
In response to those feats, I introduce a Mandarin complement that I will use in a forthcoming post about Mandarin disyllabism: bùdéliǎo 不得了 ("awesome")
Selected readings
- "Another chapter in the history of the Chinese typewriter" (8/14/21)
- "Zhou Youguang, Father of Pinyin" (1/14/14)
- "Zhou Youguang, 109 and going strong" (1/13/15)
- "Zhou Youguang 1906-2017" (1/14/17)
- "Chinese Typewriter" (6/30/09)
- "Chinese typewriter, part 2" (4/17/11) — featuring the old Chinese typewriter at the University of Kansas
- "Chinese typewriter redux" (11/5/16) — this and the preceding two posts all talk about my favorite M. C. Hammer performance, "Chinese Typewriter Dance" ("U Can't Touch This")
- "Chinese character inputting" (10/17/15)
- "Sinographic inputting: 'it's nothing' — not" (2/22/21) — with a bibliography of more than two dozen items
- "Typing by voice recognition" (8/22/20)
- "Swype and Voice Recognition for mobile device inputting" (1/22/14) — esp. ¶¶ 3-5
- "Character amnesia yet again: game (almost) over" (4/28/22) — with long, long bibliography
- "Whence cometh 'Vicotr'?"(4/5/25)
[Thanks to John Rohsenow]
Philip Taylor said,
April 6, 2025 @ 8:08 am
"[Some] people can type rapidly on those little glass plates with just two thumbs, even those with fat, pudgy fingers". And others cannot. Even on a reasonable sized screen (a Cubot X9, inherited from my wife), about 1/3 of my attempts result in my "typing" the letter to the left of the one intended. And I don’t use my thumbs (nor do I understand why others do) — I use my left index finger while supporting the telephone in my right hand. And if I have to send an SMS (a task I abhor), I infinitely prefer my Archos F24 Power, with real mechanical keys.
J.M.G.N said,
April 6, 2025 @ 8:28 am
https://en.wiktionary.org/wiki/swype
Victor Mair said,
April 6, 2025 @ 8:58 am
I like how "swype" pairs with "type".
Robert Coren said,
April 6, 2025 @ 9:13 am
[The QWERTY keyboard] had become truly ubiquitous in countries that used the Latin alphabet.
Except in Francophone countries, where you're more likely to find AZERTY, or at least that was true 50+ years ago (the last time I had occasion to type in a Francophone country).
cameron said,
April 6, 2025 @ 9:24 am
yeah, AZERTY is still the norm in France. Typewriters designed for German typically use QWERTZ, which is pretty much QWERTY, with the Z and Y having exchanged places
Bybo said,
April 6, 2025 @ 9:29 am
… which is an even greater mystery, since it is obviously a modified QWERTY, but this permutation of letters does not in the slightest make it better for French specifically.
(I mostly use Ergo-L when typing English or French, and DIN QWERTZ for German.)
Chris Button said,
April 6, 2025 @ 10:30 am
I remember my first time being exposed to an AZERTY keyboard as a kid. I had only ever seen QWERTY until then, so I think I thought it was a jumbled up keyboard at first. The dedicated keys for common accented letters made me realize it was by design though.
It's interesting that ù gets its own dedicated key though. Granted, "où" is a very common word.
Chris Button said,
April 6, 2025 @ 10:51 am
As for Chinese, I suppose that typing the 35 "strokes" (one key for each stoke) is somewhat akin to typing "letters" (one key for each letter) to spell words:
一, ㇀, 乛, ㇇, , , ㇊, ㇅, ㇍, ㇈, 乙, ㇎,㇋, ㇌, , 〡, 亅, , ∟, ㇄, , 乚, , ㇉, ノ, ㇓, ㇢, , ㄑ, ヽ, ㇏, ㄟ, ㇂, ㇃, ㇁
Chris Button said,
April 6, 2025 @ 10:56 am
(Any missing stokes above will be Unicode CJK Unified Ideographs Extension B)
Jarek Weckwerth said,
April 6, 2025 @ 11:41 am
@Philip Taylor: Your problems are very likely to be software-induced. There is a considerable amount of dictionary-based and language-model-based "prediction" in the best software keyboards, making the exact location of presses less, erm, pressing (excuse the lousy pun). (And even more so for swype-style systems.) You need to install a better keyboard; those from even very reputable manufacturers such as Samsung (who do a fair bit of language tech in-house) are usually at least a notch or two below the best offerings, and I hate to point out that Cubot doesn't strike me as a very established brand. Have you tried Gboard?
David Z. said,
April 6, 2025 @ 1:34 pm
Philip Taylor: I am one of those people who find it almost impossible to text. That's even when I turn to phone sideways to provide more space between the letter. Although I have poor coordination, I have played guitar for decades, so I should have some degree of dexterity. I don't understand how people can text at all, let alone so quickly.
Michael Vnuk said,
April 6, 2025 @ 5:22 pm
Whoever wrote Sholes's name on the pictured patent has dotted the I in Christopher and also the preceding R. I have very occasionally made such an error in my own handwriting, although probably for other letters, as my handwriting more clearly distinguishes R and I. A similar and slightly more common error for me would be adding a crossbar to a letter with an upright that is not a T. Is there a name for this specific type of error, adding a superfluous dot or crossbar?
Michael Vnuk said,
April 6, 2025 @ 5:38 pm
The quoted article states that 'The type bars connecting the keys and letter plates hung beneath the paper' and Wikipedia at 'Typewriter' says: 'As with most other early typewriters, because the typebars struck upwards, the typist could not see the characters as they were typed. This arrangement, retronymically known as "understrike" would eventually give way to so-called "frontstrike" mechanisms in later, competing machines.'
People must have really hated handwriting if they were prepared to develop or use a machine where you couldn't see what you were creating. We should be eternally grateful to those who developed frontstrike machines. The QWERTY layout seems to be only a trifling inconvenience compared to not being able to see what you are typing.
Julian said,
April 6, 2025 @ 5:42 pm
@philip Taylor
i also use the left index and most mistakes are the letter to the left of the intended one.
I think it's to do with the fact that the actual touch point is hidden by the rest of the finger.
Deborah Pickett said,
April 6, 2025 @ 7:40 pm
For anyone remotely interested in the history of typewriters and keyboard layouts, I have only praise for Marcin Wichary's immense two-volume work "Shift Happens" (2023). Chapter 6 is all about the Dvorak keyboard, its promise, and why it didn't catch on and never will. Chapter 8 is all about international layouts and how most of them were made by US manufacturers shoehorning another language into the existing four-row design that their factories were churning out.
Wichary writes of the problem of typebar collisions on early typewriters—which were built in such a way that you couldn't see the letters you'd just typed until you'd rolled the platen a row or two, so you wouldn't know you'd mistyped something until sentences later—and some recent analysis showing that the arrangement that became QWERTY was designed to minimize these typebar errors. By the time that typewriters became the shape we all know today, and the original problem had been engineered out, QWERTY was already cemented as the layout that typists learned to touch-type on, and nothing, including Dvorak, could displace it. (Surprise fact: QWERTY predates touch typing!)
His conclusion is that despite other layouts, QWERTY was Good Enough for most people, and unless you've been forced through injury or ideology to adapt to another layout, most people stick with it because the reasons to switch aren't compelling enough. (I switched to Dvorak because of RSI while typing QWERTY. Progress was excruciatingly slow at first but it saved my career.)
mdhughes said,
April 6, 2025 @ 11:02 pm
In tech there's a lot of Dvorak enthusiasts, and some can type fast but not noticeably faster than equally-skilled Qwerty users. Somewhat objectively testing this with Typing of the Dead game, Qwerty users beat Dvorak users.
But ultimately it can't matter to me, or maybe 1/3 of other developers, for a simple reason: The main editor I use is vi (Vim). The layout of keys is essential: hjkl are the cursor keys, every key is mapped to a function or a branch to many functions (g). The only problem is I need Control more, so I remap Capslock to Control.
There's a minor need for layout with Mac, where Cmd-ZXCV are the clipboard/undo keys, and keeping them in place
Alexander said,
April 7, 2025 @ 1:26 am
Been typing in Dvorak for twenty-plus years. I had never heard the Morse code theory for QWERTY, I like it!
Chris Partridge said,
April 7, 2025 @ 3:37 am
The dominance of the QWERTY layout seems to have been a succession of coincidences.
Remington used qwerty not because it was more efficient but because it was different from earlier, copyrighted layouts.
Then the inventor of touch typing, Frank McGurrin, used a Remington machine because he happened to have one. He won all the typing competitions that were the rage at the time, so everyone wanted to use his system and typing schools standardized on qwerty.
The problem with touch typing is that learning it takes months so touch typists were unwilling to learn a new system when they changed jobs. So all employers had to also standardize on the qwerty system if they wanted to employ the best typists.
Robert T McQuaid said,
April 7, 2025 @ 6:17 am
I hope this will not get me disqualified for racism. Sometimes the French do something different just to be different. In addition to the AZERTY keyboard, they modify the Linnaean naming system for living things. Trembling aspen is Populus tremuloides everywhere but France, where it becomes Populus tremuloïdes.