History
Path dependence
Early choices, including accidental ones, can lock a system onto a track that later choices cannot easily leave.
Paul David, Clio and the Economics of QWERTY (American Economic Review, 1985); Brian Arthur's models of increasing returns to adoption (Economic Journal, 1989).
David defined a path-dependent sequence as one in which the outcome is shaped by remote earlier events, including ones dominated by chance; such processes do not settle on a single predictable result, and historical accidents cannot be set aside. His case was the keyboard. Christopher Sholes patented a typewriter in 1868, rearranged its keys over six years to reduce clashing typebars, and in March 1873 Remington took the manufacturing rights. David argued that three features then locked the layout in: technical interrelatedness (the hardware only has value with typists trained to match it), economies of scale (each new adopter lowers the cost for the next) and quasi-irreversible investment in skills that nobody wants to pay to retrain. Arthur's 1989 model made the general point: under increasing returns to adoption, small early events decide which of several technologies wins.
It holds because adoption feeds on itself. Every QWERTY typist made a QWERTY machine worth more, every machine made training on QWERTY the sensible choice, and the gap compounded while switching costs rose. David reported that Apple advertised its 1980s Dvorak switch as letting you type 20 to 40 percent faster, and that 1940s US Navy experiments found retraining on Dvorak paid for itself within ten days, yet the world did not switch. The prediction is that in any market with these three features the standard that wins need not be the best one, and that the longer it stands the more it costs to leave. File formats, programming languages, railway gauges and the first database a team chose all behave this way.
The empirical heart of the story is disputed. Liebowitz and Margolis, in The Fable of the Keys (Journal of Law and Economics, 1990), found that the Navy study was run by August Dvorak himself, who held the patent, and that its initial scores were measured in ways that favoured his layout; that a 1956 General Services Administration study by Earle Strong, which David did not cite, found retrained typists needed well over 25 days of four-hour sessions to regain their old speed; and that on 25 July 1888 a touch typist won a Cincinnati contest against a rival on a non-QWERTY machine, so the layout had competed and won. They accept that lock-in can happen; they deny the evidence shows a market trapped on a worse standard. David replied that his paper was about the process, not Dvorak's merits. Keep both: inertia is real, and "locked into something worse" needs proof.
Example
Christopher Sholes patented his typewriter in 1868 and spent six years rearranging the keys to reduce jamming; in March 1873 Remington took the manufacturing rights, and David estimates there were no more than about 5,000 QWERTY machines in the United States when the 1880s opened. On 25 July 1888 Frank McGurrin, touch typing on a Remington, beat Louis Taub, who used a Caligraph with a different keyboard, in a contest in Cincinnati. A 1940s US Navy study claimed that retraining typists on August Dvorak's rival layout would repay its cost within ten days of full-time work. Liebowitz and Margolis later found that Dvorak had run that study himself, and that a 1956 General Services Administration study by Earle Strong, with ten typists retrained on Dvorak against ten matched QWERTY typists, found the Dvorak group needed well over 25 days of four-hour sessions just to recover their old speed.
Where it applies
The first database, language or cloud a team picks becomes the one everything is built around, and each year of integration raises the price of leaving; the honest check is whether the alternative is really better or just newer. In a data platform, a schema chosen in week one decides what every downstream job assumes for years, so spend the extra day on it.
Think about it
Which tool or convention in your work survives mainly because switching is expensive, and have you ever measured whether the alternative is actually better?
Applied it to something real? Log it.
Test yourselfDavid names three features that lock a standard in. What are they, and what did Liebowitz and Margolis dispute about his keyboard story?
Technical interrelatedness, economies of scale and quasi-irreversible investment in skills. Liebowitz and Margolis disputed that Dvorak was clearly better: the Navy study was Dvorak's own and a 1956 GSA study found no advantage.
Related models
- InertiaA body keeps doing what it is doing until a force acts on it, and so, by a looser analogy, do habits, defaults and organisations.
- Chesterton's fenceBefore you remove a rule, a process or a piece of code, find out why someone put it there.
- Network effectsA product becomes more valuable to each user as more people use it, which makes leaders grow faster and makes leaving hard.
- The Lindy effectFor things that do not wear out, such as ideas, books and technologies, every year they survive raises how long they can be expected to last.
- The adjacent possibleAt any moment only the ideas one step from what already exists can be built; the rest wait until their parts arrive.
Sources and further reading
- Paul David, Clio and the Economics of QWERTY, American Economic Review 75(2) (1985)
- Stan Liebowitz and Stephen Margolis, Typing Errors, Reason (June 1996), restating The Fable of the Keys (1990)
- Stan Liebowitz and Stephen Margolis, How the Lock-In Movement Went off the Tracks, CAPRI working paper (2010)