Pattern/Systems Thinking/No. 0728
Path Dependence
Path dependence is when earlier events shape a system’s later options, costs, and incentives. Studied in economics by Paul David and W. Brian Arthur, it explains how history can steer choices and foster lock-in without making an outcome permanent or inefficient.
Also called Path Dependency
- Evidence
- Well established
- Read
- 6 min
- Links
- 15 connections
01You've seen this when…
- in life
You choose a phone brand for its camera. Years later, your photos, paid apps, and family subscriptions all sit in its ecosystem. Your next phone choice now depends on the camera and everything you have built around that brand.
- at work
A temporary spreadsheet becomes the team’s main planning tool. Other teams build imports around it, new employees learn its quirks, and replacing it now requires changing six workflows.
- out in the world
A town builds shopping centers around highway exits. Residents buy cars to reach them. When the council proposes better bus service, the destinations are already scattered across miles of parking lots.
02The idea
The database may have been an ordinary choice when your team was small. Ten years of decisions have changed what that choice means. Reports depend on it. Employees know it. Other systems expect its format. A replacement must now compete with the database and the entire arrangement built around it.
That is path dependence: the sequence of earlier events changes the options, costs, or incentives that come later. Two organizations facing the same customers and software prices can make different choices because they arrived with different skills, equipment, and commitments.
The useful question is whether a different plausible sequence could have left the system on a different route under similar outside conditions. What happens first can change what becomes attractive next.
Path dependence is broader than lock-in. A history can steer a system without trapping it permanently. Nor does it automatically mean the current arrangement is bad. The route taken may have built valuable capabilities.
03Why it happens
- Using something makes it more useful. A growing user base attracts suppliers and supports more training and compatible products. Those additions attract still more users. These increasing returns can turn an early lead into a lasting advantage.
- Experience accumulates unevenly. People improve the technology or process they actually use. An alternative with equal or greater potential may look worse partly because it has received less practice and investment.
- Other choices grow around the first one. A machine determines which parts you stock. A software platform shapes hiring. Each surrounding investment raises the switching costs of changing the original choice.
- Some moves remove options. Selling land, dismantling equipment, or losing a specialized workforce can make returning expensive or impossible. The order of decisions matters because later decisions happen with a different menu.
These mechanisms often reinforce one another. An early choice attracts investment; investment makes continued use attractive; continued use attracts more investment. This reinforcing feedback loop gives users reasons to stay beyond a preference for familiar things.
04A worked example
The United States’ commercial nuclear industry largely settled on light-water reactors. These use ordinary water for cooling and to slow neutrons. Early development included the Navy’s submarine program. The USS Nautilus used a pressurized-water reactor and first went to sea under nuclear power in 1955.
Economist Robin Cowan examined how early development and adoption helped light-water designs gain a durable advantage over competing reactor technologies.
What it looks like Utilities repeatedly choosing the winning technology, suggesting that it must have been the best option all along.
What’s actually going on Early investment supported manufacturers and helped build engineering experience and a growing body of operating knowledge. Later buyers could draw on that foundation. Their purchases then strengthened it further. Rival designs had to compete on technical merits against a technology backed by accumulated expertise and infrastructure.
What would have helped Keeping credible alternatives available during early development while comparing designs before experience became concentrated in one family could have preserved more options. That approach would have carried costs, and whether it would have been better policy remains uncertain.
The case supports historical reinforcement, not a verdict that light-water reactors were inferior. Other countries developed different reactor traditions. Path dependence helps explain the divergence; assessing which route was best is a separate task.
05How to spot it
06What to do about it
- Find the decisions that narrowed the route. Sketch a short history of major commitments. Separate the original choice from the later investments that now sustain it.
- Evaluate the choice from today’s position. Compare future benefits and costs, including transition costs. Assess a useful system on those grounds even if its origins look accidental.
- Preserve alternatives before commitments multiply. Open formats, portable data, modular equipment, and limited pilots can protect option value. Focus on decisions that will be hardest to undo, the one-way doors.
- Change the centerpiece and the supporting arrangement together. A new tool may need training, compatible suppliers, and revised workflows. Gaps in those complements can cause a technically good replacement to fail.
- Use a bridge when a jump is too costly. Run old and new systems together temporarily, provide conversion tools, or move one group first. Treat these as real options: staged commitments that let you learn before committing further.
Keeping every option open also costs money. Preserve the alternatives whose future value plausibly exceeds that cost.
07When it isn’t a trap
Path dependence can produce expertise, reliable coordination, and useful standards. A common system can be worth keeping for its compatibility, even when better systems are imaginable.
Money already spent is gone. The skills, equipment, and obligations it created may affect future costs. Continuing because employees know a system can be rational; continuing merely to justify the original purchase is the sunk cost fallacy.
Hysteresis describes a more specific feature: a system’s state depends on its history, often persisting after the force that changed it is removed. Path dependence emphasizes how a sequence shapes subsequent development; the concepts overlap but are not interchangeable.
Finally, persistence alone proves little. A standard may endure because it still works well. The famous claim that QWERTY trapped users in a demonstrably inferior keyboard layout is disputed. Historical influence is easier to establish than avoidable inefficiency.
08Roots
In 1985, economic historian Paul David made a typewriter keyboard the centerpiece of an argument about technology. Computers had left much of the typewriter’s machinery behind, yet their top letter row still began QWERTY. Why did a design from an earlier machine remain embedded in a new one?
David traced the interaction of keyboard design, manufacturers, and trained typists. Once employers wanted a common layout and workers learned it, each side strengthened the other’s reason to stay. The keyboard offered a memorable challenge to the idea that competition always selects the best technology independently of history. Later critics, notably Stan Liebowitz and Stephen Margolis, disputed the evidence that QWERTY was inferior and that its survival represented market failure.
W. Brian Arthur gave the broader argument a mathematical form. His 1989 paper modeled competing technologies whose attractiveness increased with adoption. Early events could push adoption toward different lasting outcomes, making the eventual winner dependent on the sequence of adoption. The notion of dependence on a path had earlier mathematical uses; David and Arthur helped make it central to economic accounts of technology. It subsequently became a way to understand institutions, infrastructure, and organizational routines: arrangements whose present logic was partly built by their past.
09How solid is this?
The mechanisms are well established in formal models and documented historical cases. Establishing that a particular outcome was inefficient is harder: evidence of lock-in must go beyond persistence, and QWERTY’s alleged inferiority remains disputed.
10Connections
- Often confused with Sunk Cost Fallacy, Hysteresis, Slippery Slope Fallacy
- Countered byReal Options
- Can lead toLock-In
- Can follow fromReinforcing Feedback, Switching Costs, Increasing Returns
- Part ofMultiple Equilibria
- Includes Matthew Effect, Technical Debt, Two-Way vs. One-Way Doors
- See also Conway’s Law, Gall’s Law, Option Value
+ 5 more in the list
11Origin and sources
Earlier mathematical usage; developed as an economic account of technological change by Paul David (1985) and W. Brian Arthur (1989).
- [1]David, P. A. (1985). Clio and the Economics of QWERTY. The American Economic Review, 75(2), 332–337.
- [2]Arthur, W. B. (1989). Competing Technologies, Increasing Returns, and Lock-In by Historical Events. The Economic Journal, 99(394), 116–131.
- [3]Cowan, R. (1990). Nuclear Power Reactors: A Study in Technological Lock-in. The Journal of Economic History, 50(3), 541–567.
- [4]Liebowitz, S. J., & Margolis, S. E. (1990). The Fable of the Keys. The Journal of Law and Economics, 33(1), 1–25.
Suggest an edit· Updated 2026-10-02