Path Dependence

Transfers

Path dependence is the principle that the sequence of past events constrains future possibilities. Where you are now depends not just on current conditions but on the specific order in which you arrived. The model says: history is not noise to be averaged out; it is structure that limits what can happen next.

Key structural parallels:

• Early choices constrain late options — the QWERTY keyboard layout was designed to prevent mechanical jamming on 19th-century typewriters. The mechanical constraint vanished decades ago, but the layout persists because typists, training programs, keyboard manufacturers, and software input systems all co-adapted to it. The cost of switching now exceeds the benefit of any alternative layout, even if that alternative is objectively superior. The model identifies this pattern across domains: technology standards, urban layouts, institutional structures, and codebases all exhibit it.

• Increasing returns create lock-in — path dependence is powered by positive feedback loops. Each person who learns QWERTY makes it more costly for the next person to choose Dvorak. Each module built on a framework makes it more expensive to switch frameworks. Each business that opens near a highway exit makes the exit more valuable, attracting more businesses. The returns to staying on the current path increase with the number of agents already on it, creating a gravitational pull that strengthens over time.

• Multiple equilibria are possible — the model says there is no single inevitable outcome. Different sequences of early events lead to different stable states, any of which could persist. The VHS-Betamax outcome was not determined by technical superiority; it was determined by a sequence of licensing decisions, retail partnerships, and content availability that could have gone differently. Path dependence means the world we inhabit is one of many possible worlds, and the one we got was selected by historical accident rather than optimality.

• Ergodic vs. non-ergodic systems — the deepest contribution of path dependence theory is the distinction between systems where time averages equal ensemble averages (ergodic — history washes out) and systems where they do not (non-ergodic — history accumulates). Flipping a fair coin is ergodic; your long-run frequency converges to 50% regardless of early results. Building a city is non-ergodic; the location of the first well determines the street grid permanently. The model provides a diagnostic: if early events can be “undone” by later ones, history does not matter. If they cannot, path dependence applies.

Limits

• The model overpredicts lock-in — the QWERTY story is compelling, but the world is full of path breaks. Vinyl gave way to CDs gave way to streaming. Mainframes gave way to PCs gave way to cloud. Entire transportation systems have been replaced (horses to cars to potentially autonomous vehicles). Path dependence explains persistence well but explains change poorly. The model says lock-in should strengthen over time, yet industries are routinely disrupted by innovations that make accumulated switching costs irrelevant.

• Retrospective narrative bias — path dependence is far easier to identify after the fact than before. Looking backward, it is always possible to construct a story about how early events constrained later ones. Looking forward, it is rarely clear which current decisions will prove path-dependent and which will wash out. This asymmetry makes the model more useful for explanation than prediction, and it can lend a false sense of inevitability to outcomes that were genuinely contingent.

• The excuse problem — “we’re path-dependent” can function as a sophisticated way of saying “we can’t change.” Organizations use path dependence to justify legacy systems, outdated processes, and suboptimal structures. The model provides intellectual cover for inertia. But path dependence describes costs of change, not impossibility of change. The question is always whether the switching cost is worth paying, and the model alone cannot answer that.

• Falsifiability — if the current state persists, path dependence explains it (lock-in). If the current state changes, path dependence explains that too (sufficient shock). This flexibility makes the model difficult to falsify in any specific case, which is a genuine weakness for a concept used to make causal claims about why things are the way they are.

Expressions

• “We’re locked in” — the practical consequence of path dependence, describing a state where switching costs exceed the benefits of change
• “Legacy system” — in software, the accumulated result of path-dependent decisions, where each layer was reasonable when built but the aggregate constrains future options
• “Switching costs” — the economic framing of path dependence, measuring the price of leaving the current path
• “QWERTY economics” — Paul David’s term for the study of how historical accidents become permanent standards
• “Installed base” — the population of existing users, devices, or systems that creates the gravitational pull toward the current path
• “You can’t get there from here” — the folk expression of path dependence, acknowledging that the optimal destination is unreachable from the current position without prohibitive cost

Origin Story

Path dependence entered economics through the work of Paul David (“Clio and the Economics of QWERTY,” 1985) and Brian Arthur (“Competing Technologies, Increasing Returns, and Lock-In by Historical Events,” 1989). Both were responding to the neoclassical assumption that markets converge on optimal outcomes regardless of initial conditions. David’s QWERTY analysis and Arthur’s mathematical models of increasing returns showed that small early advantages could compound into permanent dominance, even when the dominant solution was not the best one. The concept was subsequently adopted in political science (Pierson, 2000), organizational theory, and technology studies, becoming one of the most influential cross-disciplinary models of the late 20th century.

References

• David, P.A. “Clio and the Economics of QWERTY” (1985) — the founding case study of path dependence in economics
• Arthur, W.B. “Competing Technologies, Increasing Returns, and Lock-In by Historical Events” (1989) — the formal mathematical treatment
• Pierson, P. “Increasing Returns, Path Dependence, and the Study of Politics” (2000) — extends the concept to political institutions
• Liebowitz, S.J. & Margolis, S.E. “The Fable of the Keys” (1990) — the principal critique, arguing that the QWERTY case is overstated