Path Dependence and Lock-In

The idea in plain words

The keyboard you are probably using was designed in the 1870s for mechanical typewriters. Whether its layout is the best possible is disputed, but nobody seriously proposes replacing it, because everyone already knows it. The value of a keyboard layout lies largely in the fact that other people use it.

When value grows with adoption, history matters. A small early lead, perhaps from luck, timing or a single large customer, makes a technology more attractive to the next adopter, which widens the lead, which attracts the next. Eventually one option dominates, and switching away becomes very costly even if something better comes along. Economists call this path dependence, and the end state lock-in.

Where it comes from

In 1985 the economic historian Paul David published a short paper on the QWERTY keyboard. He argued that the layout became standard through a sequence of historical accidents, including typing contests and the training of typists, and that once typists and machines were matched, the layout was locked in, even though alternatives such as the Dvorak layout were, he suggested, more efficient. His broader claim was methodological: economic outcomes can depend on the order of events, and cannot always be predicted from the merits of the options alone.

W. Brian Arthur gave the idea a mathematical form in 1989. In markets with increasing returns, where each adoption makes the option more attractive to the next adopter, he showed that outcomes are unpredictable in advance, that the market tends to tip towards one option, that the winner need not be the better one, and that the result is hard to reverse. This was a sharp break from standard economics, which assumed diminishing returns and a single predictable equilibrium.

At about the same time Michael Katz and Carl Shapiro formalised network externalities: the benefit one user gets from a product depends on how many others use it. Telephones, fax machines, operating systems and messaging apps all have this property.

The challenge came in 1990. Stan Liebowitz and Stephen Margolis re-examined the QWERTY story and argued that the evidence for Dvorak’s superiority was weak, partly drawn from studies of doubtful quality. They argued more generally that cases of truly inferior technologies being locked in are rare, because if a technology is clearly better, someone can profit by organising the switch. The debate has never fully closed, and that itself is instructive.

What it does to the mind

Lock-in is partly psychological. Users invest in learning a system, and that learning becomes a sunk cost they are reluctant to abandon. Habits harden. Files, workflows and integrations pile up around a product until leaving it means rebuilding a life. Status quo bias then makes the familiar option feel safer even when it is not.

For decision-makers, path dependence creates a strange humility. In increasing-returns markets, being first, cheapest or best is not enough; being the one others expect to win matters. Expectations become self-fulfilling. That is why platform contests are fought with announcements, alliances and pre-launch commitments as much as with products.

Where it shows up

Betamax is the standard example, though, like QWERTY, the details are contested. The operating systems essay and the OS/2 road show lock-in around application compatibility. The email essay describes a technology so locked in that no rival has displaced it in fifty years. Itanium shows the lock-in of a software base defeating the best-funded attempt to replace it. The compatibility gravity pattern collects these cases.

What it asks of someone deciding

Identify which markets you are in. In markets with strong network effects or high switching costs, early adoption and expectations matter far more than in ordinary markets, and the cost of losing is permanent rather than proportional.

Think about the lock-in you create. Your own architectural, vendor and data-format choices become paths too. Ask how expensive each will be to leave.

Do not fight lock-in head-on. Challengers rarely win by being a little better on the incumbent’s terms. They win by changing what the contest is about, as the web did to desktop operating systems and the smartphone did to the PC.