Path Dependence & Lock-In

Look down at your keyboard. The top letter row spells Q-W-E-R-T-Y — a layout finalised in the 1870s, on a mechanical typewriter, partly to keep fast typists from jamming the machine’s swinging arms. The arms are gone. The jamming problem died with them a century ago. Demonstrably faster layouts exist and have for decades. And yet here you are, still typing on a keyboard designed around a constraint that hasn’t existed since before your grandparents were born. Nobody chose QWERTY on the merits. We’re all just standing where the first crowd happened to stand — and the crowd is now far too large to move.

That is path dependence: the outcome a system settles on depends on the sequence of things that happened along the way, not just on what’s best right now. A small, half-random early lead gets caught in a loop of positive feedback — each adopter makes the option more attractive to the next adopter — and the lead compounds until it hardens into a standard no single person can afford to abandon. The result is lock-in: a whole world stuck on an option that a clean-sheet designer would never have picked, held in place not by anyone’s decision but by the sheer weight of everyone having already decided. History, once made, refuses to be optimised away.

This is the expert-tier meeting point of two models you’ve already built. From critical mass you know that a reinforcing loop has a threshold — cross it and the process runs away. From feedback loops you know that reinforcing loops turn small differences into large ones. Path dependence is what those two do to history: an early accident, amplified past a threshold, becomes a permanent feature of the landscape. The organising engine is increasing returns — the idea, sharpened by economists Brian Arthur and Paul David, that some things get more attractive the more they’re used, which flips the ordinary economics of “the better product wins” on its head. When returns increase, the lucky product wins, and luck is decided early.

The course builds the model floor to ceiling. We’ll pin down what path dependence actually claims — and, just as important, what it does not (the difference between “the past mattered,” which is trivially true of everything, and “a small early accident got locked into a large, hard-to-reverse outcome,” which is the real, falsifiable claim). We’ll walk the famous QWERTY story and the honest academic brawl over it, because a model you can only see confirming itself is a superstition, not a tool. We’ll dissect the four engines of increasing returns — set-up costs, learning effects, coordination, and self-fulfilling expectations — so you can spot which one is running in any given case. We’ll see why a coordination trap keeps a whole market frozen on a worse standard when no individual can profitably switch first, and tour the standards wars — VHS versus Betamax, which side of the road you drive on, railway gauges, metric versus imperial, programming ecosystems — where the pattern is naked. We’ll follow path dependence out of technology and into institutions and your own career, where laws, organisations, and résumés calcify the same way. And we’ll be rigorous about how locked-in systems break (external shocks, a discontinuous leap in quality, coordinated switching, a deep-pocketed sponsor) and about where the model lies — the just-so-story trap of declaring every winner a historical accident, the sloppy slide from “history influenced this” to “history trapped this,” and the lazy assumption that lock-in is forever. You’ll leave able to tell a genuine lock-in from a story that merely sounds like one, to name the loop holding a bad standard in place, and to judge — before you commit — whether the path you’re stepping onto is one you’ll ever be able to step off.