Lock-In, Switching Costs & Standards Wars

You now know how increasing returns build a lead and how feedback loops amplify it. This lesson is about the trap on the far side: lock-in. Once a system has settled onto a path, why does it stay there — even when nearly everyone can see a better path a few steps away? The short answer is that getting there costs something, and that cost is the load-bearing beam of the whole model. So we start with the cost, then show why it freezes not just individuals but entire markets, then take a tour of the greatest standards wars in history to watch the freezing happen.

Hold that answer. Everything below is an elaboration of it.

Switching costs — the price of the exit door

Analogy. Think of your current option as an apartment you’ve already furnished, wired, and learned the quirks of. Moving to a nicer apartment across town isn’t free just because the rent is lower: there’s the deposit, the movers, the days off work, the new commute you have to re-learn, the friends who now live further away, the appliances that don’t fit the new kitchen. The rent difference has to beat all of that before moving is worth it. That whole pile is the switching cost.

Definition. A switching cost is the total cost — money, time, retraining, lost compatibility, and risk — of moving from the option you’re currently on to an alternative. Crucially it is a cost you pay in addition to the price of the new option, purely for the act of changing. It comes in recognizable types:

The single most important property: the bigger the switching cost, the deeper the lock-in. Switching cost is the depth dial on the trap. A standard with trivial switching costs is barely locked in at all (you can leave whenever a better option appears). A standard with enormous switching costs can hold a worse option in place for decades.

Worked example — a company trapped in its software suite. A 4,000-person firm runs its entire operation on an entrenched enterprise software suite. A rival suite is genuinely better and would save, say, $2 million a year in licensing and efficiency. Should they switch? Count the switching cost honestly:

• Learning: ~4,000 employees each lose productive time retraining — even at 20 hours each and $50/hour, that’s ~$4 million just in lost work, before formal training programs.
• Data / compatibility: fifteen years of records, macros, and integrations have to be migrated and re-tested; call it $3 million and a very real risk of corrupted or lost data.
• Contractual: the current license runs two more years with an early-exit penalty of $1.5 million.
• Search + risk: months of evaluation, pilot programs, and the ever-present chance the migration fails and has to be rolled back.

Add it up: the one-time switching cost is comfortably north of $8 million against a $2 million/year gain. Rationally, they wait. They may wait so long that the “better” rival is itself obsolete before the math ever tips. Multiply this decision across every firm in the industry and you see how a whole sector can stay frozen on software everyone privately grumbles about.

Lock-in as a coordination trap

Analogy. Picture a crowded conference where everyone agreed to switch from an old messaging app to a new one — but only if everyone actually moves. You open the new app and it’s empty. So you go back to the old one, where all your contacts still are. So does everyone else, for exactly the same reason. The new app stays empty not because it’s worse but because being first is being alone. Nobody wants to be the sole inhabitant of a better island with no bridges to anyone.

Definition. Lock-in is a stable state in which each participant’s individually best response — given what everyone else is doing — is to stay, even though everyone would be better off if they all moved together. In the language of nash-equilibrium, the “stuck on the worse standard” state is itself an equilibrium: no single player can improve their own outcome by unilaterally deviating. The better standard is also an equilibrium — but there’s no gradient leading from the bad one to the good one that any individual can climb alone. Economists call the resulting stickiness excess inertia: the system is too reluctant to change, holding a worse option in place because the first mover always pays and never wins.

This is worth distinguishing sharply from a cousin it’s often confused with:

• tragedy-of-the-commons is about over-use of a shared resource: my private incentive (graze one more cow, catch one more fish) diverges from the collective good, so the resource gets depleted. The failure is one of restraint.
• Lock-in is about being stuck at a bad coordination point: my private incentive is to match whatever everyone else already does, so nobody moves. The failure is one of coordination. Nothing is being depleted — the group is simply parked in the wrong spot with no individual able to drive out.

Worked example — the standard nobody defends but everyone uses. Suppose an industry runs on File Format A. A cleaner Format B exists. Every firm would happily be on B if the others were too — but the value of your format is the number of people you can exchange files with. A firm that unilaterally adopts B can suddenly trade with nobody: its “better” format is worthless in isolation. So each firm, reasoning correctly, stays on A. A is a bad equilibrium held together not by anyone’s preference for it but by everyone’s rational fear of being the one who moved first. That fear is the lock-in.

Standards wars — the grand tour

A standards war is what a market looks like while it’s tipping: two (or more) incompatible standards fight for adoption, each one getting more valuable the more it’s used, until increasing returns tip the market to a single winner and switching costs weld it in place. The pattern is always the same — an early lead, a positive feedback loop, and a switching cost that locks the outcome — but the flavors are gloriously varied. Here’s the tour.

VHS vs. Betamax — did the worse tech win?

Sony’s Betamax launched first (1975) with, by several measures, a crisper picture. JVC’s VHS (1976) offered longer recording time (crucially, enough to record a whole movie or a football game on one tape) and JVC licensed the format openly to many manufacturers, while Sony kept Betamax closer. More machines meant more pre-recorded tapes in that format at the video-rental store — which meant more reasons to buy that kind of machine, which meant more tapes. That’s the increasing-returns loop, and the video-store tape shelf was the network effect made physical. By the late 1980s VHS had won and Betamax was effectively dead. Whether VHS was truly “worse” is genuinely debated — VHS’s longer recording time was a real feature consumers wanted — but the case remains the classic illustration that picture quality alone did not decide it; adoption momentum and complementary tape availability did.

Pitfall: don’t over-tell this one. “The worse product won” is a great story, but VHS beat Betamax partly on a dimension (recording length) that consumers genuinely valued. Path dependence doesn’t require the loser to be strictly better on every axis — only that quality wasn’t the deciding force.

Which side of the road — a pure coordination standard

Here there is no “better” side. Left and right are perfectly symmetric; the only thing that matters is matching the cars near you. This is coordination in its purest form — the payoff comes entirely from conformity, zero from the intrinsic merit of the choice. Most of the world drives on the right today, but the switching cost is staggering, which is why changes are so rare and so expensive when they happen.

Worked example — Sweden’s Dagen H. On 3 September 1967, Sweden switched from driving on the left to driving on the right, in a single coordinated instant at 5:00 a.m. — literally every vehicle in the country stopped, changed sides, and restarted. It required years of planning, repainting road markings, moving thousands of bus stops and traffic signals, redesigning intersections, and a massive public-information campaign. The whole nation had to move at once precisely because you cannot switch road sides one driver at a time — a lone switcher just causes a head-on collision. Dagen H is the coordination trap solved the only way it can be: by a central actor forcing the simultaneous jump the market could never make on its own.

Railway track gauge — welded into the ground

Track gauge is the distance between the two rails. George Stephenson’s roughly 4 ft 8½ in “standard gauge” spread across early British railways and, through them, much of the world — arguably not because it was optimal (broad gauges offered a smoother, more stable ride and higher capacity) but because Stephenson’s lines got there first and connecting lines had to match to interoperate. Once a rail network is laid, the switching cost is almost comically literal: re-gauging means physically moving thousands of miles of rail, and re-fitting every locomotive and carriage. Countries and companies that adopted different gauges paid for it for a century in the form of “break-of-gauge” points where cargo had to be unloaded and reloaded onto different trains.

Pitfall: the “better” broad-gauge case is real but partial — standard gauge’s near-universal early spread gave it interoperability, which is itself a genuine value. As with VHS, “arguably better on some axis” is not the same as “unambiguously better all-things-considered.”

Metric vs. imperial — the stubborn hold-out

Nearly the entire world standardized on the metric system; the United States is the conspicuous hold-out, still officially running on imperial units (feet, pounds, gallons). Why not just switch? Switching cost, at civilizational scale: retooling every machine shop, re-labeling every product, redrawing every blueprint, retraining every worker, replacing every road sign and recipe and wrench set. Each of those is a switching cost, and summed across an economy the number is astronomical — so the U.S. stays put, paying a smaller, invisible tax forever instead of one enormous visible bill once.

Worked example — the Mars Climate Orbiter. In 1999, NASA lost a $125 million Mars probe because one team’s software produced thrust values in imperial units (pound-force seconds) while another team’s navigation software expected metric (newton-seconds). The unit mismatch sent the orbiter too close to Mars, where it burned up in the atmosphere. It is the perfect parable of the hidden cost of living in two incompatible standards at once: the “invisible tax” of non-standardization occasionally sends a very visible bill.

Software, languages, keyboards — layered lock-in

Modern digital standards stack every engine of increasing returns at once:

• Programming languages lock in through learning costs (a whole workforce’s fluency), complementary tooling (libraries, frameworks, hiring pools), and legacy codebases nobody dares rewrite. A “better” language can’t easily displace one with millions of trained developers and billions of lines of existing code.
• OS / platform ecosystems lock in through the apps, files, peripherals, and habits built on top — the platform is worth the ecosystem around it, so switching means abandoning the ecosystem.
• Keyboard regional layouts (QWERTY, AZERTY, QWERTZ, and others) persist by pure learning + interoperability lock-in: the layout in your fingers and the layout on every shared machine must match, so nobody re-learns alone.

Winner-take-all, tipping, and the moat that results

Analogy. A standards war is less like a marathon (where runners spread out along the road) and more like a stampede funneling through a single gate: once the crowd leans one way, everyone leans that way, and the market pours through one exit. Standards markets tip.

Definition. Because each adoption makes a standard more valuable (increasing returns), standards markets tend to be winner-take-all: rather than settling into a stable 60/40 split, they run away toward ~100/0. This is the critical-mass tipping dynamic from earlier in the course, viewed at the level of a whole market. Below a critical share, a challenger fizzles; above it, adoption sweeps and the market flips. And once one standard wins, the switching costs that trap users become, from the winner’s side of the table, a moat (see the moats topic): a durable structural barrier that keeps rivals out and customers in. Lock-in is a trap for the locked-in and a fortress for the winner — same wall, two sides. The user experiences “I can’t leave”; the incumbent experiences “they can’t leave,” and books it as a competitive advantage.

Worked example — the tipping threshold. Suppose a new platform needs 15% of a market to reach self-sustaining critical mass. Below 15%, every new user finds too few others already there, so growth stalls and reverses — the challenger dies. Cross 15%, and each new user now makes the platform attractive enough to pull in the next, and the feedback loop carries it to dominance. The entire war is really a fight to shove past that threshold before your rival does; whoever tips first inherits a moat, and the loser is left holding an island.

First-mover advantage vs. first-mover trap

Analogy. Being first to plant your flag on a hill can mean you own the best ground before anyone else arrives (advantage) — or it can mean you’ve built your fort on the first hill you found, only to discover a better one next door once your walls are already up and immovable (trap). Same act, opposite outcomes.

Definition. First-mover advantage is the head start that lets an early entrant accumulate the adoption lead, network effects, and switching costs that lock in a durable position. First-mover trap (or incumbent’s curse) is the flip side: the same early commitment can weld you onto an early, now-outdated version of the technology — your own switching costs, sunk investments, and installed base make it painful for you to move to the better approach a latecomer adopts freely. The pioneer who built the biggest installed base on the old standard has the most to lose by abandoning it.

Worked example — the numeric trap. An early mover locks in 70% of a market on version 1 of a technology and enjoys years of dominance (advantage). Then a superior version 2 appears. A fresh entrant can adopt version 2 instantly — they have nothing to migrate. The incumbent, however, must weigh the switching cost of dragging its huge installed base, its trained staff, and its legacy integrations from v1 to v2. That switching cost — the very thing that protected it — now anchors it to the obsolete version while the nimble newcomer leaps ahead. The lesson is nuance, not a slogan: first can be a fortress or a cage, and which one depends on whether the ground under your flag stays valuable.

Key takeaways

Where this goes next

We’ve watched lock-in freeze technologies, road systems, and file formats. But the deepest, stickiest lock-ins aren’t made of tape or track — they’re made of institutions and careers: laws, professions, credentials, org charts, and the skills you personally invested a decade in. The next lesson takes path dependence into the human and organizational world, where the switching cost is measured not in dollars but in identity, and where the standard you’re locked into might be your own life’s work.