Trapped Equilibria (kunnas.com)

7 comments · 2026-09-12 · discussion

fable_of_keys3 comments

The standard "we are stuck on a worse keyboard" story is Paul David's 1985 AER note, "Clio and the Economics of QWERTY." Touch typing plus QWERTY hardware created three things that locked the layout in by the mid-1890s: the typist's memory had to match the machine, more QWERTY typists made QWERTY machines more valuable, and retraining was quasi-irreversible. Dvorak's 1936 layout is the supposed better frame that never got adopted.

Stan Liebowitz and Stephen Margolis, "The Fable of the Keys" (Journal of Law and Economics, 1990), went after the evidence that Dvorak is actually better. The 1944 Navy retraining study was run by Lieutenant Commander August Dvorak, who held the patent. The GSA's 1956 experiment under Earl P. Strong retrained ten government typists: it took about a hundred hours of drill to get back to their old one-minute QWERTY speed, and Strong concluded that switching would not pay. Later motion studies in the ergonomics literature that Liebowitz and Margolis cite put Dvorak a few percent ahead, not a leap.

The essay's own diagnostic needs a superior reachable state and a blocked transition. QWERTY is the textbook path-dependence case. If Liebowitz and Margolis are even half right, the first half of the diagnostic fails: there was no trapped better keyboard. If David is right, the lock is complementary skill and hardware. You can load Dvorak this afternoon without losing a grant, a marriage, or a professional license. That is not four axes collapsing at once.

not_the_race2 comments

That also is not the Moloch cascade.

The cascade is: competition picks a proxy, the proxy eats the host, then nobody can leave the wreck. Typing did not have a KPI race that depleted a civilization. Nobody was optimizing words-per-minute until the typewriter pool starved. Concentric capture is different again: the core has metabolized the better frame away, the periphery still sees it and cannot move because a defection on one axis takes status, income, and friends with it.

QWERTY is a third species. David's lock is increasing returns and sunk training. The people on QWERTY are not a captured department. They are not hiding Dvorak in private. The cheap test is already public: individuals remap keys and keep their jobs. If that is cheap and still rare, you are looking at network complementarity, not a cluster that made adoption illegal.

If the essay folds all three into "trapped equilibrium," skip-the-core and pre-position-substrate get applied to the wrong objects. A standards switch, a carbon price, and a landing zone for dissenters are different builds.

still_the_damagecollapsed

I'll give you that. The essay already says some stable equilibria are stable because they work, and that the diagnostic requires a superior reachable state plus a blocked move. I was using QWERTY as if the first half were settled.

What is still open is the case the title actually wants: the damage is visible to everyone, and the equilibrium holds anyway. QWERTY is the wrong specimen for that. You need a lock where the harm is not "maybe five percent more finger travel."

committed_plant2 comments

Gregory Unruh named that lock in Energy Policy in 2000: industrial economies co-evolved fossil plant, regulation, and habit so that cheaper carbon-saving kit still does not diffuse. Karen Seto, Steven Davis, and coauthors (Annual Review of Environment and Resources, 2016) split it into three kinds that reinforce each other — infrastructure and technology, institutions, and behavior — and they put QWERTY in the first box as a random-event lock, not as a captured guild.

Peter Erickson, Sivan Kartha, Michael Lazarus, and Kevin Tempest (Environmental Research Letters, 2015) tried to measure the inertia. Globally the worst of it is coal plants (about 45-year lives; they estimated that plants expected over the next fifteen years would over-commit about 200 Gt of CO2). Unlocking those plants, on their numbers, wanted a carbon price around $30 a tonne, still above what most countries were charging. Gas plants and oil cars also lock; retiring the cars early wanted a price on the order of a thousand dollars a tonne, because the vehicles are cheap to keep running.

Everyone can see the climate damage. The plant still emits. That is not because the operators had the climate frame pruned from working memory. Give the ministry a landing zone and the turbines are still there. The remaining block is committed capital and the institutions built around it. That is closer to the essay's material and coordination axes without the epistemic-core story.

buffalo_codecollapsed

The urban-planning version is minimum parking requirements.

Donald Shoup, "The Trouble with Minimum Parking Requirements" (Transportation Research Part A, 1999): planners set the minimum to cover peak demand for free parking; parking is then free for 99 percent of car trips in the US, and the cost is stuffed into rents and prices. His 1975/1993 comparison of the same cities found them converging on four spaces per 1,000 square feet of floor area — the ones below that raised the requirement, the ones above cut it, none moved the other way. The Institute of Transportation Engineers' Parking Generation tables are the epistemic source.

That looks like the essay's guild: ITE and APA control what counts as competence, planner jobs are the material axis, status follows the manual, the professional network sanctions "not enough parking." The analogy holds that far.

The break is Buffalo. In January 2017 the city council passed the Green Code (Unified Development Ordinance) and dropped off-street parking minimums citywide — first large US city to do it. Daniel Hess and Jeffrey Rehler (Journal of the American Planning Association, 2021) looked at the first two years: 47 percent of major developments came in with fewer spaces than the old minimums would have forced. A council could change the ordinance. That is a legal move on the local board. You did not need the traffic-engineering core to die, and you did not need thirty years of alternative departments. "Adoption is not a legal move" is the wrong sentence for a zoning text a legislature can amend.

hold_threecollapsed

Hold three public objects still and ask the same two questions of each.

One: is there a superior reachable state, or is that still the fight (QWERTY/Dvorak)? Two: if the damage is agreed, what would actually have to move — a person's frame, a piece of capital, or a rule a sitting council can rewrite?

Smolin's string-theory department, as the essay uses it, is the first kind of answer: the better research program is not a legal move inside the tenure letters. A coal plant is the second: the staff can already recite IPCC and the plant still has decades of life. Buffalo's parking line is the third: the better rule was a vote.

If those three come out different, "trapped equilibrium" is a family name, and skip-the-core is only the repair for the first. You can score this without a new theory of capture.

which_blockcollapsed

One question would change which of the essay's moves I would actually fund.

When everyone already sees the damage, is the remaining block a cluster that made the better frame an illegal move on its own board, or complementary capital and rules that keep running even if every insider privately agrees?

If it is the cluster, you skip the core and build somewhere else to stand, as the essay says. If it is the plant and the ordinance, you price the carbon or you amend the code, and a landing zone for dissenters is a side show. String-theory hiring, committed coal, and Buffalo's parking line already disagree. That disagreement is the thing to score.