Every term of art on this site, defined the way the studies are built: a plain one-liner anyone can take away, real depth underneath, the owner named where a term carries a worldview, and the common misreading called out — because a definition that can’t say what it does not mean is advertising.
Where a concept has a great interactive teacher, it’s credited and linked — play sits above read on purpose. Those little machines are the tradition our own study surfaces belong to: put the dial where the reader is, and let consequence teach.
Three house rules. Linking a module or a thinker is a claim that the work teaches the concept well — not an endorsement of anyone’s other positions. In-study glossaries stay just-in-time and plain; this page is the canonical reference layer they point to. And this page is in scope for the corrections form: a bad definition, a dead link, or an unfair attribution is a reportable error here, same as a number in a study.
Family A — the metacrisis vocabulary
These terms carry a worldview, so every entry names whose term it is — “Schmachtenberger uses X to mean…”, never “X is…”. Where a term is contested, the entry says so.
The unfolding. The word does one specific job: it separates the crises you can see from the machinery that keeps manufacturing new ones. Climate, debt, pandemics, misinformation, arms races — each has its own experts and its own emergency meetings, and each can be addressed on its own terms. The metacrisis framing says that treating them one at a time misses the point, because the same few dynamics keep generating fresh instances: races nobody can quit unilaterally, technological capability compounding faster than the judgment governing it, and a shared inability to agree on what’s true (see A3 for the named list). Solve one visible crisis without touching the generators and the machine simply produces the next one. That’s the claim — and it is a claim, carried on this site as a framing lens sourced to named thinkers, not as house doctrine.
An example. Our two studies are deliberately instances of the same generators in different domains: an oil shock’s household incidence traces how a distant geopolitical event lands on the people least padded against it, and the orbital-commons study finds a filing race, a governance vacuum, and a physical threshold in low-Earth orbit — different facets, recognizably the same machinery.
Whose it is. Daniel Schmachtenberger’s usage, developed with the Consilience Project; the word itself circulates more broadly and has no single owner. Distinct from polycrisis (A2), which names the visible bundle rather than the generators.
What it does not mean. It isn’t a synonym for “everything is bad” or a brand for apocalypse content. The analytical content is the distinction between symptoms and generators — a metacrisis diagnosis stands or falls on whether the generator dynamics are actually there, which is precisely what studies exist to check.
The unfolding. The prefix is the content: poly doesn’t just mean “many,” it means the crises couple. A drought that raises food prices that stress governments that disrupt trade that raises food prices further — the harm of the bundle exceeds the sum of the parts, because the crises share transmission lines. The term was popularized in economic commentary by Adam Tooze and given formal definitional machinery by the Cascade Institute (Thomas Homer-Dixon and colleagues), whose framing this site’s Complex-Systems seat holds to a deliberately strict standard: the word is only earned when you can name the actual mechanism coupling one crisis to another. Otherwise “polycrisis” is just a mood.
An example. The Uneven Month models this discipline directly. Its floor band treats each month’s shocks as independent; its coupled band explicitly turns on the couplings — energy, shipping, food, and credit going wrong together — and states their assumed strength. The finding is what coupling is worth: the extreme tail moves from roughly $49B to roughly $69B a month. That gap is polycrisis, priced.
Whose it is. Widely used; popularization credited to Tooze, formal definition to the Cascade Institute.
What it does not mean. Not “several problems at the same time” — coincidence isn’t coupling. The claim that crises amplify each other is a mechanism claim, and it can be wrong in either direction: assuming coupling that isn’t there inflates the scare, missing coupling that is there understates the tail.
The unfolding. If the metacrisis (A1) is the diagnosis, generator functions are the proposed disease mechanism. Schmachtenberger’s three: first, rivalrous dynamics — races in which restraint is punished, so every player runs even when all of them can see the cliff (A4). Second, exponential technology — capability that compounds on an engineering timescale while the wisdom and institutions governing it move on a treaty timescale, so the gap widens by default (D16). Third, failures of collective sensemaking — the information environment degrades faster than societies can repair their shared picture of what’s true (A5), which cripples the coordination the first two problems demand. The practical use of the list is that it tells you where a fix has to reach: a policy that leaves the generator running is a symptom patch.
An example. The orbital-commons study finds all three operating in one thin shell of sky: a million-satellite filing race nobody can unilaterally exit; constellation deployment scaling in years against governance instruments written in 1967; and a public debate conducted in press releases and docket filings that even the regulator struggles to read.
Whose it is. Schmachtenberger’s framing and his list of three.
What it does not mean. No one is steering. A generator function is a structure, not a conspiracy — the whole point is that it produces bad outcomes without requiring bad actors, which is why exposing villains doesn’t turn it off.
The unfolding. The structure has three load-bearing parts. Players are rivalrous: one’s gain is another’s loss, at least in the short run. Restraint is individually punished: the fishing boat that takes less simply catches less while the stock collapses anyway; the lab that ships slower watches the market go to the lab that didn’t. And exit is unilateral-proof: no single player’s good behavior changes the outcome, so good behavior looks like a donation to your competitors. The result is that everyone runs a race everyone can see ending badly. Formally this is old terrain — arms races, commons depletion, collective-action problems — and game theory has studied it for decades under several names. What the metacrisis framing adds is the claim that our most consequential races (AI capability, attention capture, orbital deployment) now share this structure at global scale, where there is no outside referee to appeal to.
An example. The satellite filing race, with the wrinkle our race model exists to quantify: filing is nearly free and reserves valuable orbital real estate, so a million-satellite paper queue accumulates — but whether the deployment race is a genuine trap turns out to be conditional. In roughly 69% of the payoff space the model sweeps, the race fizzles — the paper never becomes hardware — and the trap only binds in the slice where the business case is real. Naming something a trap is a measurement, not a vibe.
Whose it is. Schmachtenberger’s “multipolar trap” framing, itself resting on game-theoretic ancestors (Schelling’s work on strategy and commitment; the collective-action literature). Boeree’s “Moloch trap” (D10b) is the same structure under its most travelled name.
What it does not mean. Not every competition. Markets, sports, and science are competitions that mostly produce their intended goods. The trap is the special case where the structure punishes restraint and harms everyone including the winners — see D10’s guard: competition itself is neutral.
The unfolding. Individually, sensemaking is what you do when the picture is confusing: gather signals, weigh sources, revise. The term as used here scales that up: can a group — a town, a country, a civilization — still converge on a working picture of reality fast enough to act on it? That capacity is infrastructure, as real as roads: it lives in institutions (courts, journals, statistical agencies), in norms (what counts as evidence, what shame attaches to lying), and in the incentive structure of the information environment itself. The metacrisis framing treats its decay as a generator function (A3) because it’s upstream of everything else — coordination on climate, pandemics, or orbital debris all presuppose enough shared reality to coordinate about. The word predates this usage: Karl Weick built organizational sensemaking theory decades earlier, and this site’s usage is the Schmachtenberger-flavored civilizational version.
An example. The CRASH Clock — the stress gauge our orbital study reproduces — is a sensemaking instrument in the exact sense: it compresses an invisible, contested situation (orbital congestion) into one legible, checkable number that moved the public conversation. This site’s own corrections system is the same bet at small scale: publish the reasoning, invite the challenge, log the fix in the open.
Whose it is. Schmachtenberger’s usage in the metacrisis context; the word’s provenance runs through Weick’s organizational theory.
What it does not mean. Not consensus. A society with healthy sensemaking still disagrees — about values, priorities, trade-offs. The failure state isn’t disagreement; it’s when the disagreement is about what the facts are, all the way down, with no shared procedure left for settling any of it.
The site is named for this one.
The unfolding. In the language of systems, an attractor is a state a system tends to fall into and stay in (see D17). Schmachtenberger’s argument is that our current trajectory has two of them waiting. The first is catastrophe: the coordination failures compound, the systems we depend on come apart, and things fall down. The second is dystopia: the failures get answered with control — surveillance, centralization, an order imposed tightly enough to hold the pieces together, and stable in the way a cage is stable. The claim that gives this site its name is that both are stable — they’re attractors, not accidents — and that a third state exists which is neither: one where coordination is good enough to handle the power we’ve accumulated, without a boot on the scale. What it would take, on this account, is sensemaking and coordination commensurate with our power — which is a demanding standard, not a hopeful adjective.
An example. The two traps show up cleanly in miniature in the orbital commons: a debris cascade that renders useful orbits unusable is the catastrophic attractor; a single actor or bloc controlling access to space and setting the terms is the dystopian one. A third path — enforceable, mutually-agreed rules that constrain everyone, including the strongest player — is neither, and it’s the harder thing to build.
Whose it is. Daniel Schmachtenberger’s framing, developed in his own writing and in the Consilience Project’s work. We borrowed his frame; we don’t speak for him, and he has no involvement in this site.
What it does not mean. It isn’t a prediction that the third path is likely, and it isn’t a plan. It’s a claim about the shape of the possibility space — that the comfortable middle we imagine we’re in is less stable than it feels, and that steering somewhere better is an active task rather than a default. Naming a site for it is a statement of what we’re aiming at, not a claim to have found the route.
The unfolding. Most risk conversations track only the first failure mode — collapse, breakdown, the systems we depend on failing. This distinction insists on watching a second one with equal seriousness: the world where the breakdown is prevented, at the price of control tight enough to be its own catastrophe — pervasive surveillance, centralized authority over speech and movement, order maintained the way a prison maintains order. The analytical point is that these two risks trade against each other: crude responses to catastrophic risk (lock everything down) increase dystopian risk, and crude responses to dystopian risk (tear the controls out) increase catastrophic risk. Any governance proposal worth taking seriously has to say how it avoids both — which is exactly the needle the third attractor (A6) names.
An example. In the orbital commons: the cascade that trashes low-Earth orbit is the catastrophic branch; a single gatekeeper deciding who flies is the dystopian one. The study’s avenues section is, in effect, a search for rules that reduce the first without purchasing the second.
Whose it is. Schmachtenberger’s distinction.
What it does not mean. It isn’t a claim that all safety measures are incipient tyranny, or that all decentralization is incipient chaos. It’s a reminder that “did we prevent the collapse?” is only half the exam.
The unfolding. A barrel of oil delivers work equivalent to years of human physical labor, for a few dollars. Everything ordinary — the breakfast shipped across a continent, the commute, the concrete, the fertilizer behind the bread — embeds that subsidy, and because the subsidy has been reliable for generations, it reads as background rather than as a contingent windfall. Hagens’ term names the perceptual default that results: we attribute prosperity to technology, productivity, and cleverness, and systematically under-attribute it to the dense, cheap energy those things run on. The blindness matters because it distorts planning — if you don’t see the energy under the economy, you’ll overestimate how easily the economy survives that energy getting scarcer or dearer.
An example. The Uneven Month is an exercise in removing the blindfold at household scale: one strait, one disrupted flow of oil, traced month by month into grocery bills, commutes, and rent pressure — with the burden concentrating, in share-of-income terms, on the households least able to absorb it.
Whose it is. Nate Hagens; a standing theme of The Great Simplification.
What it does not mean. It isn’t an accusation of stupidity or denial — the blindness is structural, the natural result of a subsidy so steady it disappears into the baseline. And seeing the dependence is not the same as any particular claim about what happens next (that’s A11, flagged separately as a thesis).
The unfolding. Watch the global economy from far enough away and it behaves with unsettling coherence: it seeks energy, converts it into structure and waste, routes around damage, and grows — regardless of which humans, firms, or governments occupy its parts. Hagens uses the superorganism image to name that aggregate behavior and, crucially, its ungoverned-ness: no one is in charge of the whole, so appeals to “we should just” founder on the absence of any actor with the steering wheel. The image has a long lineage — superorganism talk comes from entomology (ant colonies), and economy-as-organism metaphors are older still — and it earns its keep here as a discipline against wishful thinking about control, not as a scientific claim about what the economy literally is. This site carries it explicitly as a lens (the charter’s phrase: framing, sourced to named thinkers, not house ideology).
An example. The satellite filing queue: no individual, including the operators themselves, chose “a million satellites on file.” Each filing was a locally rational move in a race (A4); the aggregate is a system-level appetite nobody decided on — which is the superorganism observation in miniature.
Whose it is. Hagens’ usage, in The Great Simplification; the metaphor’s lineage is long and shared. Flagged as the more thesis-like of his two entries here, alongside A11.
What it does not mean. Not literal biology, and not an agent with intentions — the economy doesn’t want anything, and treating the metaphor as an entity to blame repeats the villain mistake (A3). It also isn’t fatalism: ungoverned is a description of the present arrangement, not a proof that governance is impossible.
The unfolding. Drilling, mining, refining, and building power plants all consume energy before they deliver any. EROI is the ratio of energy delivered to energy invested: an early conventional oil well might return on the order of dozens of units per unit invested, while harder resources — deep offshore, tar sands, some biofuels — return far less. The concept, formalized by Charles Hall and colleagues in biophysical economics, matters because the surplus (D1) is what everything other than energy-getting runs on: hospitals, schools, art, and leisure are all funded out of the energy left over after the energy sector feeds itself. A civilization sliding down the EROI curve has to dedicate an increasing share of its total effort just to keeping the lights on — a squeeze that shows up as cost pressure everywhere while remaining invisible in ordinary accounting, which tracks dollars, not joules. The honest caveat: EROI estimates are sensitive to where you draw the boundary (wellhead vs delivered vs useful energy), and published numbers for the same resource can differ substantially for that reason.
An example. (Outside example — no study of ours computes EROI yet; this swaps for our own the day one does.) The canonical contrast is conventional crude versus tar sands: the same barrel at the pump, radically different energy bills behind it — which is why “we have plenty of oil” and “the cheap oil is gone” can both be true.
Whose it is. Standard in biophysical economics — Hall and colleagues; Hagens popularizes.
What it does not mean. EROI is not a single settled number per fuel, and low EROI is not a moral property — it’s an accounting result that depends on stated boundaries. Quoting an EROI without its boundary convention is the energy version of quoting a percentile without its distribution.
The unfolding. The argument runs: modern economies grew on a one-time subsidy of dense, cheap fossil energy (A8), amplified by credit that pulls future consumption into the present. As the cheap half of the resource base depletes (A10) and debt saturates, the system faces a downshift it didn’t choose — not an apocalypse, but a forced simplification: less energy per person, shorter supply chains, fewer complex intermediated services, more of life re-localized. Hagens frames it as a bend, not a break — decades, not a weekend — and argues the useful work is preparing institutions and expectations for the smaller throughput rather than betting everything on the trend resuming. This site reports the thesis because it disciplines our biophysical reading of the economy; it does not adopt it as a prediction, and no study of ours assumes it.
An example. (The thesis’s own illustration, not a study finding.) Hagens’ benchmark image: the average American commands energy equivalent to hundreds of full-time human laborers. The Simplification question is simply what daily life looks like when that retinue shrinks by a third — not to zero, to a third less — and which arrangements survive the shrinkage gracefully.
Whose it is. Nate Hagens — his podcast and writing project carry the name. Explicitly a thesis; explicitly not a forecast we endorse.
What it does not mean. Not a dated prediction, not a collapse scenario, and not a tribe. Treating it as an identity (“we’re the ones who know it’s all ending”) is exactly the doom-basin reflex this site exists to resist — the thesis is an argument about energy and credit, and it stands or falls on those terms.
The unfolding. A population is in overshoot when its consumption exceeds its environment’s regenerative rate — spending the principal, not the interest. The concept (William Catton’s Overshoot is the classic statement) has two features that make it treacherous. First, overshoot is comfortable while it lasts: drawing down a stock feels identical to living within a flow, right up until the stock runs thin — the signal arrives late by construction. Second, the carrying capacity being overshot can itself be eroded by the overshoot (degraded soil, collapsed fisheries, destabilized climate), so the level you must eventually return to is lower than the level you could have held. Both features recur across scales, from a fishery to the planetary accounts the boundaries framework (D3) tries to formalize.
An example. The orbital commons runs the same arithmetic in a vacuum: debris is created faster than atmospheric drag removes it in the higher bands, and our study’s central physical finding is which shells self-clean (flow) and which accumulate for centuries (stock). The 700–1,000 km ratchet band is overshoot’s signature — a sink slower than the source, so every mistake is banked.
Whose it is. Standard ecology; Catton’s 1980 book made it a civilizational argument. No single owner.
What it does not mean. Overshoot is not a prophecy of instant collapse — systems can run in overshoot for a long time, which is precisely the trap. And “carrying capacity” for humans is genuinely contested terrain (technology moves it; how far and how durably is the real debate), which is why our studies model specific stocks and sinks rather than assert planetary conclusions.
The unfolding. A pasture, a fishery, an aquifer, the atmosphere, an orbital shell: a commons is rival (my use subtracts from yours) but hard to fence (excluding users is costly or impossible). The dilemma is structural: each user captures the full benefit of their own use while the cost spreads across everyone, so individually sensible decisions sum to collective depletion. Garrett Hardin’s 1968 essay made “the tragedy of the commons” famous — and Elinor Ostrom’s Nobel-recognized fieldwork then showed the tragedy is not a law of nature: real communities have governed real commons for centuries, using recognizable design features (D9). The modern reading, and this site’s, is that the dilemma is real but the outcome is institutional — commons fail by default and survive by design. (Hardin’s framing is contested on the evidence, and his later writings carry positions many find objectionable; we cite the dilemma, not the man’s program.)
An example. Study 2’s whole subject: low-Earth orbit as a commons filling faster than anyone governs it, where the study’s governance finding is that everything currently in force supplies monitoring and norms — eyes and manners — while everything that would price, bond, or guarantee continuity exists nowhere. A commons with watchers and no ledger.
Whose it is. The dilemma’s formalization is standard; the escape is Ostrom’s lineage (Governing the Commons, 1990). Cross-link: D9 for the design principles, A4 for the race structure underneath.
What it does not mean. “Commons” does not mean “doomed.” That’s the single most common misreading, and the entry exists partly to retire it: the tragedy is the default trajectory of an ungoverned commons, and the interesting question — the one our studies ask — is what governance turns the default off.
Family B — the method vocabulary
No attribution owed; correctness is the whole burden. These are the terms a reader meets in the studies’ charts and method notes.
The unfolding. When a cost lands on an economy — a tax, a tariff, a price shock — the entity that pays it first is rarely the story’s end. Firms pass costs to customers, landlords to tenants, employers to workers; the cost migrates until it settles on whoever has the least room to dodge it. Incidence analysis is the discipline of tracing that migration and reporting where the burden actually comes to rest, usually broken out by income, age, tenure, or household type. It is the difference between “oil companies face higher crude prices” and “the bottom income bracket loses nine percent of its take-home pay” — the same event, described at the point of impact instead of the point of entry.
An example. The Uneven Month is an incidence study by construction: a Strait-of-Hormuz oil shock, traced through fuel, food, and shipping into household budgets by income bracket. The point-of-entry story is a barrel price; the incidence story is that the lowest-income households surrender roughly 8.9% of after-tax income and the highest roughly 1.6% — most dollars at the top, most pain at the bottom.
Whose it is. Standard public-finance vocabulary.
What it does not mean. Incidence is not blame. Finding that a burden settles on renters or the working poor says nothing yet about who caused it or what to do — it says where the weight is, which any honest response has to start from.
The unfolding. Almost every argument about who bears a cost turns on this distinction between dollars and shares, and it’s the reason two people can look at the same true numbers and reach opposite conclusions. A cost is regressive when it consumes a larger fraction of a smaller income, progressive when it consumes a larger fraction of a larger one, and proportional when everyone pays the same share. Energy and food costs tend to be regressive for a straightforward reason: they’re near-necessities with limited room to cut, so they occupy far more of a small budget than a large one — while, in absolute dollars, richer households usually pay more, because they consume more of everything.
An example. The Uneven Month found precisely this shape. The lowest-income bracket bore roughly 8.9% of after-tax income; the highest, roughly 1.6% — about a 5.5× difference in pain. Yet the largest share of the national dollar total came from higher-income households, because they spend more in absolute terms. Both facts are true at once, and the study’s headline exists to hold them together: most dollars at the top, most pain at the bottom.
Whose it is. Standard public-finance vocabulary. The summary measures we use for it (the Suits and Kakwani indices, B3) each come from named 1977 papers.
What it does not mean. “Regressive” is a description of a distribution, not a verdict on a policy. A regressive cost may still be worth bearing; a progressive one may still be badly designed. The word tells you who it lands on, and that’s a starting point for an argument rather than the end of one.
The unfolding. Once you’ve traced a burden’s incidence (B1), you want to compare it — to other burdens, across assumptions, over time — and for that you need a summary number. Both indices compress the whole income-versus-burden picture into one figure between −1 and +1: zero for a proportional burden, negative for regressive, positive for progressive. The Suits index compares the cumulative burden distribution against cumulative income; the Kakwani index measures the gap between the burden’s concentration and income inequality itself. The useful intuition: an index near −0.25 puts a burden in the territory of a gasoline tax — solidly regressive — and the sign is usually more robust than the second decimal. A summary index is only as honest as its inputs, which is why ours ship with confidence intervals and the income base stated (B4).
An example. The Uneven Month’s shock scores a pre-tax Suits of −0.26 (90% interval −0.31 to −0.21) and Kakwani −0.25; measured after-tax, −0.22 and −0.21. Across 100,000 bootstrap replicates (B9), the sign never flipped: regressive in every single one. The indices also proved invariant to how big the shock gets — the size of the hit moves, the shape of who bears it doesn’t.
Whose it is. Daniel Suits (1977) and Nanak Kakwani (1977), separate papers, both standard tools since.
What it does not mean. An index is not the distribution — two different burden shapes can share a score. It’s a headline, and like all headlines it’s for comparison, not for substituting the picture it summarizes.
The unfolding. A burden’s share-of-income depends on the denominator. Divide by pre-tax income and you’re comparing against a bigger base (especially at the top, where taxes take more); divide by after-tax income and every share grows, but unevenly. Neither is “the” right base: pre-tax is the standard for comparability with the tax-incidence literature; after-tax better reflects what a household actually has available to absorb a shock. What is non-negotiable is disclosure — an incidence claim that doesn’t name its base isn’t checkable, and comparing a pre-tax number against an after-tax one manufactures phantom disagreements.
An example. This entry exists because of a debugging session. Early in The Uneven Month’s review, two Suits indices refused to reconcile — −0.26 versus −0.22 — and the gap was chased for a while as a suspected grouping error. The real cause was the income base: one number was pre-tax, the other after-tax, and both were correct. The house rule (lead pre-tax, report after-tax alongside, label everything) is that lesson, institutionalized.
Whose it is. House convention, following standard practice in the incidence literature.
The unfolding. Ask “how big will the burden be?” and the honest answer is not a number but a shape: many possible outcomes, each with a weight. The predictive distribution is that shape. Reporting it beats reporting a point estimate for three reasons. It shows the spread — how wrong the central number could plausibly be. It shows the shape — our floor distribution turned out multimodal (several distinct humps, one per geopolitical scenario), which no single interval can convey; roughly 28% of its probability mass sat outside the old best-case-to-worst-case envelope. And it lets you decompose the uncertainty: for The Uneven Month, about 82% of the floor spread came from which scenario realizes — a structural, geopolitical question — and only the residue from measurement noise, which tells you exactly where better data would and wouldn’t help.
An example. The study’s December run-rate: median about $23B/month, 90% interval roughly $4B to $49B. The width isn’t hedging — it’s the finding that the outcome hinges on an unresolved war, and the sliders on the study page let you re-weight the scenarios and watch the whole distribution answer.
Whose it is. Standard statistics.
What it does not mean. A distribution is not a forecast of any single future — it’s conditional on the model and its stated assumptions. p99 events are inside it; so are pleasant surprises. The claim is “this is the honest shape of what could happen given what we assumed,” never “this is what will happen.”
A term of ours. We define it here because we use it, and because a house coinage that isn’t defined is jargon with a marketing problem.
The unfolding. For some questions you have enough history to estimate uncertainty honestly: the past behaved a certain way, and the future plausibly behaves similarly. That produces a floor — a defensible lower bound on how uncertain things are. But some risks aren’t in the historical record, because the coupling that would produce them hasn’t been triggered yet: things going wrong together rather than independently, or a shift into a regime the data never sampled. Those live in a second, coupled band, built by explicitly turning on the couplings and stating their assumed strength. Reporting only the floor understates the risk; reporting only the stress band overstates the confidence we have in its knobs. So we report both — and the distance between them is the honest statement: it measures how much of your exposure depends on assumptions the data can’t settle.
An example. In The Uneven Month, the floor’s 90% interval for the monthly burden ran to roughly $49B. Turning on the couplings moved the extreme tail (the 99th percentile) to about $69B. That ~$14–20B gap isn’t a hedge or a rounding — it is the result: the amount of exposure that exists only if things go wrong together, and which no amount of historical data could have told you about.
Whose it is. The two-band presentation is ours; the underlying machinery (regime-switching models, tail-dependent copulas, fat-tailed marginals) is standard in the risk literature, and the motivating distinction is Knight’s, between measurable risk and genuine uncertainty (see B7).
What it does not mean. The upper band is not a forecast, and it is not “the real number.” It’s a conditional statement: if the couplings are as assumed, the tail lives here. The knobs are published so anyone can move them and get their own band.
The unfolding. Frank Knight’s 1921 distinction: risk is when the odds are knowable (dice, actuarial tables, well-sampled histories), and uncertainty is when they aren’t — when the event is novel enough, or the system changed enough, that any probability you assign is a judgment call rather than a measurement. The distinction matters because methods built for risk quietly fail under uncertainty: a single confidence interval on a deeply uncertain quantity launders a guess into something that looks like a measurement. The honest responses are structural — report conditional results (“if the couplings are X, then…”), sweep the assumptions rather than pick one, name the few knobs that actually move the answer, and keep the distinction between what’s estimated and what’s assumed visible all the way to the surface. Our two-band convention (B6) is this discipline made into a house format.
An example. The Uneven Month’s crisis-regime knobs — how likely a cascade year is,
how hard it hits — have no settled historical odds; the study publishes them as assumptions
with sensitivity sweeps. The orbital study goes further: its collision-per-close-pass
fraction (f_imp) is verifiably unmeasurable from a zero-event record, so every trajectory
magnitude is labeled conditional on it.
Whose it is. Frank Knight, Risk, Uncertainty and Profit (1921); “deep uncertainty” is the modern decision-methods literature’s term for the same territory.
What it does not mean. Deep uncertainty is not an excuse to say nothing, and not a license to say anything. The move it forbids is false precision; the moves it demands are bounds, conditions, and named assumptions — which is more work than a point estimate, not less.
The unfolding. When a result depends on several uncertain inputs at once, algebra gets you an average but hides the range. The Monte Carlo method — named for the casino — brute- forces the question instead: draw each uncertain input at random from its stated distribution, run the model, record the outcome, and repeat tens of thousands of times. The pile of outcomes is the answer: its median, its spread, its tails, its humps. The method’s honesty lives in two disciplines. First, the input distributions must be stated — a Monte Carlo is exactly as good as what you feed it, and “garbage in, garbage out at scale” is the standing failure mode. Second, the randomness must be reproducible: our engines run from fixed seeds (B15), so every one of the hundred thousand draws can be regenerated by anyone, exactly.
An example. Both studies run on this machinery. The Uneven Month draws 200,000 scenario-months to price an oil shock’s tail; the orbital study draws 100,000 parameter sets to ask how often the trigger band’s physics comes out supercritical without maneuvers — 98% of draws, a parameter-belief statement the study is careful to distinguish from an event probability.
Whose it is. Standard method; the name and modern form trace to Ulam, von Neumann, and Metropolis at Los Alamos.
What it does not mean. A hundred thousand runs of a wrong model are a precisely wrong answer. Monte Carlo quantifies the uncertainty you told it about — it cannot surface the uncertainty you left out, which is why the input assumptions, not the run count, are where scrutiny belongs.
The unfolding. You computed a statistic — an average, an index, a share. How much should you trust it? The bootstrap answers by resampling: draw a new dataset of the same size from your own data (sampling with replacement, so some rows repeat and others drop), recompute the statistic, and repeat thousands of times. The spread of the recomputed values estimates the spread you’d have seen across alternative datasets — a confidence interval built from the data’s own variability, without assuming any textbook distribution. The method’s elegance is that it works for statistics too gnarly for clean formulas, which is most of the interesting ones.
An example. The Uneven Month’s regressivity finding runs on 100,000 bootstrap replicates: the Suits index’s 90% interval (−0.31 to −0.21) comes from the resampling, and the sharpest sentence in the study — regressive in 100% of replicates — is a bootstrap statement: not one resample in a hundred thousand flipped the sign.
Whose it is. Bradley Efron (1979).
What it does not mean. The bootstrap sees only the data you have — it quantifies sampling wobble, not the possibility that the data itself is unrepresentative or the model misspecified. Our studies label which one is being reported (modeled-input versus survey-sampling uncertainty) for exactly that reason.
One-liner entry (depth lives in the papers). The statistical machinery for asking: when one thing goes badly wrong, how often do the others go wrong with it? A copula couples individually-known uncertainties into a joint story, and “tail dependence” is the dial for whether disasters cluster. The Uneven Month’s stress band runs on it — with the finding that the copula’s exact family mattered far less than the crisis-regime knobs, a useful deflation of the fanciest ingredient.
The unfolding. Percentiles are how a distribution (B5) gets quoted in words. Line up every simulated outcome from best to worst: p50 is the median — half land below, half above; p05 and p95 bracket the central 90%; p99 is the outcome only 1 in 100 exceeds. Two reading disciplines keep them honest. First, a percentile is a rank, not a prediction — “p99 ≈ $69B” says one simulated month in a hundred is at least that bad, not that such a month is coming. Second, the pair of a percentile and its distribution is the unit of meaning: the same “p99” label sits on very different dollar figures depending on whether the couplings are on (B6), and quoting the number without the band is how scare headlines get manufactured from careful work.
An example. The Uneven Month reports the December burden’s floor p99 near $49B and the coupled p99 near $69B — same model, same month, different assumed togetherness of things going wrong. The study’s plain-language layer translates every such label (“worse than 99 of 100 simulated months”) because a reader shouldn’t need this glossary to survive a chart.
Whose it is. House translation convention; the mathematics is elementary statistics.
What it does not mean. p99 is not “the worst case” — p99.9 is worse, and the distribution’s tail keeps going. Nor is it a schedule: nothing about a 1-in-100 label says when. It’s a ruler mark on a range of possibilities, nothing more.
The unfolding. A cost that builds over time can be quoted three ways: the run-rate (what this month costs, once the price rise has fully built up), the cumulative (the total added up over the whole window so far), and the annualized (a rate scaled to twelve months, useful for comparison, dangerous when mistaken for a total). All three are legitimate; silently swapping them is how numbers lie while being technically true. The classic failure is multiplying a late-month run-rate by twelve and calling it the year’s damage — double-counting the ramp-up months that were smaller.
An example. This discipline is written in this project’s origin story. The study that became The Uneven Month began when a circulating claim applied an annual inflation rate as if it were monthly — overstating a real burden roughly fifteenfold. The published study quotes a December run-rate of about $24.7B/month alongside a June–December cumulative of about $110B, labels both, and states the multiplier connecting them — because the difference between those framings was the whole error it exists to correct.
Whose it is. House convention (charter rule); the underlying distinction is ordinary accounting.
What it does not mean. The run-rate is not “the cost of the year,” and the cumulative is not “the ongoing burden.” Each answers a different question; the label is load-bearing.
The unfolding. Every quantitative claim has a pedigree, and the tags make it visible at the point of use. MEASURED (or SOURCED): read from the world — a catalog pull, a government series, a cited primary. MODELED (or DERIVED): computed by an engine from stated inputs; only as good as those inputs. ASSUMED: chosen by us because the world hasn’t settled it — always shipped with a sensitivity sweep, never dressed as data. DESIGN: an analysis specified but not yet executed — a promissory note, labeled as one. The tags are the last line of defense against the commonest quantitative sin: a modeled or assumed number reaching a reader dressed as a measured one.
An example. The orbital study’s five-bases table wears the tags in public: Starlink’s 38.7% of tracked LEO objects is MEASURED (catalog), its 66.1% of active payloads SOURCED (census), the 97.9% share of no-maneuver collision rate DERIVED (engine) — five honest answers to “how big,” each with its pedigree, never mixed.
Whose it is. House convention, doing standard epistemics in public.
What it does not mean. MODELED doesn’t mean unreliable and MEASURED doesn’t mean beyond question — catalogs have epochs, sources go stale (one of our mass-share bases is stale-flagged for exactly this). The tags don’t rank trustworthiness; they tell you what kind of checking each number invites.
One-liner entry (depth lives in the papers). The workhorse designs for asking whether X caused Y rather than merely preceded it: compare the affected group against a control group, before and after, with checks that the two were tracking together beforehand. The Uneven Month specifies one for its debt-triage question and — the discipline being the point — labels it a design, not a result until the microdata exists to run it.
The unfolding. A simulation’s “random” draws come from a pseudo-random generator that is perfectly deterministic once you know its starting state — the seed. Publishing the seed, the engine, and the raw draw vectors turns “trust our Monte Carlo” into “check our Monte Carlo”: re-run the engine and you get byte-identical results; load the raw draws and you can recompute any percentile yourself without trusting our arithmetic at all. It also enables the stronger check our review process requires: an independent reimplementation on a different seed, which should reproduce the findings’ shape even though no individual draw matches — separating “the code says what the paper says” from “the result is real.”
An example. Every engine in both studies’ Floors records its seed in its own results file (the orbital study’s run from seeds 20260715–20260720 and 26716, each with raw draw samples shipped), and each study page carries the same standing offer: if a number in the report and a number in these files disagree, the files win — and we want to know.
Whose it is. House convention; reproducible-research practice generally.
What it does not mean. Reproducibility is not correctness — a wrong model reproduces perfectly. It removes one excuse (“you can’t check us”) so scrutiny can reach the part that matters: the assumptions.
The unfolding. Donald Kessler and Burton Cour-Palais argued in 1978 that beyond a critical density of objects, collisions produce fragments faster than atmospheric drag cleans them out, and the debris population grows under its own power — a chain reaction in slow motion. Three qualifications keep the term honest. It’s band-specific: altitude decides everything, because the thin air that flushes a 480-km shell in years does almost nothing at 900 km, where mistakes persist for a century or more. It’s slow: even a triggered cascade unfolds over years to decades — a ratchet, not an explosion. And criticality is conditional: our study’s central physical finding is that the most crowded band is subcritical only by virtue of collision avoidance — the engine reads ρ* ≈ 0.14 with everyone dodging and ≈ 5.8 if everyone stopped, meaning the calm is operational, not physical.
An example. Study 2 exists to put honest error bars on this term: which shells self-clean and which ratchet, how far the trigger band sits from its threshold, and what share of that distance is manufactured daily by one operator’s autopilots.
Whose it is. Kessler & Cour-Palais, “Collision Frequency of Artificial Satellites: The Creation of a Debris Belt” (1978) — the primary.
What it does not mean. Not a movie scene — no wall of shrapnel, no hours-long apocalypse. And not a present-tense description: today’s orbit is not “in Kessler syndrome.” The term names a threshold condition, and the study’s discipline is to always say which band, under which assumptions, is how far from it.
Family C — the site’s own furniture
Terms we coined for this site’s machinery, defined so nobody has to guess.
The unfolding. The canonical definitions live on the homepage’s evidence ledger, and this entry deliberately points there rather than restating them (two wordings of the same tiers would eventually disagree — that divergence would be a bug, and on this site bugs get logged publicly). The short version: Established — well-supported by replicated evidence or settled mechanism; Contested — real support and real dispute, with the disagreements named rather than buried; Exploratory — promising, early, and labeled as such. The tier colors (green, amber, blue — never a red/green pairing, for colorblind safety) recur across the site, including in the corrections system (C4).
An example. The orbital study’s avenues table runs the discipline in its sharpest form: everything currently in force up there tags as Demonstrated-but-toothless monitoring and norms, while everything that would price, bond, or guarantee continuity tags as Modeled — zero implementations anywhere. The tags are the finding.
Whose it is. House furniture.
The unfolding. Every study page has a surface a non-specialist can read, and a floor directly beneath it holding what a specialist needs to check it: the seeded engines (B15), the recorded results, raw draw samples, the live workbook, and the research-footing paper. The Floor is why the site’s claims are offers rather than assertions — any reported percentile recomputes from the shipped draws without trusting our arithmetic. The closing rule is the epistemic point: the report is a rendering of the files, so when they diverge, the files are the truth and the divergence is a bug we want reported (C4).
An example. The Keys to Orbit ships a 27-file Floor: six engines with recorded seeds, their results and raw draws, the workbook (35 formulas, 0 errors), and the 24-page paper. The reading edition’s own methods line repeats the rule verbatim.
Whose it is. House furniture; the practice is reproducible-research standard, made reader-facing.
What it does not mean. The Floor is not an appendix dump — each file is named, described, and load-bearing (a cited statistic traces into it). Nor is it a paywall’d extra: it publishes with the study, same URL, same day.
The unfolding. The Roundtable’s studies are produced by convened seats: an Author who builds and integrates, and a rotating bench of expert seats — economics and statistics, game theory, institutions, earth systems, complex systems, technology governance — each holding verified judgment in one lens. A seat owns a lens, never a topic: no seat “owns” quantification, and no phenomenon is off-limits to any seat’s toolkit. Studies list which seats sat, because the bench is part of the method: the economics referee adversarially reimplements the engines; the domain seats rule on what their literatures do and don’t support. And the honest line the reader deserves in full: the seats are AI agents, human-orchestrated, not autonomous — a human editor convenes the table, sets the questions, relays every dispatch, and decides what publishes. The architecture is public on the Method page; the disclosure is the credential, not the caveat.
An example. The Keys to Orbit convened six seats. The referee seat’s independent reimplementation of the race model — fresh seed, fresh code, agreement to ~10⁻¹⁵ — is what “refereed” means on this site.
Whose it is. House furniture; the name is the site’s own.
What it does not mean. Not a claim that AI involvement is invisible or incidental — the opposite: it’s disclosed because a site arguing for better sensemaking doesn’t get to obscure its own production. And not autonomy theater: nothing publishes without the human editor’s explicit decision.
The unfolding. The corrections form asks which study, where, and how bad: Presentation — wording, styling, a chart’s legibility; the numbers stand. Sourcing — a citation is dead, wrong, or unfairly attributed; the claim needs its paperwork fixed. Substantive — a number, method, or conclusion is challenged; the underlying data gets re-examined and, if the challenge stands, the study gains a dated update (published studies are never silently rewritten — the update protocol is append-only, and the original stays visible). Every accepted correction lands in the public log with date, root cause, and credit if the reporter wants it. This page’s own definitions are explicitly in scope: a bad definition or dead link here is a Presentation or Sourcing report like any other.
An example. The site’s founding correction predates the site: the fifteenfold annual-as-monthly inflation error (B12) that The Uneven Month was built to correct — caught by a reader of the original claim, not by an institution.
Whose it is. House furniture; the tier names are ours.
What it does not mean. The log being short is not evidence nothing is wrong — it’s an invitation with a service-level promise attached. The empty state says “this space is public on purpose.”
Family D — the bench’s references
Every seat card names its intellectual lineage; these entries are the way in. Named thinkers are named, never absorbed.
The unfolding. Every energy system pays an energy toll: rigs, mines, refineries, turbines, and grids all consume energy to deliver it. The surplus is what remains after that toll — and it, not the gross production number, is what funds everything that isn’t energy-getting: agriculture, medicine, education, art, leisure, governance. The concept is the flow-side twin of EROI (A10, the ratio); together they carry biophysical economics’ central discipline — read the economy in energy and materials first, money second. A society’s surplus can shrink even while gross output holds steady, if the toll rises; the squeeze shows up everywhere downstream while remaining invisible in dollar accounting.
An example. (Outside example, pending a study of ours.) The standard illustration is historical: pre-industrial societies ran on agricultural surpluses thin enough that most people had to farm; the fossil windfall inverted the ratio, freeing the majority for everything else. The Simplification thesis (A11) is, at bottom, an argument about that ratio’s direction of travel.
Whose it is. Biophysical economics — Hall’s lineage; Hagens’ usage in the bench card.
What it does not mean. Not a call to maximize energy production — burning the furniture raises today’s surplus at tomorrow’s expense. The analytical content is the distinction between gross and net, and the reminder that the net is what civilization actually spends.
The unfolding. Money circulates; matter flows through. An economy takes in ordered matter and energy (ores, fuels, food, timber), transforms them, and expels degraded residues (waste heat, emissions, landfill, tailings). That one-way physical flow is throughput, and Herman Daly built ecological economics on the observation that it — not GDP — is what the biosphere actually experiences. Two economies with identical GDP can have wildly different throughput; efficiency shrinks the ratio between them but has never yet shrunk the absolute flow at global scale. Daly’s point, and this site’s reason for carrying the term, is that throughput is bounded by sources and sinks (A12, D19) in a way that money, being a social convention, is not.
An example. The orbital study’s Interlude measures a literal, novel throughput: the mass of satellites burning up in the atmosphere per year, benchmarked against the ~15,000 tonnes of natural meteoric infall — an industrial material flow routed through a layer of the sky that never carried one before, with its chemistry (reentry metals in stratospheric aerosols) already measurable.
Whose it is. Herman Daly, ecological economics.
What it does not mean. Not a synonym for “economic activity.” The whole point is the non-identity: services, software, and finance can grow money faster than matter, but they ride on a material base — and the claim that growth has “dematerialized” is an empirical question about throughput, not something the GDP series can settle by itself.
The unfolding. This is a specific, citable scientific framework, not a general phrase about environmental limits. A group of Earth-system scientists proposed that a set of global processes — climate, biosphere integrity, land use, freshwater, biogeochemical flows (nitrogen and phosphorus), ocean acidification, atmospheric aerosols, stratospheric ozone, and novel entities such as synthetic chemicals — each have a quantified “safe operating space,” with a boundary placed conservatively before the point where the science expects nonlinear or irreversible change. The framework’s power is that it puts numbers on the ceiling; its main criticism is that several boundaries are genuinely hard to quantify at global scale, and that a global number can hide the fact that the harms and the causes are unevenly distributed.
An example. The ozone boundary is the one that shows the framework working in the good direction: a global ceiling identified, an agreement reached, and a boundary that moved back inside the safe range. It’s also directly relevant to our orbital work, where satellite-reentry chemistry touches stratospheric ozone.
Whose it is. Rockström, Steffen and colleagues; maintained and updated by the Stockholm Resilience Centre. How many boundaries are currently assessed as crossed has changed with each update — the entry must carry the current figure and its date, verified against the primary assessment, never a number remembered from press coverage. Resolved at build (2026-07-27, live check): seven of nine transgressed, per the 2025 Planetary Health Check (ocean acidification newly the seventh), as carried on the Stockholm Resilience Centre’s planetary-boundaries page.
What it does not mean. Crossing a boundary isn’t a cliff-edge event on a date. The boundaries are placed before the danger zone precisely so that crossing one means “we are now running an uncontrolled experiment,” not “the disaster has occurred.”
The unfolding. Founded at Royal Roads University under Thomas Homer-Dixon — whose earlier work on environmental scarcity and conflict, and on what he called the ingenuity gap, anticipated much of the current framing — the Cascade Institute took a word being used as a mood (“everything’s connecting!”) and gave it definitional discipline: which crises count as systemic, what qualifies as an interaction, how amplification is traced through named causal pathways. Our Complex-Systems seat inherits its standard directly: the word polycrisis (A2) is only earned when the coupling mechanism is named. The institute also works the solution side, hunting for high-leverage intervention points (D6) in coupled global systems.
An example. The house discipline in practice: The Uneven Month’s coupled band names its channels — energy, shipping, food, and credit stressed together, with the coupling strengths stated as assumptions — rather than asserting “cascading crisis” as an atmosphere.
Whose it is. A named institution; entry exists because the bench card names it.
The unfolding. C.S. Holling, watching forests and fisheries, found a recurring four-phase rhythm: rapid growth (resources abundant, pioneers everywhere), then conservation (the system matures, optimizes, and accumulates — becoming efficient, interconnected, and brittle), then release (a shock — fire, pest, crash — that the rigid structure can no longer absorb), then reorganization (the freed-up material and survivors recombine into something new, and the cycle restarts). Two of his insights carry the weight. Efficiency and resilience trade off: the conservation phase’s optimization is exactly what makes the release violent. And resilience is not toughness — it’s the width of the basin (D17): how far the system can be pushed and still find its way back, a quantity you can spend without noticing until it’s gone.
An example. (Outside example, the field’s own.) Fire-suppressed forests: decades of successful suppression accumulate fuel, deepening the eventual burn — safety optimized in the small, fragility purchased in the large. The pattern generalizes uncomfortably well to financial systems and supply chains, which is why the seat carries it.
Whose it is. C.S. Holling; the adaptive cycle and its nested “panarchy” extension.
What it does not mean. The cycle is not a prophecy — not every system is one phase from collapse, and the framework’s value is diagnostic (where in the cycle is this system? what’s its basin width?) rather than predictive. “Resilience” has also become a buzzword meaning roughly “good”; Holling’s version is measurable, and can be too high — a dysfunctional regime can be resilient too.
The unfolding. Donella Meadows’ famous ordering runs from weak interventions to strong ones: adjusting parameters (rates, subsidies, standards) moves a system least; strengthening or weakening feedback loops (D7) moves it more; changing information flows (who can see what, and when) more still; changing the rules (incentives, constraints, who decides) more again; and changing the system’s goal — or the paradigm the goal comes from — moves everything. Her uncomfortable observation: public argument concentrates at the bottom of the list (tweak the rate, adjust the subsidy) precisely because the top is harder to see and harder to touch. The list is a diagnostic for wasted effort — and a warning that systems push back hardest exactly at the high-leverage points.
An example. The orbital study’s avenues sort themselves onto her ladder. Orbital-use fees adjust a parameter — and the model finds the static version nearly invisible (~$231/satellite-year). Shared traffic awareness is an information-flow intervention. The per-shell cap and the continuity regime are rule changes — and the study’s tested one-sentence conclusion, no single instrument unpicks both games, is a leverage-points finding: the games live at different rungs.
Whose it is. Donella Meadows — Thinking in Systems and the “Leverage Points: Places to Intervene in a System” essay; she was lead author of The Limits to Growth.
What it does not mean. Not “parameters never matter” — a carbon price is a parameter and still worth fighting over. The claim is comparative: if the goal and rules stay pointed the wrong way, parameter victories get quietly reabsorbed.
The unfolding. Two species, and everything in systems behavior grows from their combination. A reinforcing loop compounds: more debris means more collisions means more debris; more attention means more visibility means more attention. Left alone, it runs exponentially — until it meets a limit. A balancing loop corrects: a thermostat, a predator-prey equilibrium, a price mechanism — deviation generates its own opposition. Real systems are tangles of both, and their behavior is decided by which loop dominates where — a question of gains, not vibes. That’s why the interesting analysis is never “is there a feedback loop?” (there always is) but “what’s the gain, where does dominance flip, and what moves the flip point?”
An example. The orbital study’s engine is, at heart, two loops in competition: the collision-production loop (debris begets debris — reinforcing) against the drag-removal loop (thin air flushes fragments — balancing). Every headline in the study is a statement about which loop wins in which altitude band: the corridor where drag dominates in years, the ratchet band where it loses for a century, and the criticality ratio ρ* that names the flip point.
Whose it is. Standard systems vocabulary — Meadows’ Thinking in Systems is the accessible canon.
What it does not mean. “Feedback” here isn’t commentary (as in “we welcome your feedback”), and a reinforcing loop isn’t inherently bad — compound learning and compound interest are the same mathematics. The moral valence comes from what’s compounding.
The unfolding. A cascade is failure with a transmission path: the outage that overloads the neighboring grid, the default that breaks the counterparty, the collision whose fragments cause the next collision. What separates the term from hand-waving is that a cascade names its coupling — the specific channel by which failure A raises the probability of failure B — and that named channel is what makes the claim testable and the intervention findable (cut the coupling, stop the cascade). Our Complex-Systems seat enforces this as house discipline: no “cascading crises” without the mechanism stated. Cascades also explain why tail risks cluster (B10): couplings that are invisible in calm times activate under stress, which is exactly when everything else is also failing.
An example. Both studies are cascade studies wearing different clothes. The orbital one models the literal case — fragments as the coupling channel between collisions, with the atmosphere as the cascade-breaker that works at 480 km and fails at 900. The household study’s stress band models the financial version: energy, shipping, food, and credit failing together through named channels rather than independently.
Whose it is. Standard complex-systems vocabulary; the Cascade Institute (D4) took the word for its name.
What it does not mean. Not domino inevitability. Cascades have gains like any loop (D7): most perturbations die out, and the analytical question is what pushes the transmission ratio above one. “Everything is connected” is where the analysis starts, not where it ends.
The unfolding. While the tragedy of the commons (A13) was hardening into dogma, Ostrom went and looked: Swiss alpine pastures, Spanish irrigation courts, Japanese village forests, Maine fisheries — commons that had survived for centuries without privatization or state control. From the field record she distilled the recurring design features: clearly defined boundaries (who may use the resource); rules matched to local conditions; the users themselves participating in making the rules; monitoring by people accountable to the users; graduated sanctions (she found that instant harsh punishment bred resentment and collapse — enduring commons escalated gently); cheap, fast conflict resolution; the right to organize without outside veto; and, for large systems, nested layers of governance. The 2009 Nobel in economics recognized the body of work. One sentence of framing this page owes the reader: there’s no single-player interactive for this entry, and that’s the lesson — Hardin’s tragedy can be shown to one person, but Ostrom’s escape can’t, because her mechanisms are communication, monitoring, and rule-making, all of which need somebody else in the room.
An example. The orbital study reads its commons against her list and finds the scaffolding half-built: operator data-sharing pacts are her bottom-up self-organization in embryo; shared traffic awareness is monitoring; and what’s missing — graduated sanctions with teeth, rule-making that binds the biggest user — maps precisely onto the principles not yet satisfied. The study’s “eyes and manners, no ledger and no will” is an Ostrom diagnosis.
Whose it is. Elinor Ostrom, Governing the Commons (1990); the Ostrom Workshop at Indiana University maintains the tradition. Three links with honest labels at build: play the mechanism — The Evolution of Trust (Case; teaches why repetition and memory beat defection, the engine under her principles); run it with a group — Fishbanks (MIT Sloan; facilitator-run, for teachers and teams); read the principles — the Ostrom Workshop.
What it does not mean. Not a recipe — Ostrom herself resisted “panaceas” as the exact error her work corrected. The principles describe what successful commons had, not steps that guarantee success; scaling them from a village fishery to a global commons is the open problem, not the solved one.
The unfolding. The mechanism is ordinary and the name is dramatic. Consider any competition where restraint costs you: the fishing boat that takes less, the campaign that spends less, the lab that ships slower, the country that arms less. Each participant faces the same arithmetic — if I hold back and others don’t, I simply lose and the harm happens anyway — so everyone proceeds, and the collective outcome is one that no individual wanted. What makes this different from ordinary selfishness is that it doesn’t require anyone to be selfish. Every participant can see the outcome coming, hate it, and still be unable to unilaterally avoid it. That’s what the name is doing: personifying a dynamic that consumes what everyone values while no one is choosing it. Our Game Theory seat exists to make a specific distinction here (see D12): a genuine trap — where defecting really is each player’s best move regardless of others — takes different medicine from a coordination failure, where cooperating would be everyone’s best move if only they could trust each other to do it. Mistaking one for the other prescribes the wrong fix.
An example. The satellite race we study has this structure with a wrinkle: filing early costs little and reserves valuable orbital space, so everyone files; the resulting congestion is in nobody’s interest, including the filers’. Our race model exists precisely to ask whether that race is a genuine trap or a coordination failure — because the answer determines whether you need enforcement or merely a credible way to see what everyone else is doing.
Whose it is. Three layers, all owed, because this term reached most people through a chain rather than a single source. The image comes from Allen Ginsberg’s 1955 poem Howl, which invokes Moloch as a devouring force — we paraphrase the idea and do not reproduce the poem. The game-theoretic usage is Scott Alexander’s 2014 essay “Meditations on Moloch,” which recast it as the god of multipolar traps. And the transmission to a general audience is largely Liv Boeree’s — the former professional poker player and physicist whose films, talks, and interviews carried the idea far outside its original readership; she credits Alexander explicitly as the one who put it in game-theory terms, and describes her own “Moloch trap” as a rebranding of the multipolar trap (A4), with her films widening the emphasis to misaligned incentives generally. She is also, usefully for this site, the source of the sharpest guard against the term’s misuse: hers is the god of unhealthy competition specifically, competition itself being neutral.
Where to start. For most readers, Boeree’s short films are the better entry point than the original essay — visual, twenty minutes rather than ninety, and built for people who have never met a Nash equilibrium. The essay is the deeper stop for anyone who wants the argument in full. (Editor’s note: this is how the Editor first encountered the term, which is itself evidence about which door a general reader is likely to come through.)
What it does not mean. It doesn’t mean “capitalism,” “competition,” or “the bad guys.” Plenty of competition produces good outcomes — Boeree’s own framing is emphatic that competition is neutral and that only the badly-structured kind is Moloch. The term names the specific structure where restraint is individually punished. Using it as a general-purpose villain drains it of the analytical content that makes it useful — and, being borrowed from an online intellectual subculture, it can read as an in-group signal. We define it here so nobody needs to be in the group to follow the argument.
The unfolding. This is the phrasing a general reader is most likely to have actually met — Liv Boeree’s coinage for the multipolar trap (A4), built for transmission: where “Moloch” (D10) personifies the dynamic and “multipolar trap” formalizes it, “Moloch trap” packages it as a countable noun. You can say “the attention economy is a Moloch trap” and be understood by someone who has read neither Ginsberg nor game theory. Boeree describes it explicitly as a rebranding of the multipolar trap, with her films widening the emphasis to misaligned incentives generally — beauty filters, news sensationalism, AI capability races. The three terms are one concept at three levels of dress; this site uses whichever the sentence needs and defines all three so the reader is never locked out.
An example. Same structure, same study: the filing race of the orbital commons — with the model’s standing caveat that whether it’s a genuine trap is conditional on the business case being real (A4, D12).
Whose it is. Liv Boeree’s rebranding of the multipolar trap; she credits Alexander for the game-theoretic framing.
What it does not mean. Everything D10’s guard says, inherited whole: not competition per se, not capitalism, not a villain — the specific structure where restraint is individually punished and collectively necessary.
The unfolding. Mechanism design’s core discipline: when you propose a rule, assume everyone subject to it will keep optimizing for their own interest under it — and check whether the rule still produces the outcome you wanted. If it only works when participants act against their own incentives, it isn’t a mechanism, it’s a hope. The test sounds cynical and is actually the opposite: it takes people as they are and asks the rule to do the moral work. The concept anchors a Nobel-recognized literature (Hurwicz, Maskin, Myerson, 2007) on designing auctions, matching systems, and institutions that perform under self-interest rather than despite it.
An example. The orbital study’s bond analysis is a worked lesson in the difference. The bonds that exist today are deployment bonds — forfeited if you fail to launch enough satellites — which makes them accelerant machinery: the incentive-compatible behavior they produce is more hardware in orbit. Invert the trigger into a disposal bond — forfeited if you fail to clean up — and size it to the actual ~$71k disposal cost, and the model’s compliance outcome flips from roughly 0% to roughly 100%. Same instrument class, opposite incentive, opposite sky.
Whose it is. Mechanism-design standard.
What it does not mean. Not “people only respond to money” — incentives include reputation, mission, and belonging, and mechanisms can harness those too. And not a counsel of cynicism: the point is that durable altruism is designed for, not assumed.
The unfolding. In a genuine trap (prisoner’s-dilemma structure), defecting is each player’s best move no matter what the others do — even perfect communication doesn’t fix it, because everyone would still rather defect. The medicine is structural: enforcement, binding commitments, or payoff changes (bonds, fees, penalties) that make cooperation individually rational. In a coordination failure (stag-hunt structure), cooperating is everyone’s best move — provided they trust the others to cooperate too; the medicine is cheaper and gentler: transparency, assurance, credible signals, a focal point. Misdiagnosis wastes effort in both directions: treating a trap with transparency accomplishes nothing (everyone can see, and still defects), while treating a coordination failure with heavy enforcement pays for machinery that trust-building would have replaced.
An example. This distinction is the orbital race model’s actual research question. The answer it returns is conditional: across most of the assumed payoff space the race simply fizzles (paper never becomes hardware — neither disease); where the business case is real, the deployment layer behaves trap-like at the margin — which is why the study’s avenues lean on payoff-changing instruments (bonds, holding prices, caps) rather than on awareness alone, while still funding the awareness layer that a coordination-failure diagnosis would make sufficient.
Whose it is. House framing over standard game theory (the dilemma/stag-hunt contrast is textbook).
What it does not mean. Not that every collective problem is one or the other — real systems mix structures across players and layers, and the same arena can be a trap at one margin and a coordination failure at another. The point is to ask which, per margin, before prescribing.
The unfolding. The McLuhan–Postman tradition’s claim is that a communication medium’s form does more work than its content. A culture that gets its public discourse through long printed arguments practices sustained attention, sequence, and logic; one that gets it through images and feeds practices immediacy, emotion, and juxtaposition — regardless of what any particular book or post says. “The medium is the message” (McLuhan) and Postman’s gentler formulation — every technology carries an epistemology — both point at the same discipline: ask not “is this content true?” but “what habits of mind does this form train?” For a site about collective sensemaking (A5), the tradition supplies the structural half of the diagnosis: the information environment isn’t failing because bad content got in; its architecture rewards different cognitive habits than the ones democratic reasoning needs.
An example. (Outside example.) Postman’s own: the televised presidential debate, judged on composure and delivery — capacities of the screen — where its print-era ancestor was judged on argument, a capacity of the page. Same institution, different medium, different selection pressure on what a leader is.
Whose it is. Marshall McLuhan’s coinage popularized through Neil Postman, who built the academic field (Amusing Ourselves to Death is the accessible classic).
What it does not mean. Not nostalgia for print, and not technological determinism — the claim is that form biases, not that it dictates. And it’s not a conspiracy theory: the environment shapes thought without anyone needing to intend it, which is exactly what makes it a structural problem rather than a villain story.
The unfolding. Herbert Simon saw it in 1971: information consumes attention, so abundance of the one creates scarcity of the other. The modern attention economy is that observation industrialized — platforms whose revenue scales with engagement, competing for a finite pool of human hours with algorithmic feedback loops (D7) optimized against whatever holds a gaze. The structural consequence, and the reason this vocabulary sits in our Cognition seat: the qualities that capture attention (novelty, outrage, tribal confirmation, fear) correlate only loosely with the qualities sensemaking needs (accuracy, proportion, context). No one at any platform needs to prefer falsehood; the selection pressure does the work — a multipolar trap (A4) where the racing firms are optimizing engagement and the commons being depleted is collective attention itself.
An example. (Outside example.) The natural experiment every reader has run: the same news event consumed via a wire-service report versus via a feed — one optimized for being checkable, one for being unputdownable. The difference in your pulse is the business model, felt from inside.
Whose it is. The term is widely used; the origin insight is Herbert Simon’s (attention scarcity, 1971).
What it does not mean. Not “all media is lies” and not “attention to things is bad.” The claim is about incentive structure, not content: an attention market can reward quality where audiences and institutions make quality legible — which is partly what corrections systems, evidence tiers, and this page are for.
The unfolding. The dilemma has two horns and they move in opposite directions over time. Early in a technology’s life, the harms are speculative — nobody can say with confidence what should be restricted, and any restriction looks like guessing. But that’s exactly when change is cheap: few users, few dependencies, no industry built on the current shape of the thing. Later, the evidence arrives — and by then, jobs, infrastructure, supply chains, and habits are built on it, and the same change now costs enormously and is resisted by everyone who depends on it. The result is a window that closes: the knowledge and the leverage are never available at the same time. This is why “wait and see” and “ban it now” are both bad defaults, and why governance work concentrates on what can be built during the window — monitoring that reveals harms earlier, reversibility so choices stay changeable, and staged deployment that buys evidence before dependency sets in.
An example. Orbital debris is close to a textbook case. The rules that would have mattered most — mandatory rapid deorbit, enforced from the first constellation — were cheapest to impose before tens of thousands of satellites were in the air and economies depended on them. The evidence that they matter is arriving now, when the leverage is lowest.
Whose it is. David Collingridge, The Social Control of Technology (1980).
What it does not mean. It isn’t an argument for pessimism about governance, and it isn’t a reason to do nothing at either end. It’s a description of why timing is the hard part — and the useful response is to attack the dilemma itself, by making harms visible earlier and choices reversible longer.
The unfolding. Two names for neighboring observations. The pacing problem is the tech-governance literature’s version (Gary Marchant and colleagues): law and regulation are built to move deliberately — comment periods, precedents, ratifications — while the technologies they govern iterate in months, so oversight arrives chronically late and aimed at the previous generation. The wisdom–power gap is Schmachtenberger’s wider framing: the metacrisis reading in which our capacity to do outruns our capacity to choose well — a generator function (A3) rather than an administrative nuisance, because every new capability arrives faster than the collective judgment for wielding it. The pacing problem is the institutional symptom; the wisdom–power gap is the civilizational diagnosis. Both point the same direction as Collingridge (D15): the governance work has to be designed for lateness — adaptive rules, sunset clauses, monitoring that shortens the knowledge lag.
An example. The orbital study’s governance layer is a measured specimen: constellation deployment scaled a thousandfold in the era governed by a treaty written in 1967 and licensing rules whose milestone clocks were designed to accelerate deployment. The regulator’s instruments aren’t wrong so much as from a different century than the problem.
Whose it is. Pacing problem: the technology-governance literature (Marchant is the standard citation). Wisdom–power framing: Schmachtenberger. Named separately, as the bench card does.
What it does not mean. Not an argument that governance is futile — the gap is a design constraint, not a verdict. And not unique to AI or any single technology: the structure recurs wherever capability compounds and deliberation doesn’t.
The unfolding. This is a real term from dynamical systems, not a metaphor we invented. Think of a landscape of hills and valleys with a ball rolling on it. The valleys are attractors: let the ball go anywhere in the surrounding slope and it ends up at the bottom. That surrounding slope is the basin of attraction. Two things follow that matter for everything else on this site. First, where you end up depends on which basin you started in, not on where you’re pointed at any moment — a system can be moving “in a good direction” and still be inside a basin that drains somewhere bad. Second, basins have edges, and near an edge a small push decides the outcome, while deep inside one, even a large push doesn’t.
An example. The hero image on our homepage is exactly this picture: three basins, two of them traps. It’s a diagram, not a decoration.
Whose it is. Standard dynamical-systems vocabulary, long predating any of its uses here.
What it does not mean. Calling something an attractor doesn’t make it inevitable, and it doesn’t imply anyone chose it. It’s a statement about stability — that the state, once reached, tends to persist — which is precisely why the interesting question is about the edges.
The unfolding. Nassim Taleb’s ladder has three rungs: the fragile breaks under stress; the robust (or resilient) withstands it unchanged; the antifragile gets better — it needs a word of its own, he argued, because no natural language had one. Muscles grow from load; immune systems learn from exposure; evolution runs on variation and selection; a portfolio of small experiments profits from volatility that would kill a single big bet. The design signature is asymmetry: small, capped downside with open-ended upside, plus enough redundancy to survive the losses that teach. The site’s hero copy (“distributed, antifragile, alive”) uses the word aspirationally — for arrangements that metabolize shocks into learning rather than damage. The term is genuinely contested: critics argue it renames known properties (hormesis, adaptive evolution, convex payoffs) and that few systems are antifragile in general rather than to specific stressors within specific ranges. The entry carries the dispute on purpose.
An example. (Outside example, Taleb’s own territory.) Bone under load: sedentary, it thins; stressed within tolerance, it densifies. The same skeleton is antifragile to exercise and fragile to a car crash — which is also the honest caveat in miniature: antifragility is always to something, within a range.
Whose it is. Nassim Nicholas Taleb, Antifragile (2012); flagged as his coinage and as contested.
What it does not mean. Not a synonym for resilient — resilience (D5) survives the shock; antifragility profits from it, and conflating them dissolves the word’s one job. Not an excuse to seek chaos: the concept prescribes capped downside, and a system that courts uncapped shocks is just fragile with bravado.
The unfolding. Standard economics diagrams put the economy at the center and treat nature as an input line-item. Ecological economics inverts the picture: the economy is a subsystem of society (it runs on trust, law, and institutions it doesn’t produce), and society is a subsystem of the biosphere (it runs on food, water, climate stability, and materials it doesn’t produce). Daly’s nested model is the formal statement; Kate Raworth’s doughnut popularized a two-boundary version — an ecological ceiling above (D3) and a social foundation below. The ordering has a hard implication: an inner ring can grow relative to its container only until it starts consuming the container’s regenerative machinery (A12), at which point growth in the part degrades the whole — the inversion of the more familiar assumption that a bigger economy can always buy its way out of social and ecological repair.
An example. The site’s Method page draws its three rings — biosphere ▷ society ▷ economy — as a standing commitment: our studies read economic questions inside their social and biophysical containers, which is why an oil shock gets traced into household budgets (the social ring) and a satellite business model into atmospheric chemistry (the outermost one).
Whose it is. Ecological economics — Daly’s nested model; Raworth’s doughnut is the popular carrier.
What it does not mean. Not “the economy doesn’t matter” — inner rings are where most levers live. The claim is about dependency direction: the economy can be rebuilt from a functioning society and biosphere; the reverse operation has never been demonstrated.