Haute Lumière
Commerce · VI.05 · MMXXVI · daylight
The Commons That Fell
Volume VI — Governance and the Commons
THE PLATE
THE LETTER
Chapter VI.04 walked the commons that held — the Törbel alps kept under one set of rules since 1483; the huerta of Valencia with a water court that has sat on Thursdays for longer than most nations have existed, the Japanese iriai forests, the Alanya inshore fishery. Those are the reason this volume exists. They are proof that a shared resource is not automatically a tragedy, and that Hardin described one particular institutional arrangement rather than a law of nature.
This chapter reads the other file. Four commons that did not hold, examined the way an air accident is examined: not to assign blame, and not to be frightened, but because a failure that has been read properly is worth more than a success that has only been admired.
The method stays appreciative, and that is not a softening. A commons that fell is a commons that worked for a while, sometimes for centuries, which means the honest question is never why do people ruin things — it is what were these institutions doing right, for how long, and what specifically changed. The Newfoundland inshore fishery ran for four hundred and fifty years. The Reddit moderator corps built the eighteenth most-visited site on earth for nothing. Ask a system what was working before you ask it what failed and you will get a different, better answer.
One instrument runs through all four cases and it is a duration. Call it the lag: the time between the first public, quantified measurement showing the draw exceeding the return, and the moment the commons stopped being one. In every case the lag was measured in years. In every case the measurement existed and was in a filing cabinet. Nothing here fell in the dark.
— The Editors
DISCOVERY
What was working, and for how long
The northern cod fishery worked for four and a half centuries. From the early sixteenth century, when Basque, Breton, Portuguese and West Country boats began working the Grand Banks, to the middle of the twentieth century, the inshore fishery off Newfoundland was a functioning common-pool resource under rules that were largely unwritten and largely obeyed. Cod were taken from small boats with hook and line, and later cod traps, within a few miles of a home port. Access was governed by berth — a trap berth was a named place, drawn by lot in many outports each spring, and a family fished it. That is Ostrom's first principle, arrived at independently: boundaries defined by people who had to live with each other afterwards. It was not idyllic and it was not rich. It was durable, and durability is the property under examination.
Rapa Nui worked for six hundred years. Current archaeology — Terry Hunt and Carl Lipo's chronometric work on the radiocarbon record, published in Science in 2006 — places settlement around 1200 CE rather than the fourth-to-eighth-century range of the older accounts. In that window a population on sixty-three square miles of basalt with no permanent streams built and moved eight hundred and eighty-seven monumental statues, and fed itself by an agriculture of extraordinary sophistication: lithic mulching, rock gardens, manavai stone enclosures that trap moisture and moderate soil temperature. Stevenson and colleagues, working the land-use record, describe a system finely tuned to local variation. That is Ostrom's second principle — congruence between rules and local conditions — executed in stone, at scale, on one of the most isolated inhabited places on the planet.
The High Plains had a working rule and abandoned it. Under the Ogallala formation, before the deep turbine pump and rural electrification, the constraint was the lift. A windmill could raise stock water; it could not irrigate a section. The binding rule was physical and therefore self-enforcing, which is the cheapest kind of rule there is. Kansas then did something better: the Groundwater Management District Act of 1972 created five districts with the power to make their own rules, governed by boards the irrigators elect. That is Ostrom's third and seventh principles together — collective choice, with the state recognising the right to organise. The machinery for governing the aquifer was built, by statute, half a century ago, and it is still there.
Reddit's commons was the largest volunteer institution of its kind. By the early 2020s something on the order of sixty thousand unpaid moderators ran more than a hundred thousand active communities, wrote their own rules, ran their own appeals, trained their own successors, and built the tooling themselves — the bots, the automod configurations, the wikis, the shared blocklists. Li, Chancellor and Hecht put a minimum-wage floor under the logged portion of that work and got roughly $3.4 million a year, about 2.8 per cent of the company's 2019 revenue. That estimate is a floor on the visible part and everyone including its authors says so. The thing worth noticing is not the number. It is that a stranger's institution ran monitoring, graduated sanction and conflict resolution — Ostrom's fourth, fifth and sixth — at a scale no paid trust-and- safety department has matched, and did it for free, for fifteen years.
Four commons. Four hundred and fifty years, six hundred years, fifty years of statutory self-governance, fifteen years of volunteer institution-building. None of these were fragile. Three of the four had genuinely excellent governance for most of their lives. Hold that firmly, because the forensics only work from here: you cannot read a failure in a system you have not first seen working.
And hold one more thing. In every case that follows, somebody measured the problem, wrote it down, and published it, while there was still time to act. The record is not missing. The record is the most striking thing in the file.
THE ARITHMETIC
The lag, its three geometries, and the case where somebody closed it
Define the instrument.
L = t(collapse) - t(first public, quantified warning)Not the first worry. Not the first anecdote. The first published number showing appropriation above regeneration. Here are the four, computed in lib/verify/VI_05.py:
commons warning end L geometry
---------------------------------------------------------------------
northern cod 2J3KL 1986 1992 6 yr exponential
Rapa Nui, the population 1862 1877 15 yr step
High Plains aquifer 1965 — 50 yr+ linear
Reddit third-party API 2015 2023 8 yr step
---------------------------------------------------------------------
stratospheric ozone (held) 1985 1987 2.3 yr heldThe lag is not the finding. The geometry is. What a delay costs depends entirely on what kind of stock is being drawn down, and the three cases are arithmetically different animals.
Exponential — a living stock under excess mortality. Fishing mortality is an instantaneous rate; survival is e^(−(F+M)t). Through the late 1980s realised F on northern cod ran at roughly two to three times the F0.1 reference point in use. Take F = 0.50 against a target of 0.20, natural mortality M = 0.20, and the six-year lag:
cohort survival at realised F 0.0150
cohort survival at F0.1 0.0907
cost of the lag = e^(ΔF · L) 6.05 ×Six years of advice not taken left about one sixth of the biomass the same six years at the reference point would have left. That is the whole shape of the thing: the cost of delay is exponential in its length, so the fifth year of a lag is worth more than the first four together. State the denominator plainly — this is a single-cohort survival model carrying mortality only, with no recruitment, no growth and no depensation, which is why it understates what Hutchings and Myers found in the real series.
Linear — a mined stock with negligible recharge. The High Plains aquifer covers 174,000 square miles. USGS puts predevelopment storage at 3,183 million acre-feet and 2015 storage at 2,910 — a depletion of 273.2 million acre-feet, 8.58 per cent, over sixty-five years. Divide:
mean depletion rate 4.20 million acre-ft / yr
remaining / rate 692 years of aggregate life
Texas portion, mean decline 41.5 ftSix hundred and ninety-two years. That number is arithmetically correct and it is a lie, because in parts of the Texas Panhandle the saturated thickness is already too thin to run a centre pivot. An average taken across a commons with no shared boundary is not a measurement of anything, which is why Ostrom's first principle is not a nicety about neighbourliness — it is the precondition for a denominator existing at all. One aquifer, eight states, no common rule, and therefore no honest aggregate.
And the High Plains row has no end date. In 1965 the Fifth Circuit, in United States v. Shurbet, allowed Southern High Plains irrigators to deduct groundwater depletion from taxable income as the exhaustion of a capital asset. The United States tax code formally recognised the overdraft sixty years ago, and the pumping continued. A stock commons with no terminal event has an unbounded lag, and that is the most dangerous class of all, because a collapse with no date cannot be put in anyone's calendar.
Step — a rule-governed commons, where the cost is zero until it is total. Reddit announced API pricing at $0.24 per 1,000 calls. Apollo's developer published his own volume — seven billion calls a month, about 1.5 million users:
7,000,000 thousand-call units × $0.24 = $1,680,000 / month
= $20,160,000 / year
per user per month = $1.12Against Reddit's own numbers — FY2023 revenue of $804.0 million over 73.1 million daily actives — ARPU was $11.00 a year, $0.92 a month. The toll on the commons was set at 1.22 times what the landlord could earn from the same user itself. A price above your own ARPU is not a price. It is a notice to quit, and it was read correctly: 8,829 subreddits went dark on 12 June 2023, 3.88 times the 2,278 that went dark in the July 2015 rehearsal eight years earlier.
Now the failure order, which is the forensic payload. Reddit's commons did not fail Ostrom's principles in sequence. It never had the third one. Moderators monitored, sanctioned, resolved conflict and nested their communities — four, five, six and eight, done well — while never at any point holding the right to modify the rules they operated under. A commons in which the appropriators cannot change the rules is not a commons. It is a tenancy, and the tenancy ended when the landlord found a buyer: the company listed in March 2024 at a $6.4 billion valuation. The volunteer labour valued at 2.8 per cent of 2019 revenue is 0.42 per cent of 2023 revenue and 0.053 per cent of that valuation — the commons' share of the thing it built fell 6.6-fold without a single rule being changed.
The cut: Rapa Nui did not fail
The most-taught commons collapse on earth is the one in the file that did not happen.
The popular account — an island that cut down its last tree, fell into warfare and cannibalism, and destroyed itself before Europeans arrived — is a parable about internal overshoot, and it is taught because it is a useful parable. It puts the cause inside the society, where a lesson can be extracted.
The record does not support it. Hunt and Lipo's radiocarbon work moves settlement to about 1200 CE, compressing the deforestation window to roughly 450 years and implicating the Polynesian rat, which eats palm seeds, alongside human clearing. DiNapoli, Crema, Lipo, Rieth and Hunt's Bayesian analysis of the radiocarbon and palaeoenvironmental record, in Nature Communications in 2021, found population growth followed by a plateau — no pre-contact collapse. Moreno-Mayar and colleagues sequenced ancient Rapanui genomes in Nature in 2024 and found no genetic signature of a pre-European crash either. The palms went, and the islanders adapted: the rock gardens and lithic mulching are the adaptation, and they carried the population through to 1722, when Roggeveen arrived and found a functioning society.
Here is what actually happened, and here is the arithmetic:
ecological interval settlement 1200 -> palms gone 1650 450 yr
demographic interval first raid 1862 -> census 1877 15 yr
ratio 30.0 ×
population 1862 (estimate range) 2,000 to 4,000
people counted, 1877 111
loss 94.5 - 97.2 %Peruvian slave ships took roughly fifteen hundred people beginning in December
- Smallpox came back with the handful who were returned. Fifteen years
later there were one hundred and eleven people alive on Rapa Nui. The ecological change took thirty times as long as the demographic one and did not cause it.
So: which of Ostrom's conditions failed? Not the first six — they were working, which is precisely why the island was inhabited at all after the palms. The seventh failed. Minimal recognition of the right to organise, by external authorities. The external authority arrived armed, recognised nothing, and the rest followed in a decade and a half.
That is the cut, and it turns the whole chapter over. We teach Rapa Nui as self-destruction because self-destruction is a lesson we can act on, and we edit out the raid because the raid is somebody else's fault. The commons did not fall. It was taken. Which means the seventh principle — the dull procedural one about external recognition, the one that reads like boilerplate — is not the eighth-most-important. In this file it is the only one that killed anybody.
The honest positive: a commons diagnosed in time
Failures are only instructive if the alternative is real, and it is.
In 2012 the irrigators of Sheridan County, Kansas — inside Groundwater Management District 4, on the same aquifer, under the same hydrology — used the 1972 statute to petition for a Local Enhanced Management Area. Sheridan 6 took effect on 1 January 2013: ninety-nine square miles, an allocation of 55 inches per irrigated acre over five years, eleven inches a year, about 20 per cent below prior use.
Deines, Kendall, Butler and Hyndman measured it against a matched counterfactual and published in 2019:
intended cut on prior use 20 %
measured cut, 2013-2017 31 %
overshoot on the rule 11 points
crop production no detectable loss
decision to published result 5 yearsThey cut their own draw by nearly a third, beat their own rule by eleven points, and lost nothing. The rule was written by the people who pump the water, monitored with meters they accept, and enforced against neighbours. That is principles one through six, assembled in one county, in one year, by people with a direct financial interest in not doing it.
Two honest negatives sit inside that positive and both belong on the page. The water table under Sheridan 6 is still falling — the LEMA changed the slope, not the sign. And it governs 0.0569 per cent of the aquifer. A working correction at one twentieth of one per cent of the problem is a proof of mechanism, not a solution, and calling it more than that is how a good finding gets discredited.
The second positive is the fastest response on record. Farman, Gardiner and Shanklin published the Antarctic ozone measurements in May 1985; the Montreal Protocol was signed in September 1987. L = 2.3 years — one quarter of Reddit's, and Solomon and colleagues documented the September hole shrinking by more than four million square kilometres between 2000 and 2015. The commons that hold are not the ones with better people. They are the ones whose institutional response time is shorter than their lag.
DREAM
What becomes ordinary
Picture the version where the lag is the thing everybody watches, and describe it in the present tense, because a dream in the future tense is a wish.
The fishery has a published reference point and a standing rule that a stock assessment crossing it triggers the quota reduction rather than initiating a discussion about one. The minister does not decide; the minister signs what the trigger already produced. Nobody experiences this as a loss of authority, because the authority was never usable inside a six-year window anyway.
The aquifer has a boundary. Not eight state lines drawn by surveyors in the 1860s across a body of water nobody had measured, but a management boundary that follows the hydrology, with the districts nested inside it — Ostrom's eighth principle, built rather than hoped for. Each district writes its own allocation. The basin holds the aggregate. There are meters, the irrigators accept them because they wrote the rule that requires them, and the water-level record is public monthly with the same seriousness as a commodity price.
The platform commons has a written constitution with an amendment procedure, and the people who do the moderation hold a defined share of the votes on it. Not a seat on an advisory council. A share of the votes, with a notice period on API terms measured in years rather than weeks, and an export right that works. The company finds this improves its position rather than weakening it, because a community that can change its own rules does not need to go dark to be heard, and a platform that cannot be left by its contributors was never worth what the prospectus said.
And every one of these institutions keeps a single number on the front page of its reporting pack: how long, at present, between a measurement crossing the line and a rule changing. Not a target. A measured figure, published, going down.
None of this requires a new law, a new technology, or anyone's conversion. Sheridan County did the hardest version of it in one year with a statute from 1972.
DESIGN
The forensic apparatus, and the institution it implies
Four moves. They are in order and the order matters.
One — establish the failure order, not the failure. "Governance failed" is not a finding. Take Ostrom's eight and date each one. For northern cod: the boundary went first, in 1949 and again in the 1950s, when the offshore factory fleet arrived and the trap berth ceased to define anything; congruence went next, when a rule fitted to a small boat was replaced by a quota fitted to a stern trawler; monitoring failed third, because catch-per-unit-effort from a trawler fleet using improving electronics rises while the stock falls, which is a monitoring instrument that lies in exactly the direction that comforts you; collective choice failed fourth, when the 1986 advice went to a federal department and the inshore fishermen reporting empty traps had no standing to be believed. Four principles, four dates, one decade. A failure with dates is a failure you can build against.
Two — name the instrument that lied. Every case has one. Cod had CPUE from a modernising fleet. The High Plains has the area-weighted average. Rapa Nui has a deforestation date used as a proxy for a demographic one. Reddit had engagement metrics that rose through 2015 because the blackout was engagement. In each, the number people watched moved the wrong way relative to the number that mattered. Ask of every governing metric: under what circumstance does this rise while the thing it represents falls? If there is no such circumstance, you have not understood the metric yet.
Three — measure your own response time. Not the lag; the lag is the resource's. R is yours: from a measurement crossing a published line to a rule actually changing, how many months? Find the last three rule changes and count backwards from each. Most institutions have never computed this and are shocked by it. A commons holds when R < L and fails when R > L, and since you cannot lengthen L — the resource sets that — the entire art is shortening R.
Four — pre-commit, because R is dominated by the decision, not the implementation. Sheridan 6's allocation took effect in twelve months because the decision had been made in advance, by the people it bound, under a statute that said what would happen. The Montreal Protocol had a built-in adjustment procedure precisely so the next tightening would not need another treaty. This is Ostrom's fifth principle read forward: graduated sanctions are not punishment, they are a decision taken early so it does not have to be taken late.
The sequence, for a governing body with a resource in its care:
| Step | Action | Artifact |
|---|---|---|
| 1 | Date each of the eight principles: working, degraded, gone | The failure-order table |
| 2 | Name every governing metric that can rise while the stock falls | The lying-instrument list |
| 3 | Compute R from the last three rule changes | One number, in months |
| 4 | Publish a trigger: line, measurer, automatic step | The standing rule |
| 5 | Escrow the transition cost before the trigger fires | The standstill facility |
DESTINY
How the correction holds when nobody is pushing
The Sheridan 6 rule has now outlived its own novelty, been renewed, and been copied into further LEMAs. Three things did that, and they are the same three in every case in this file where something held.
The rule binds the people who wrote it. Not a regulator's rule accepted under protest — a rule the irrigators petitioned for, which is why compliance costs almost nothing to enforce. Ostrom's finding, over and over: rules made by appropriators are obeyed at rates that astonish people who model enforcement.
The measurement is independent of the appropriator and public. Meters read by the district, water levels published, the counterfactual estimated by university hydrologists with no stake. The moment measurement sits inside the appropriating body you have rebuilt the CPUE problem with new equipment.
The trigger is automatic. A rule that requires a fresh decision each time is a rule with your own R embedded in it, and R is what killed the cod.
Now the honest part, because this is where these corrections fail. They fail when the trigger is set at the level where harm begins rather than at the level where the response time begins — if R is three years, a trigger at the harm line is three years late by construction. They fail when the boundary is smaller than the resource, which is Sheridan 6's own exposure: 0.0569 per cent of an aquifer, with every neighbouring section free to pump. They fail when the compensating escrow is not funded, so the first hard year converts the rule into a hardship petition. And they fail — most often — when the institution mistakes having published a measurement for having acted on one. Every commons in this chapter had the measurement. The measurement is not the intervention, and a filing cabinet is not a governance mechanism.
DELIGHT
What it feels like
There is a specific pleasure in reading an old assessment volume and finding that somebody already knew. It is not vindication and it is not grief. It is closer to the feeling of meeting a colleague — a person at a desk in 1986, or 1965, doing the arithmetic properly, writing it down clearly, and filing it where it could be found. They did their part. The part that failed was downstream of them.
And then the better pleasure, which belongs to Sheridan County. There is something genuinely funny about a room full of irrigators voting to take a fifth of their own water away, and then quietly taking a third, and then finding at harvest that the corn did not notice. The comedy is that everyone expected it to hurt. It did not hurt. Most of the sacrifice we assume a commons requires turns out on measurement to be slack nobody had costed — which is the same finding as Chapter I.01, arriving from the opposite direction.
That is what makes this work absorbing rather than mournful. You are not handling a tragedy. You are handling a filing problem with a deadline, and the deadline is negotiable if you start early enough.
OPERATIONALIZE THIS
At the level of finance
The instrument: a standstill facility — a pre-committed, trigger-indexed reduction with an escrowed transition fund.
The problem every fallen commons shares is not ignorance and not greed. It is that the cost of restraint is immediate and concentrated on named people, while the benefit is deferred and diffuse. That is a financing problem, and financing problems have instruments.
The mechanics.
- The trigger. One published index with one published line. A stock-status ratio (
SSB / B_lim), a saturated-thickness rate (feet per year over a rolling five), a contractual notice metric. Written before the line is near, because a line drawn while approaching it will be drawn in the wrong place. - The step. Graduated and automatic. Crossing the line reduces the allocation by a stated percentage on a stated date. No meeting. The meeting is what
Ris made of. - The measurer. Named, independent of every appropriator, paid from the escrow rather than by the operator, and obliged to publish whether or not the line was crossed. A measurer paid to look only when there is trouble reports trouble only when it is paid.
- The escrow. Funded in advance out of the resource rent in good years, at a rate set so it covers the gross margin on the foregone units for the first
nyears after a trigger. This is the whole instrument. The escrow is what converts a rule people will fight into a rule people will write. - Term and review. Set the term past the resource's own recovery time constant. Hard review at one third of it.
- Security. The rent stream itself. If the resource does not recover, the escrow is not replenished, and the facility should say so in terms.
Balance-sheet treatment. The escrow is restricted cash, not a provision — it is a funded obligation with a named trigger, and treating it as a provision invites it to be released into earnings in a good year, which is the failure mode. The allocation right is an intangible with a stated life, and its life is the resource's regeneration time, not the licence period. That single alignment is frequently the entire argument with the auditors, and it is a conversation about useful economic life, which they have every year anyway.
The counterparty. The management district or authority first, internally funded. Two completed internal cycles give you a track record. Only then take it outward — to a reinsurer, who prices parametric triggers routinely, or to a sustainability-linked bond where the coupon steps with the same index. The capital markets can already price a parametric trigger; what they have never been offered is a commons with one.
The number that decides it. Buy the standstill when
C_standstill < V · ( 1 − e^(−ΔF · L) )V the present value of the sustainable rent, ΔF the excess appropriation rate per year, L your own response time in years. Worked in lib/verify/VI_05.py on the cod parameters — V = $100m, ΔF = 0.30/yr, L = 6 — the value at risk is $83.5 million, a three-year escrow at $8 million a year is $24 million, and the decision ratio is 0.29. The lag costs about three and a half times what closing it costs.
And the number that decided it in practice: Sheridan 6's C was approximately zero, because income was maintained. A standstill that costs nothing needs no escrow at all. It needs a rule the appropriators wrote themselves — which is the cheapest instrument in this volume and the hardest one to buy.
The first ninety days.
| Day | Action | Artifact |
|---|---|---|
| 1–15 | Date the eight principles for your resource | Failure-order table |
| 16–30 | List every metric that can rise while the stock falls | Lying-instrument list |
| 31–45 | Compute R from the last three rule changes | One number, in months |
| 46–60 | Draft the trigger: index, line, step, measurer | The standing rule |
| 61–75 | Size the escrow against V · (1 − e^(−ΔF·L)) | Facility memo |
| 76–90 | Put the trigger to the appropriators for their vote | Their signatures |
APPRECIATIVE QUESTIONS
Twelve, for a room
Discovery — what has already held here
- What rule in this organisation has been obeyed for years without anyone enforcing it, and what makes it obeyable?
- Think of a time we changed a rule quickly and it stuck. How many weeks did it take, and who made that possible?
- Which measurement do we already publish that somebody outside could check — and who checks it?
Dream — what becomes possible
- If our response time were three months instead of three years, which decisions would we make differently, and which would simply make themselves?
- Imagine the people who draw on this resource wrote next year's allocation themselves. What would they write that we would not have dared to propose?
- What would it be worth to us to know, on the day it happened, that a line had been crossed?
Design — what we build
- Which of our governing numbers can rise while the thing it stands for falls? Name one, out loud, now.
- Where is the boundary of the thing we are actually managing, and where is the boundary of our authority over it? What sits in the gap?
- What is the smallest trigger we could publish this quarter that would fire without a meeting?
Destiny — how it holds
- If everyone in this room left, which of our rules would still be running in five years, and why those ones?
- Who would have to be paid, and how much, for a reduction to feel like a decision rather than a loss?
- What is the first measurement we would see if this were beginning to fail, and whose desk does it currently land on?
WORKS CITED
Butler, J. J., Whittemore, D. O., Wilson, B. B. and Bohling, G. C. (2018). "Sustainability of aquifers supporting irrigated agriculture: a case study of the High Plains aquifer in Kansas." Water International, 43(6), 815–828.
Deines, J. M., Kendall, A. D., Butler, J. J. and Hyndman, D. W. (2019). "Quantifying irrigation adaptation strategies in response to stakeholder-driven groundwater management in the US High Plains Aquifer." Environmental Research Letters, 14(4), 044014.
Diamond, J. (2005). Collapse: How Societies Choose to Fail or Succeed. Viking.
DiNapoli, R. J., Crema, E. R., Lipo, C. P., Rieth, T. M. and Hunt, T. L. (2021). "Approximate Bayesian Computation of radiocarbon and paleoenvironmental record shows population resilience on Rapa Nui (Easter Island)." Nature Communications, 12, 3939.
Farman, J. C., Gardiner, B. G. and Shanklin, J. D. (1985). "Large losses of total ozone in Antarctica reveal seasonal ClOx/NOx interaction." Nature, 315, 207–210.
Finlayson, A. C. (1994). Fishing for Truth: A Sociological Analysis of Northern Cod Stock Assessments from 1977 to 1990. Institute of Social and Economic Research, Memorial University of Newfoundland.
Hardin, G. (1968). "The Tragedy of the Commons." Science, 162, 1243–1248.
Harris, L. (1990). Independent Review of the State of the Northern Cod Stock: Final Report. Department of Fisheries and Oceans, Ottawa.
Hunt, T. L. (2007). "Rethinking Easter Island's ecological catastrophe." Journal of Archaeological Science, 34(3), 485–502.
Hunt, T. L. and Lipo, C. P. (2006). "Late Colonization of Easter Island." Science, 311(5767), 1603–1606.
Hunt, T. L. and Lipo, C. P. (2011). The Statues That Walked: Unraveling the Mystery of Easter Island. Free Press.
Hutchings, J. A. and Myers, R. A. (1994). "What can be learned from the collapse of a renewable resource? Atlantic cod, Gadus morhua, of Newfoundland and Labrador." Canadian Journal of Fisheries and Aquatic Sciences, 51, 2126–2146.
Li, H., Chancellor, S. and Hecht, B. (2022). "All That's Happening behind the Scenes: Putting the Spotlight on Volunteer Moderator Labor in Reddit." Proceedings of the International AAAI Conference on Web and Social Media (ICWSM), 16, 584–595.
McGuire, V. L. (2017). Water-Level and Recoverable Water in Storage Changes, High Plains Aquifer, Predevelopment to 2015 and 2013–15. U.S. Geological Survey Scientific Investigations Report 2017-5040.
Moreno-Mayar, J. V., et al. (2024). "Ancient Rapanui genomes reveal resilience and pre-European contact with the Americas." Nature, 633, 389–397.
Myers, R. A., Hutchings, J. A. and Barrowman, N. J. (1997). "Why do fish stocks collapse? The example of cod in Atlantic Canada." Ecological Applications, 7(1), 91–106.
Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action. Cambridge University Press.
Ostrom, E. (2009). "A General Framework for Analyzing Sustainability of Social-Ecological Systems." Science, 325, 419–422.
Rose, G. A. (2007). Cod: The Ecological History of the North Atlantic Fisheries. Breakwater Books.
Solomon, S., Ivy, D. J., Kinnison, D., Mills, M. J., Neely, R. R. and Schmidt, A. (2016). "Emergence of healing in the Antarctic ozone layer." Science, 353(6296), 269–274.
Stevenson, C. M., Puleston, C. O., Vitousek, P. M., Chadwick, O. A., Haoa, S. and Ladefoged, T. N. (2015). "Variation in Rapa Nui (Easter Island) land use indicates production and population peaks prior to European contact." Proceedings of the National Academy of Sciences, 112(4), 1025–1030.
United States v. Shurbet, 347 F.2d 103 (5th Cir. 1965).
Note on figures. Every number above is computed in lib/verify/VI_05.py, which prints its inputs with their units and their sources before its results. The cod cost-of-delay is a single-cohort survival model carrying mortality only and therefore understates the true cost; the Sheridan 6 reduction is a matched- counterfactual estimate; Apollo's call volume and user count are the developer's own public statements; the Rapa Nui 1862 population is a missionary estimate with a stated range, which is why the loss is printed as a range and not a point. Ostrom's eight principles are set out in full in Chapter VI.02 and are named here only where a case turns on one. Chapter VI.04 carries the systems that held.