Course Arc

Working plan, v0.3. Two levels:

Sessions — the 6 live slots, ~1.5h each. What people sign up for. Wed 7 Oct to Wed 14 Oct 2026, weekdays only (7, 8, 9, 12, 13, 14), 15:30-17:00 CEST.
Modules — the units of material. Smaller than a session; a session holds 3–6 of them. Research and preparation is organised around modules, not sessions, so they can be moved, cut, or promoted as the material develops.

The spine: why this matters now → what the failure actually is → how we already cooperate → how cooperation has been made to work among strangers → cultural evolution as the mechanism behind every jump in scale → and the planetary jump we now need.

We're interested in the problems because we want the solutions. Diagnosis is a stage the course passes through rather than the thing it dwells on.

Audience: advanced, already familiar with the idea of a collective action problem, not necessarily with the game theory. The work is bridging the applied case, the theoretical background, and — where I actually want to get to — the cultural evolution narrative.

Public version of this: landing.md. Raw braindump it came from: ../raw/2026-09-11-course-outline-braindump.md.

Two devices running through the whole course

The diagnostic. Five questions, introduced in Session 1 and re-run out loud on every case in every session:

How many players?
Can you see defection?
Can you punish it?
Will you meet again?
Is anyone big enough to act alone?

This is the portable instrument people take away. Without it the course is cases, theory and history with no tool at the end of it.

The same diagram at two scales. Every case gets drawn next to a case several orders of magnitude away. A handful of people in a room next to a handful of AI labs in a race. Lobster grounds next to global fisheries. Ilahita next to the planet. The scale jump is the course's whole subject, so make it visible every session rather than announcing it once.

Session 1 — Metacrisis Motivation

Why should we care about collective action problems? How do they relate to climate change, runaway AI and the metacrisis in general?

Module: Introduction to collective action

What the thing is, stated plainly, with a couple of cases. Successes as well as failures — the course is about opportunities as much as problems, and if the first hour is all disaster the rest of the arc reads as consolation.

[Research pass needed: opening cases and video clips. This is the module with the least material so far.]

Module: War

The oldest and largest of them. An arms race is a collective action problem with a body count, and war is also — awkwardly — one of the main things that has ever driven humans to cooperate at larger scales. Both faces here, briefly. Session 5 collects on this with Turchin.

Module: The pattern beneath the metacrisis

The connection made explicitly, for this audience.

This is the central pattern of dysfunction. It sits beneath climate change. Beneath the AI arms race, nuclear weapons, bioweapons — each one of us might want to stop, or go slower, and we can't. And beneath nihilism, anomie, loneliness too: it makes sense to me to sit in my room and watch Netflix when everyone else is doing whatever they're doing.

There is a collective action problem of us.

Module: Why this matters so much now

Why the present moment is different: the speed of technology relative to governance, the number of players who can now do civilization-scale damage, and the fact that the remaining problems are the ones that are global by construction and so can't be solved at any smaller scale.

Prep and materials

If a Tree Falls: A Story of the Earth Liberation Front goes out as prep viewing rather than being taught as a module. People who did a genuinely hard piece of cooperation for years, then informed on each other within weeks of arrest. It's the cleanest illustration available that this isn't about bad character, and it's worth having in the room without spending session time on it.

Session 2 — Stag & Moloch

Why do rational, decent people end up in outcomes none of them want? The basic dynamics of cooperation and defection, with the theoretical frameworks and key terminology.

Module: The games

Prisoner's dilemma, stag hunt, tragedy of the commons, public goods and free riding. Three or four done properly, not a survey.

Keep the distinction that does real work: dilemma vs coordination vs assurance. Different problems, different fixes — and the stag hunt is what the optimistic half of the course title rests on.

Module: Some basic game theory

The minimum apparatus: payoff matrices, dominant strategies, Nash equilibrium, why an equilibrium can be stable and terrible at the same time. Built from scratch, no prior exposure assumed.

The point underneath it: defection is not a deficiency. It's rational, it makes sense, no villain required. And yet we do cooperate — that's the puzzle, not the defection. Any account that locates the problem in people being unenlightened has misread the situation and will prescribe accordingly.

Module: The reframe

The framing in the literature isn't attractive. Even the prisoner's dilemma is a story about prisoners betraying each other. Turned around, it is a cooperation opportunity, a cooperation challenge. Not a collective action problem but a collective action opportunity.

And the direction of the question changes with it. Not "there's a good state and we fell into the bad one" — the bad state is the default. What's the bad state, and how do we get into the good?

Module: The many names, and a critique of Moloch

Moloch and the Scott Alexander piece, which many people will know. The multipolar trap. Race to the bottom, arms race, social dilemma, coordination failure — the same structure named in a dozen literatures.

Moloch is great branding and conceptually problematic: it personifies a structure, and turns a design problem into fate. Worth saying plainly that half the course title is a word I intend to argue with.

Session 3 — How Do We Ever Cooperate?

We do cooperate, constantly. What makes that possible, and why does it hold at village scale?

Module: The cooperative ape

Start from the fact rather than the puzzle. We are an unusually cooperative species, and cooperation is not a human invention either — the animal cases go here. Then the actual question: given that we're built for it, what conditions does it need, and what makes those conditions fail?

Module: Olson

The Logic of Collective Action — the founding text, and the cleanest answer to why small groups manage and large ones don't. Group size, latent groups, selective incentives, and the privileged group where one player is large enough to provide the good alone.

Module: Repeated interaction and the folk theorem

Repetition changes everything: the shadow of the future. What it actually needs — regular interaction, the ability to observe, the ability to punish people who don't cooperate. Note that this is questions 2, 3 and 4 of the diagnostic, which is not a coincidence.

Module: The evolution of cooperation

Axelrod's tournaments; Nowak, repeated interaction on a grid, spatial structure, and the conditions under which cooperation emerges at all.

Module: Ostrom and the commons

The design principles as the empirical version of the same ingredient list, and what real communities actually do to govern a commons. [I don't know Ostrom's work as well as I probably should — this is the module needing most research.]

Distinguish kinds of commons, because they are not the same problem:

Positive commons — the information commons. Things we need to create, where the failure is undersupply.
Subtractable commons — lobster grounds, alpine grazing. Things we can overuse, where the failure is depletion.

Module: Why this gets harder at scale

Everything in this session depends on repetition, observation and punishment among people who know each other. Hand over to Session 4.

Session 4 — Cooperation Among Strangers

How do you make cooperation work when the group is larger? What have we built, and what has actually worked?

The session where the practitioners in the room get paid.

Module: Coercion

Hobbes. And Hardin's own answer, which is usually forgotten by people who quote him: mutual coercion, mutually agreed upon. The state is a collective action technology — the most consequential one we've ever built deliberately — and jumping from village norms straight to religion would skip it.

Module: Changing the payoffs

Taxes, subsidies, property rights, Coase. Markets as cooperation technology, and where that runs out. Home turf.

Module: Schelling

Focal points, credible commitment, cooperating with an adversary you cannot trust, the strategy of conflict. He was writing about nuclear arms races and the AI version rhymes closely. Given how many people will come for the AI framing, this may be the highest-value module in the course.

Module: Polycentricity

Ostrom's answer to scale, and the bridge the course would otherwise be missing. Session 3 ends "it gets harder at scale"; Session 5 answers "culture". This is the second answer, and the practical one: you don't scale a village, you stack villages. Federation all the way up.

Module: Did it ever work?

The applied record, and the most directly useful hour in the course. Montreal against Kyoto. Nuclear non-proliferation. Fisheries. Why some worked and most didn't, and what the difference turned on — number of players, cost of monitoring, availability of substitutes, observability of defection. Run the diagnostic on each, live.

Session 5 — Cultural Evolution

How did we get from bands of 150 to a planet of 8 billion? What innovations made each jump possible?

Module: Multi-level selection

Group-level selection — the largely genetic story of why we select for cooperation at all. Then cultural group selection: cultures that cooperate outcompete cultures that don't. And gene–culture coevolution between them. Boyd and Richerson here.

Module: Ritual and the manufactured tribe

Ilahita — ritual, initiation, fictive kinship holding together a village of thousands that on the standard account shouldn't exist.

Module: War as ratchet

Turchin, Ultrasociety. War as the brutal selection pressure that ratchets cooperating group size up and up. Collects on the war module from Session 1.

Module: Costly signals and Big Gods

Sosis on religious communes and hard-to-fake signals; Norenzayan on moralizing high gods. This is the mechanism by which religion does collective action work. Without it, the claim that religion is a cooperation technology is asserted rather than shown — and the claim is load-bearing for Session 6.

Module: Henrich and the trust of strangers

The WEIRDest People in the World. The Church's marriage and family rules dissolve kin networks, and trusting strangers becomes possible. Culture as the technology.

Frame for the whole session: religion, ritual, law, money, the firm are all collective action technologies, and each one buys another order of magnitude of group size.

Module: The bill

The sting, and the strongest teaching moment in the course. Don't paper over the contradiction — surface it.

Cooperation scales by building in-groups, and an in-group needs an out-group. Turchin's ratchet is war: cooperation manufactured for the purpose of killing the neighbors. Every enlargement of the circle so far has been paid for by whoever was left outside it.

Agriculture belongs here. Hunter-gatherers had a pretty good life, and then there was the fall into farming, which no one chose and which made most people worse off for millennia. It's the earliest case of cooperation at larger scale leaving most participants worse off, so it's evidence for the bill rather than a separate story. (Moved out of Session 1 — it works better as the sting than as the opener.)

End the session there and let it hang overnight.

Session 6 — The Planetary Turn

What would cooperation at planetary scale require? Do we need a planetary religion?

Module: What modernity already did

Human rights, international institutions, we're all brothers and sisters. That enlargement was real. It's also incomplete.

Module: Why the current stack can't hold

Nation-states and markets against AI, climate, the biosphere.

Module: The first enlargement with no outside

The answer to Session 5's sting, and the reason this jump is not simply the next in a series. Every previous enlargement had an outside — to define itself against, to compete with, to send the bill to. This one has none. No external enemy, nobody to defect against, nobody to pay. Which is exactly why more of what worked before will not do.

Module: Enlarging the sphere of concern

The integral point, and the actual mechanism of the whole course: at each cultural paradigm shift — pre-modernity to modernity and onward — we enlarge the zone of our moral concern and care.

Module: Interbeing

Thich Nhat Hanh, and Interbeing as a candidate for the binding we'd need. If planetary cooperation requires something that holds us together as humanity, this is the most developed living attempt at it, and the honest question is whether 'religion' is the word for what's required.

Open questions and to-dos

Session 1 examples. The research pass still to do — cases and video clips, successes and failures. Session 1 needs the most and currently has the least.
Ostrom. The module I know least well. Needs real reading.
Session 4 is dense. Five substantial modules including the applied record. If anything overflows, Changing the payoffs is the compressible one, since the audience is least likely to need convincing on it.
Format. Prepared material plus discussion, versus lecture and improvise. Sessions 3 and 4 are where prepared reading would pay off; 1 and 6 probably work better cold.
Prep. If a Tree Falls before Session 1. Consider one short reading before Session 4.