Part II - The Games#
Contents
Five strategic structures sit on top of the physical substrate. Each has a different game-theoretic character, and getting the character right matters - most public analysis applies prisoner's-dilemma logic to situations that are not prisoner's dilemmas, and then recommends coordination mechanisms that cannot work on the actual payoff structure.
The five#
| Players | Structure | Key result | |
|---|---|---|---|
| 1 - Labs | Frontier labs | Tullock contest | Over-dissipation; consolidation to 3–5 by 2029; open weights ~3–6 mo measured (authored ~9–15; ground note r25) |
| 2 - Nations | States | Security dilemma with leakage | No binding agreement before a salient accident; compute-governance trap |
| 3 - Firms | Adopters | Red Queen / Bertrand | Adoption mandatory; surplus to consumers (competitive markets) |
| 4 - Labor | Workers, firms | Commons tragedy in human capital | Apprenticeship gap; aggregate displacement is an upper bound once demography is included |
| 5 - Information | Signalers, evaluators | Signaling collapse → lemons market | Verification replaces generation; → meaning |
What they have in common#
Four of the five are public-goods problems with private costs: safety (Game 1), restraint (Game 2), junior training (Game 4), and provenance infrastructure (Game 5). Each will therefore be undersupplied by the market and supplied late, after a forcing event. → C2 shelf-readiness, C8 provenance
Game 3 is the exception and the one that explains the most: it is a pure competitive-dissipation result, and it is why the aggregate statistics will look disappointing even while the technology works exactly as advertised. Scope: competitive markets specifically (steelman §2).
The games compose in a specific order#
The five are not parallel; they feed each other, and the coupling is directional. The substrate sets the entry price of Game 1. Game 1's output - frontier capability plus an open-weight tier that trails it - is an exogenous parameter of every other game: it sets what Game 2 must try to control, what Game 3's firms adopt, what Game 4's tasks face, and how cheap Game 5's forgeries are. Game 3's dissipation then feeds back into Game 1 by shaping the prize: if adopter surplus flows to consumers rather than to adopters, the frontier prize is smaller than the raw capability suggests, which is one quiet reason consolidation pressure builds even while usage grows.
The practical rule this ordering yields: an intervention in an upstream game moves every game downstream of it; an intervention downstream moves almost nothing upstream. Compute and energy policy touch all five. Content-provenance mandates (Game 5) touch only Game 5. Most proposed governance operates downstream because downstream is where harms are visible, which is a structural reason to expect regulation that is simultaneously popular and weak.
Timescale mismatch as a shared failure mode#
Each game equilibrates on a different clock: Game 1 in financing cycles (quarters to a few years), Game 2 in incident-and-legislation cycles (years), Game 3 in competitive diffusion (years), Game 4 in career and cohort time (decades), Game 5 in norm time (fast to break, slow to rebuild). Analysis that samples all five at one moment will always find some games "not happening" - usually 4 and 5, whose equilibria arrive last. The failure mode for this framing itself: if capability gains stall, the slow games never reach the equilibria described here, and the snapshot critics were right for the wrong reason.
Cross-links the short table hides#
| Interaction | Where developed |
|---|---|
| Labs × nations: export controls spend the verification option | Game 2, bipolar |
| Labs × capital: correction culls commercial players, leaves state-adjacent | Game 1, Capital |
| Nations × insurance: correlated failure as forcing event | Game 2, Uncertainty 6 |
| Firms × prices: two-economy split | Game 3, Prices |
| Labor × demography: aggregate softens, composition hardens | Game 4, Demography |
| Information × education/media/meaning | Game 5, B6, B11 |
| Labs × firms: lag compression shrinks prize, raises over-dissipation | Game 1, inference |
| Nations × finance: jurisdiction brands the incident | Game 2, finance |
Reading order when short on time#
Game 3 first (where surplus lands), then Game 4 (who loses the pathway), then Game 1 + Game 2 as the supply-side race. Game 5 last - it is the atmosphere everything else breathes, easiest to underweight until education and media make it concrete.