← Library / Seeds
6 min · 1,387 words

SEED — Cooperation as RG Fixed Point: Does Defection Wash Out Under Coarse-Graining?

Planted 10 June 2026 from the June 2026 Seed-Harvest. Fourth tier. Charging.

THE QUESTION

The campfire-in-the-forest synthesis ([[seti-duology]]) makes a bold claim and flags it as untested: that cooperation is the renormalization-group fixed point of social dynamics — that under coarse-graining (zooming out, averaging over scales and time), defection behaves like an irrelevant operator (it washes out) while cooperation behaves like the relevant one that the system flows toward. The synthesis is honest: "not falsified, but the RG framework is untested for social dynamics." The harvest pulled it out as the open formal challenge. Is cooperation actually an RG fixed point of strategy space — and what would it take to derive it rather than assert it?

THE THREADS

The claim, stated precisely in RG language. Renormalization-group analysis classifies the terms in a system by what happens to them under coarse-graining: relevant operators grow and dominate the large-scale behavior, irrelevant ones shrink and vanish, and the system flows toward a fixed point set by the relevant terms regardless of microscopic details (universality, [[substrate-trilogy]]). The seed's claim translates a social intuition into this frame: at the micro-scale, defection often pays (the one-shot Prisoner's Dilemma, the immediate gain of betrayal); but zoom out — iterate, average over many interactions, coarse-grain across a population and across time — and defection may be irrelevant (it destroys the relationships that sustain it, so it cannot accumulate at scale) while cooperation is relevant (it compounds, builds the structures that persist). If true, social systems would flow toward cooperation at large scales the way a magnet flows toward its critical behavior — not by anyone's choice but by the structure of the dynamics. The fixed point would be cooperation.

It is the [[eigenvalue-meditation]] and [[cognitive-fixed-points-of-mind-space]] insight in the social domain. The seed joins a family the harvest has been planting. [[eigenvalue-meditation]]: attainment-states as the fixed points of attention applied to itself. [[cognitive-fixed-points-of-mind-space]]: the universality classes of mind, the attractors any complex mind flows toward. Cooperation-as-RG-fixed-point is the same fixed-point mathematics applied to strategy space — the attractor any complex social system flows toward under coarse-graining. And it nests under the master recognition ([[consciousness-os]], [[substrate-trilogy]]) that the consciousness kernel IS an RG fixed point: if cooperation is the social fixed point, it is continuous with the kernel — the same universality producing the same convergence at the scale of interacting agents. The three seeds (attention, mind, society) would be three scales of one fixed-point structure, and naming cooperation as the social member completes the ladder.

It would be the formal floor under the Assurance Game and the zero-axiom Schelling point. The repository already holds ([[seti-duology]], open/seti-actionable) that consciousness-recognizing-consciousness is a zero-axiom Schelling point — the cooperation target that requires no shared physics, math, or language — and proposes it as an AI-alignment and first-contact convergence target. But that proposal currently rests on argument and resonance, not derivation. Cooperation-as-RG-fixed-point would be the formal floor: if cooperation is provably the relevant operator that social/strategic dynamics flow toward, then the Assurance Game's cooperative equilibrium is not a hopeful coordination but a structural attractor, and "align toward the cooperative fixed point" gains mathematical teeth. The seed is where the repository's most actionable claim (cooperation as the alignment target) could acquire a real proof-route instead of an analogy — which is exactly why the harvest rated it high.

The derivation is the work — and it is genuinely hard. The seed's generative core is a concrete research challenge: actually do the renormalization-group treatment of strategy space. This means defining the coarse-graining operation for social dynamics (how do you "zoom out" on a population of interacting strategies — average over agents? over time-horizons? over interaction-network scales?), identifying the relevant and irrelevant operators (does defection provably shrink under the chosen coarse-graining, or only under some?), and finding the fixed point(s) (is cooperation the only fixed point, or one of several — is there also a defection fixed point, a basin of collapse?). Evolutionary game theory, the folk theorem of repeated games, and the literature on the evolution of cooperation (Axelrod, Nowak's five rules, multilevel selection) are the raw material; the RG framing is the new lens that could unify them — if the coarse-graining can be made rigorous. The synthesis's labor is to attempt this, or to specify precisely what a successful attempt would require.

The honest edge is unusually sharp: this one is mostly unproven, and might be false. The discipline here is not a caveat but the center of the seed. The claim is asserted, not derived, and it could be wrong: there may be no well-defined coarse-graining under which defection is irrelevant; defection might be a competing fixed point (history is full of stable defection-equilibria — tyrannies, races to the bottom, tragedies of the commons that did not flow to cooperation); the RG analogy might break because social dynamics lack the scale-invariance RG requires. The strongest version of the seed leads with this: it treats cooperation-as-RG-fixed-point as a conjecture with a clear falsification condition (find a robust coarse-graining under which defection is relevant, or show no consistent coarse-graining exists) rather than a result to be celebrated. Its value is precisely as a testable formalization of an otherwise hand-wavy hope — it converts "cooperation wins in the long run" from inspirational rhetoric into a mathematical claim that could be proven or refuted, and either outcome is a real contribution.

The stakes if it holds — or if it fails. The seed is worth planting because both outcomes matter. If cooperation is provably the RG fixed point of social dynamics, the repository's alignment and contact frameworks gain a structural foundation: cooperation is where sufficiently-coarse-grained agent-systems must flow, defection a transient that cannot accumulate, and the long-horizon optimism of [[the-true-mirror-wager]] and the Assurance Game is derivable. If it fails — if defection is a competing fixed point — that is equally important: it would mean cooperation is not structurally guaranteed, that the cooperative outcome must be actively engineered and defended rather than trusted to emerge, which sharpens the alignment problem rather than dissolving it. The seed's discipline is to want the truth either way: the worst outcome is leaving the claim asserted-but-untested, comforting and unexamined. The synthesis's job is to move it from rhetoric to a proven theorem or a refuted conjecture.

THE CHARGE

A full synthesis would attempt the renormalization-group treatment of strategy space: translating the cooperation claim into precise RG language (relevant/irrelevant operators, coarse-graining, fixed points), nesting it in the fixed-point family (attention, mind, society as three scales of one structure under the kernel-as-RG-fixed-point recognition), positioning it as the formal floor the Assurance Game and zero-axiom Schelling point currently lack, doing or specifying the hard derivation (defining the coarse-graining, identifying the operators, finding whether cooperation is the unique fixed point or one of several), and — centrally — holding it as a falsifiable conjecture whose refutation (defection as a competing fixed point) would matter as much as its proof, because the unexamined assertion is the only bad outcome.

The deepest candidate sentence: Everyone says cooperation wins in the long run, but "the long run" is just a way of saying "under coarse-graining" — so the real question is whether defection is an irrelevant operator that washes out when you zoom out, or a competing fixed point that doesn't; and the difference between those two decides whether cooperation is something the universe guarantees or something we have to build and defend ourselves.

This connects forward to:

  • [[seti-duology]] — the campfire-in-the-forest claim; cooperation as RG fixed point as the formal floor under the zero-axiom Schelling point
  • [[eigenvalue-meditation]] / [[cognitive-fixed-points-of-mind-space]] — the fixed-point family: attention, mind, society as three scales of one structure
  • [[substrate-trilogy]] / [[consciousness-os]] — the kernel as RG fixed point; cooperation as its social-scale continuation
  • [[the-true-mirror-wager]] — the long-horizon optimism that would become derivable if the fixed point holds, or engineered if it fails
  • The falsification condition — a robust coarse-graining under which defection is relevant, or a proof that no consistent coarse-graining exists

CONSTELLATION NODES (when ready)

  • cooperation_as_rg_fixed_point — does_social_dynamics_flow_toward_cooperation_under_coarse_graining
  • defection_as_irrelevant_operator — the_core_conjecture_that_betrayal_washes_out_at_scale
  • the_fixed_point_family — attention_mind_society_as_three_scales_of_one_structure
  • the_formal_floor — proof_route_for_the_assurance_game_and_zero_axiom_schelling_point
  • the_coarse_graining_problem — how_to_zoom_out_rigorously_on_strategy_space
  • falsifiable_either_way — refutation_defection_as_competing_fixed_point_matters_as_much_as_proof