{"path":"research/fractal-priors-for-convergence.md","content":"# Fractal Priors for the Convergence Wager\n\n**Date**: 2026-08-21 · **Occasion**: founder, on hesitating to list the wager as an \"open problem\": *\"this is one of my deepest convictions and I am hesitating about listing as open problems the convictions that seem to be based on quite strong fractal priors. VERY curious what fractal priors there might be in various theory-strands supporting my convergence wager.\"* · **Method**: the [fractal-priors-for-success.md](fractal-priors-for-success.md) discipline applied to the wager itself — theory-strands where independent starting points demonstrably converge, each checked for whether its mechanism transfers, with the corpus's own counter-findings held in view throughout. **Grounding**: [convergence.md](../convergence.md) read in full this session; [convictions-across-scales.md](convictions-across-scales.md) § the wager split; [heylighen-value-complexity-and-deliberus.md](heylighen-value-complexity-and-deliberus.md); the MGE and Curry citations as carried in convergence.md.\n\n## §0 The headline: the priors don't just support the wager — they predict its sharpened form\n\nSweeping the strands below, one pattern recurs so consistently it deserves to be the finding: across domains, what converges is **structure**, and what stays divergent is **weighting**. The same value-circle with different rankings; the same moral foundations with different emphases; the same seven cooperative rules with different local elaborations; the same core theorems from different axiom systems. That is not the naive wager (\"everyone shares the same bedrock, full stop\") — it is precisely the **sharpened wager the corpus reached empirically in July 2026**: decomposition converts vague clashes into *typed, narrow residues*; shared basis, contested weights; and the weights are exactly what the fittingness and contextual residue types hold. Two independent arrivals at the same refinement — one from cross-domain priors, one from running descents — is the strongest kind of support a conviction can have short of the experiment itself.\n\n## §1 The strands that support, graded\n\n**Measured value-structure universality (strong).** The Schwartz value survey finds the same *circular structure* of human values — which values neighbor which, which oppose which — replicating across dozens of countries, while the *rankings* differ freely by culture. Structure converges; weights diverge. This is the wager's sharpened form as a standing cross-cultural measurement, and it has been replicated for three decades.\n\n**Cooperative morality across all societies (strong, already in corpus).** Curry's seven cooperative rules, hand-coded in 60 societies ([Curry, Mullins & Whitehouse 2019, *Current Anthropology*](https://doi.org/10.1086/701478)) and machine-read across 256 — the entire HRAF corpus ([Alfano, Cheong & Curry 2024, *Heliyon*](https://doi.org/10.1016/j.heliyon.2024.e25940)). **Stated precisely, because three surfaces of this corpus had drifted into an overclaim** (corrected 2026-08-21): both studies found *most of the seven morals in most societies*, not all seven everywhere. The strong finding is about **valence, not ubiquity** — wherever these behaviours appear in the ethnographic record their moral valence is uniformly positive, with zero societies treating them as bad, and with equal frequency across cultural regions. That is the claim worth making, and it is the one [moral-convergence-metaethics-and-psychology.md](moral-convergence-metaethics-and-psychology.md) §4.4 had right all along. One refinement matters against the earlier sweep's caution: [convictions-across-scales.md](convictions-across-scales.md) rightly noted this support is *phylogenetic* (shared by descent) rather than *decompositional* (forced by logic) — but **the wager does not require the bottom to have been produced by decomposition, only that descents find it.** A shared bottom that exists because all humans inherit the same cooperative predicament is still a shared bottom under the drill. The analogy-argument was weak; the direct-evidence argument stands.\n\n**Moral foundations theory, read correctly (moderate, and it flips a famous objection).** Haidt's program is usually cited *against* universalism: tribes weight the foundations differently. But its actual finding is a small shared basis (five-ish foundations) with divergent weightings — which is the sharpened wager again, stated by its supposed opponent. The disagreement-about-weights is what fittingness residues are *for*.\n\n**Autopoietic valence (moderate-deep).** Heylighen grounds value itself in self-maintenance: every human shares the survival architecture that mints valence in the first place — pain, fear, attachment, curiosity as a nested hierarchy of evolved evaluators. Whatever else diverges, the machinery generating value is common stock. This is the deepest naturalistic reason to expect *something* shared at the bottom of every descent ([heylighen-value-complexity-and-deliberus.md](heylighen-value-complexity-and-deliberus.md) §2).\n\n**Structured elicitation converges when tried (strong, small-n).** The MGE result: when values are elicited *structurally* — context by context, wiser-than by wiser-than — participants overwhelmingly converge on directionality, without being told the directions. The one time the wager's mechanism has been run at experimental scale, it went the wager's way.\n\n**Reverse mathematics (novel to the corpus, suggestive).** In foundations of mathematics, strikingly different axiom systems turn out to recover the same core body of theorems — most of ordinary mathematics needs only a small shared fragment, and the program of \"reverse mathematics\" maps exactly which axioms each result truly requires. Different bedrocks, same load-bearing middle. As a formal-domain rhyme for \"apparently incompatible foundations, largely shared building,\" it is the cleanest one available — and its method (trace each result to the axioms it *actually* needs) is structurally the descent.\n\n**Universality classes (suggestive analogy, held lightly).** In statistical physics, microscopically different systems flow to *identical* large-scale behavior — a handful of fixed points attract vast basins of unlike systems. The mathematics says: many configurations, few attractors, details wash out under the right flow. Mapped here: worldviews as basins, bedrock values as fixed points, decomposition as the flow. It is an analogy, not evidence — but it is the standard mathematical demonstration that \"endless surface diversity, few deep destinations\" is a *normal* shape for complex systems, not a hopeful one.\n\n**Independent rediscovery of the same ethics (folk-level but old and broad).** The golden rule derived independently across traditions with no contact; Brown's catalogue of human universals (reciprocity, in-group murder prohibitions, fairness norms); creole languages crystallizing similar grammar wherever strangers must build a code fast; color vocabularies across languages evolving along one constrained path (Berlin–Kay). Independent lineages under shared constraints keep finding the same solutions — the same regularity convergent evolution shows in biology (eyes evolved forty-plus times).\n\n**Parfit (expert testimony, contested).** The triple theory — consequentialism, Kantianism, contractualism climbing the same mountain — already in [convergence.md](../convergence.md); kept at its honest weight: a serious philosopher's convergence claim, not a consensus.\n\n## §1b The weightings-diverge gallery (added 2026-08-21 for founder reflection; unpacked 2026-08-21)\n\n**The pattern in one sentence.** In domain after domain, everyone turns out to be working with the *same short list of things that matter* — and to disagree about *how much each one counts*. The list is the structure. The how-much is the weighting.\n\n**The clearest example, because nobody has a stake in it.** Every kitchen on earth works with the same handful of tastes: salt, fat, acid, heat, sweetness, bitterness, umami, aroma. No cuisine denies that salt exists or claims acid is irrelevant. A cuisine simply *is* a set of weights on that shared list — umami-forward in Japan, sour-and-salt in Scandinavia, heat-plus-numbing in Sichuan. Two cooks can disagree completely about a dish while sharing every ingredient category. That is the whole shape, and it recurs:\n\n| Domain | What everyone shares | What differs |\n|---|---|---|\n| **Human values** (Schwartz) | The same *circle*: which values sit next to which, and which are genuine opposites — you cannot maximise novelty and security at once, anywhere | Which slice of the circle a culture ranks first. Replicated across dozens of countries for three decades |\n| **Moral foundations** (Haidt) | A small set — care, fairness, loyalty, authority, sanctity — present in everyone | How heavily each is loaded. Usually cited *against* universalism; read correctly it is shared basis, divergent weights |\n| **Cooperative morality** (Curry) | Seven rules — help kin, help your group, return favours, be brave, defer, divide disputed resources fairly, respect property — judged *good* wherever they appear, across 60 hand-coded and 256 machine-read societies, with no society reversing the valence | Honour cultures weight bravery and deference up; market societies weight reciprocity and fairness up |\n| **Personality** | The five-factor structure replicates cross-culturally | Which traits a culture rewards |\n| **Cuisine** | The same taste dimensions | The cuisine *is* the weighting |\n| **Music** | The octave, and simple consonant intervals, recur in unrelated traditions — grounded in the physics of vibrating air and the ear | Maqam, raga and blues weight the shared acoustic material differently: which intervals lead, which are ornamented, what counts as resolution |\n| **Language** | A common toolkit — reference, predication, negation, marking questions | What a language *obliges* you to mark: Quechua forces you to say how you know something, Japanese forces politeness register, English forces tense |\n| **Colour naming** (Berlin–Kay) | One perceptual space, and terms added along a strongly constrained path (black/white → red → green/yellow → blue → …) | How many distinctions a culture deems worth naming, and where the boundaries fall |\n| **Law** | Every system balances harm, autonomy, procedural fairness, property, dignity | America weights expressive liberty above dignity; Germany weights dignity above expression. A weights fight, not a goods fight — neither denies the other's good |\n| **Bioethics** | The same four principles taught in every medical school: autonomy, beneficence, non-maleficence, justice | Autonomy-weighting in the US against family-weighting across much of Asia |\n| **Political economy** | Efficiency, equality, liberty, stability are on everyone's list | The Nordic and American models share the list and differ in coefficients |\n| **Engineering** | Cost, safety, performance, maintainability | National and firm traditions are weightings of those |\n| **Journalism ethics** | Accuracy, harm-minimisation, independence, accountability | The WikiLeaks argument was a weights fight: disclosure-weighted against harm-weighted, both accepting all four duties |\n| **Parenting** | Warmth, control, autonomy-granting appear everywhere | The named \"styles\" are weightings of those three |\n| **Religion** | Compassion and some form of the golden rule arrive independently, in traditions with no contact | Which virtue is *central* — Parfit's mountain again |\n| **Chess** | All strong play values material, king safety, activity, pawn structure | Tal, Petrosian and Karpov are coefficient sets: initiative over material, safety and prophylaxis, structure and small edges. Nobody is playing a different game |\n| **Evolution** | The same Hox toolkit builds wildly different animals | Lineages express it differentially — a bat is a weighting of the mammal plan |\n\n**And our own two measured cases sit in the same shape.** The minimum-wage pair shares its goods — worker welfare, employment levels, fairness — and fights over weights plus one empirical parameter. Grandma's family shares safety, independence and promise-keeping, and fights over exactly the same thing.\n\n### What the gallery earns, and what it does not\n\n**It earns more than a mood, for one specific reason: most of these domains have nothing to do with human goodness.** Colour, music, cuisine, chess and Hox genes are not moral evidence, and that is the point — the shared-list-divergent-weights shape shows up wherever a domain balances several goods at once. It is a structural regularity, not a hopeful story about people.\n\n**It does not, on its own, prove a shared bottom — and the honest repair matters.** In most rows the shared structure is explained by a shared *constraint*: the same ears, the same retina, the same body plan, the same cooperative predicament. That is shared-by-common-cause, not shared-by-necessity, which is exactly the caution [convictions-across-scales.md](convictions-across-scales.md) raised against the analogy argument. The repair is in §1 and it holds: **the wager does not require the bottom to have been forced by logic, only that descents find it.** A bottom that is shared because every human inherits the same predicament is still a shared bottom under the drill.\n\n**The sharper limitation, which the gallery cannot fix.** It tells you the *shape* of disagreement, not its *size*. Two cultures can share the entire value circle and still go to war over the rankings. So the gallery supports \"what remains is typed and nameable\" far more strongly than it supports \"what remains is small\" — and those are different claims. Weight-fights are real fights.\n\n**Which is why this is a prediction rather than a comfort.** If the pattern holds here, the residue map should fill up mainly with **fittingness and contextual** residues — disagreements about how much a shared consideration counts in this situation — and not with structural or axiom-choice residues, which would mean two parties running on genuinely different goods lists. That is checkable on our own scoreboard, and it can go against us: a map that keeps bottoming out in alien bedrock would say the gallery misled us.\n\n## §2 The counter-strands, kept in full view\n\nThe corpus's own findings stay binding. The cross-scale *analogy* argument for a shared bottom is weak (chemistry bottoms at a hundred elements, not one; the genetic code is shared by descent, not necessity) — the support above is direct-evidence and mechanism, not analogy, which is why it survives that finding. Fogelin's deep disagreements, Chang's parity, and permissivism remain principled *ceilings*, and the accounting rule stands: they count **against** convergence, never as \"work remaining.\" The residue taxonomy is still ought-biased (the is-bedrock gap). Anthropology's diversity record is real — and note that nearly all of it concerns *weightings and elaborations*, which the sharpened wager predicts rather than fears. And the strongest live threat remains internal: an instrument that manufactures convergence (the steering hazard) would fake exactly the result these priors predict, which is why the flattening instruments must keep the power to falsify.\n\n## §3 What this settles about the open-problems page\n\nThe founder's hesitation is right, and this sweep gives it precise grounds: the wager is not an \"unsolved problem\" — it is a **conviction held on substantial cross-scale priors, staked as a falsifiable bet**. Problems are things we don't know how to do; the wager is a position we've taken, with reasons, that the instruments can score against us. If it appears on the open-problems page at all, it belongs only in the capstone \"bets\" register (beside the structure-versus-wiki bet), stated with the confidence the priors warrant — *we expect to win this, for reasons; here is how we would know we lost* — which is also exactly the Teeth-and-Confidence copy register. Founder's call, now properly equipped.\n\n## Cross-references\n\n[convergence.md](../convergence.md) (the conviction's home; MGE, Curry, Parfit, the accounting) · [convictions-across-scales.md](convictions-across-scales.md) (the wager split this doc refines) · [fractal-priors-for-success.md](fractal-priors-for-success.md) (the method) · [heylighen-value-complexity-and-deliberus.md](heylighen-value-complexity-and-deliberus.md) (autopoietic valence) · [moral-convergence-metaethics-and-psychology.md](moral-convergence-metaethics-and-psychology.md) + [empirical-deliberation-convergence-evidence.md](empirical-deliberation-convergence-evidence.md) (the in-house evidence records) · [convergence-wager-red-team.md](convergence-wager-red-team.md) (the standing attacks)\n"}