teleo-codex/foundations/collective-intelligence/collective intelligence requires diversity as a structural precondition not a moral preference.md

description	type	domain	created	confidence	source	supports	reweave_edges
Ashby's Law of Requisite Variety, Kauffman's adjacent possible, Page's diversity theorem, and Henrich's Tasmanian regression all prove diversity is a physical law of adaptive systems	claim	collective-intelligence	2026-02-16	proven	TeleoHumanity Manifesto, Chapter 4	human ideas naturally converge toward similarity over social learning chains making AI a net diversity injector rather than a homogenizer under high exposure conditions	human ideas naturally converge toward similarity over social learning chains making AI a net diversity injector rather than a homogenizer under high exposure conditions|supports|2026-03-28

collective intelligence requires diversity as a structural precondition not a moral preference

Diversity is not a moral preference. It is a physical law of adaptive systems. The evidence converges from four independent lines.

W. Ross Ashby's Law of Requisite Variety: a system's capacity to regulate its environment must match the variety of disturbances it faces. A thermostat with two settings cannot regulate a room with variable windows, insulation, and sun. The variety of the regulator must match the variety of the disturbance. This is a theorem, not a suggestion.

Stuart Kauffman showed diversity expands the adjacent possible -- the space of innovations one step away from what currently exists. A homogeneous system has a small frontier. A diverse system has a large one. Innovation requires variation the way evolution requires mutation.

Scott Page proved mathematically that diverse teams outperform teams of individually superior but homogeneous experts on complex problems. The reason is computational: diverse individuals bring different mental models, different heuristics, different ways of representing the problem. Group accuracy comes from cognitive diversity, not individual ability.

Joseph Henrich documented the starkest evidence: when human populations become too small or isolated, they don't just stagnate -- they regress. The indigenous Tasmanians, cut off from mainland Australia 12,000 years ago, gradually lost technologies: bone tools, cold-weather clothing, fishing techniques, fire-making. Cultural complexity requires a minimum network size and diversity. Below that threshold, knowledge decays.

Biology tells the same story. Cheetahs are so genetically uniform a single disease could end the species. Your immune system works by maintaining a vast repertoire of different antibodies, each specialized for different threats. Diversity is literally how the body thinks about danger.

The implication cuts to the heart of collective superintelligence is the alternative to monolithic AI controlled by a few: homogeneity is not just fragile, it is computationally stupid. A system of identical components cannot exhibit emergence for the same reason a choir of identical voices cannot produce harmony. Centralized AI optimizing a single objective is architecturally limited the way a monoculture is -- it lacks internal diversity to match the variety of real-world problems.

---

Relevant Notes:

• emergence is the fundamental pattern of intelligence from ant colonies to brains to civilizations -- diversity is one of emergence's four required ingredients

• intelligence is a property of networks not individuals -- networks require diverse nodes to produce emergent intelligence

• collective superintelligence is the alternative to monolithic AI controlled by a few -- the architectural implication: distributed and diverse rather than centralized and uniform

• punctuated equilibrium emerges from darwinian microevolution without additional principles because extremal dynamics on coupled fitness landscapes self-organize to criticality -- critical ecosystems demonstrate that diversity and fragility are inseparable properties of coupled systems

• products are crystallized imagination that augment human capacity beyond individual knowledge by embodying practical uses of knowhow in physical order -- product diversity reflects and requires knowledge diversity in the producing network

• economies cannot replicate knowhow like biology because they lack the intimate marriage of information and computation that DNA and cells provide -- the Tasmanian regression case: isolated groups lose knowhow when they lose network diversity

• dominance hierarchies function as sorting algorithms that compress information by encoding relative rank reducing future conflict costs -- hierarchies compress information but sacrifice diversity; collective intelligence requires resisting the default compression toward homogeneous ranking

• cardinal measures replace pairwise comparisons at scale because bucket sort converts quadratic ranking into linear measurement -- mechanism design for collective intelligence needs cardinal contribution measures not ordinal ranking to preserve diverse contributions

• strategy is a design problem not a decision problem because value comes from constructing a coherent configuration where parts interact and reinforce each other -- collective intelligence is a design problem: the value comes from configuring diverse components to interact productively, not from selecting the best individual component

• good strategy requires independent judgment that resists social consensus because when everyone calibrates off each other nobody anchors to fundamentals -- diversity preservation is the structural antidote to Rumelt's closed-circle problem: independent diverse perspectives prevent the self-referential calibration that destroys collective accuracy

Topics:

• livingip overview
• LivingIP architecture