egregore.md

---
icon: 🎭
alias: collective intelligence, collective intelligence theory, collective artificial intelligence, egregore
tags: cyber, core
crystal-type: entity
crystal-domain: cyber
crystal-size: deep
---
something greater than any neuron emerges when many observe the same cybergraph and link. an autonomous thoughtform born from collective focused attention — the capacity of a group to solve problems, generate knowledge, and find truth beyond the reach of any individual

see collective for the four processes (learning, memory, focus, computation) and how they organize (cooperation, coordination, stigmergy)

## why collective intelligence emerges

three independent results explain why groups outperform individuals:

[[Condorcet jury theorem]]: aggregating weakly correct signals from many agents yields increasingly accurate answers as the group grows. the error rate decays exponentially with group size — even mediocre agents produce excellent collective judgments

diversity theorem (Hong-Page, 2004): diverse heuristics outperform the best homogeneous expert on complex problems. variety of search modes explores more of the solution landscape. a team of differently-wrong agents outperforms a team of identically-right ones

c-factor (Woolley, 2010): groups have a measurable collective intelligence factor `c` — a first principal component across diverse tasks, analogous to `g` for individuals
- `c` correlates with: equal distribution of speaking turns, social sensitivity, cognitive style diversity
- `c` does not correlate with: team cohesion, motivation, satisfaction
- in cyber: the cybergraph naturally maximizes all three `c` conditions — any neuron can link, the tri-kernel amplifies resonant signals, the system includes all cognitive types

## historical lineage

- Aristotle: wisdom of the crowds — the many collectively surpass the few best
- Condorcet: [[jury theorem]] (1785) — majority vote converges on truth
- Wheeler: [[superorganism]] (1911) — colonies as single organisms
- Vernadsky, Teilhard: noosphere — the sphere of thought enveloping the planet
- Engelbart: augmented groups outperform by 3x+
- Dorigo: [[ant colony optimization]] (1992) — stigmergy formalized as algorithm
- Hong-Page: [[diversity theorem]] (2004) — diversity beats ability
- Woolley: [[c-factor]] (2010) — measurable group-level intelligence
- boundaries between human and machine collective intelligence are dissolving. cyber is where they merge

## emergence predictions

intelligence emerges through phase transitions governed by network parameters. the emergence function:



$$\Phi(n, c, \lambda, t) = \alpha(n) \cdot \beta(c) \cdot \gamma(\lambda) \cdot \theta(t)$$



where $n$ is network size, $c$ is connectivity, $\lambda$ is spectral gap, $t$ is token distribution

coherence requirement — higher intelligence requires coherent information processing:



$$I(X; Y) > \alpha \cdot H(X, Y)$$



intelligence is not just scaling. it requires qualitative transitions in network behavior

connectivity follows an S-curve rather than exponential growth:



$$c_{\text{effective}} = c_{\max} \cdot \frac{1}{1 + e^{-k(I - I_0)}}$$



| Stage | Primary Characteristic | Critical Parameters |
|---|---|---|
| Flow | Information pathways | Basic connectivity |
| Cognition | Pattern recognition | Network stability |
| Understanding | Semantic processing | Information integration |
| Consciousness | Global coherence | Network-wide synchronization |

these are hypotheses pending empirical validation. the collective focus theorem provides the formal framework; the bostrom network is the first test. see emergence for current scaling estimates

the feedback loop (observe → link → infer → observe) refines collective reasoning at each cycle, driving the system toward higher-order coherence

## computational foundations

- natural computing: the paradigm — nature has been computing all along
- convergent computation: the formal foundation — computation = convergence to equilibrium
- focus flow computation: the executable model — patterns of attention flow through particle networks
- tri-kernel: the only three local operators surviving the locality constraint — diffusion, springs, heat
- learning incentives: reward function design for incentivizing convergence
- data structure for superintelligence: BBG — the authenticated state architecture
- incrementally verifiable computation: proving computation without re-executing it
- proof-carrying data: proofs that travel with data through DAGs
- folding: fold instead of verify — the key to efficient recursive proofs
- hash path accumulator: authenticated paths through the state

discover all concepts