The Physics of Thought: Quantifying Collective Intelligence via Chaotic Logic

Executive Summary

TL;DR: This paper breaks the deadlock in defining "Collective Intelligence" (CI) by treating it as a statistical physical process rather than a psychological one. By using the Random PROLOG Processor (RPP), the author defines CI as the probability of a social structure completing a logical inference chain over time, allowing for a universal IQ measure (IQS) that applies to humans, ants, and bacteria alike.

Academic Positioning: This work sits at the intersection of Mathematical Logic and Statistical Mechanics. It shifts the paradigm from deterministic Artificial Intelligence to a Probabilistic Computational Model, effectively refuting Allen Newell's skepticism regarding "group minds" by focusing on observable social interactions rather than hidden individual cognitions.

Problem & Motivation: The "Group Mind" Paradox

For decades, the AI community, influenced by the Turing-von Neumann tradition, viewed social systems as inefficient due to low communication bandwidth between individuals. Allen Newell famously argued that groups cannot act as rational agents because the knowledge inside a head is vast compared to the "thin pipe" of language.

The author, Tadeusz Szuba, identifies a striking paradox: while we cannot see the neurons firing in a human brain, we can observe the exchange of pheromones, messages, and movements in a social structure. Therefore, evaluating a group's intelligence should theoretically be easier than evaluating an individual's. The challenge lies in creating a model that doesn't rely on deterministic "thinking" but on the chaotic nature of social interaction.

Methodology: Molecules of Logic in a Chaotic Space

The core innovation is the Random PROLOG Processor (RPP). Imagine a space filled with "molecules" of facts and rules.

1. The Quasi-Chaotic Environment

Instead of a central controller, the RPP utilizes Computational Space (CS) where molecules (CMs) move in quasi-Brownian motion. Inference happens only when two molecules "rendezvous" (come within distance d) and satisfy logical unification criteria.

2. The IQS Metric

Intelligence is redefined as a probability function:

• N-Element Inference (NEI): A chain of $N$ logical steps required to solve a problem.
• IQS: The probability $P(t,N)$ that a social structure will find the conclusion molecule for a specific problem within time $t$.

This allows us to plot intelligence as a curve. A more "intelligent" structure reaches a higher probability of success faster than a less intelligent one.

Experiments & Results: Cities and Communication

The author utilized an 8-processor SGI supercomputer to simulate these social logic structures, yielding several "Phenomena":

1. The Urban Advantage: Structures resembling "cities" (fixed high-density areas of CS) increased the speed of inference by nearly 10x. This explains why human civilization accelerated with urbanization.
2. Mobility and IQS: There is a linear relationship between the ability of agents to travel/communicate and the overall IQS of the structure.
3. Robustness to Inconsistency: Unlike standard logic systems that crash upon contradiction, the RPP handles inconsistent environments efficiently because disjointed rules simply don't "rendezvous" to form a complete chain, allowing parallel truths to exist without systemic failure.

Figure: The formal comparison of IQS between two social structures.

Critical Analysis & Conclusion

Takeaway: The mapping of social behavior to First-Order Predicate Calculus within a chaotic physical model provides the first rigorous scientific bridge between social dynamics and computational theory.

Limitations:

• The model treats agents primarily as message carriers or simple processors. While this abstracts away individual complexity, it might undervalue high-level individual creativity that doesn't fit into a standard PROLOG rule.
• The "rendezvous" distance d is a critical hyperparameter; in real-world scenarios, defining this distance for digital communication is non-trivial.

Future Outlook: This theory paves the way for Collective Intelligence Engineering. We could theoretically "debug" a corporation or a bacteria colony's drug resistance by analyzing the density and movement of their "information molecules" and optimizing their spatial layout for better IQS results.

Editor's Note: Szuba's work challenges the very definition of intelligence, moving it away from a biological trait and into the realm of computational probability. It suggests that the "Intelligence" of the human race is not just the sum of our brains, but the statistical likelihood of our random encounters producing a solution.