#40 — Complexity & Stupidity

Episode Summary

Executive Summary: Sam Harris interviews David Krakauer about information, complexity, intelligence, and stupidity. Krakauer distinguishes mathematical concepts from metaphors, argues that brains and cultures really do process and store information, and frames intelligence as problem solving that is better than chance. He also claims culture and cognitive artifacts can augment human intelligence, while some modern institutions may be degrading it.

Main Topics: The Santa Fe Institute and consilience across disciplines (Priority: 5/5): Krakauer describes SFI as an institution built to study complex adaptive systems across biology, economics, physics, archaeology, and computation, rejecting rigid disciplinary boundaries. Information as a mathematical concept (Priority: 5/5): He explains information via Shannon and thermodynamics: information reduces uncertainty and can be understood as the negative of entropy, not merely as a loose metaphor. The brain as an information-processing system (Priority: 5/5): Krakauer defends the claim that brains encode, transmit, combine, and store information, while clarifying that the brain is not a literal von Neumann computer. Complexity as description length (Priority: 4/5): Complex systems are characterized by long descriptions that sit between highly regular and highly random phenomena; mice, societies, and brains are hard to compress into simple equations. Intelligence vs. stupidity (Priority: 5/5): Intelligence is defined as behavior that solves problems efficiently and better than chance; stupidity is rule-following that is worse than chance or interminably inefficient. Culture and cognitive artifacts as intelligence amplifiers (Priority: 5/5): Numbers, language, and other cultural tools are presented as artifacts that extend human reasoning and physically reshape cognition over time. Plasticity, education, and limits of innate IQ (Priority: 4/5): Krakauer argues that training and deliberative practice can greatly expand capabilities, suggesting IQ reflects trainable working memory more than fixed innate intelligence.

Key Arguments: Scientific categories are increasingly porous, so institutions like SFI are designed to focus on common principles of complex adaptive systems rather than disciplines. Information has a precise mathematical meaning: it is the reduction of uncertainty, historically linked to entropy through Shannon's work. The brain does process information in the relevant mathematical sense: sensory inputs are transduced, combined, and stored as structured representations. Claims that biological systems do not process information often confuse metaphor with mathematical model and fail to account for neuroscience and neuroprosthetics. Complexity is partly a function of description length; complex phenomena are neither highly ordered nor fully random. Intelligence should be judged relative to chance and problem efficiency, not by species-specific IQ tests alone. Culture is not just passive background; it is a machine for amplifying cognition through cognitive artifacts like numbers and language. Human intelligence is highly plastic, and education/practice can materially increase working memory and skill, making culture and upbringing central to cognitive development.

Data Points: SFI founding period: mid-1980s - The Santa Fe Institute was founded in the mid-80s by Nobel laureates and others. IAS founding period: 1930s - Krakauer contrasts SFI with the Institute for Advanced Study, which dates to the 1930s. Entropy/information historical origin: 1870s - Boltzmann’s H-theorem and molecular chaos are described as early foundations for the information-entropy relationship. Turing's breakthrough year: 1936 - Turing answered Hilbert’s question and created the Turing machine. Hilbert's question year: 1928 - Hilbert asked whether a machine could decide the truth or falsity of mathematical propositions in reasonable time. von Neumann's book year: 1958 - John von Neumann’s The Computer and the Brain is cited as a major metaphorical bridge between computation and neuroscience. Roman numeral era in Europe: about 1500 years - Krakauer cites the long use of Roman numerals in Western Europe as a cultural example of stupidity. Ancient Sumerian cuneiform numbers: about 5,000 years old - Used to illustrate the ancient history of numeric systems. Cognitive artifact training example: up to 300 numbers - Krakauer says some people trained with encoding techniques can remember around 300 numbers, versus the older limit of about 10-11. Short-term number memory baseline: 10-11 numbers - He references classic working-memory findings about how many digits people can hold in mind. Brain energy use: about 20% of body energy - He notes the brain consumes roughly one-fifth of the body's energy. Brain power consumption: about 20 watts - He compares brain energy use to about the power of a light bulb. Cube solution space: on the order of 10 quintillion solutions - Used in the intelligence/stupidity example to contrast random search with efficient algorithms.

Pivotal Quotes: "we're grappling with the problem of the evolution of intelligence and stupidity on Earth" — David Krakauer: Krakauer summarizes his scientific agenda near the start of the conversation. "information is in fact the negative of thermodynamic entropy" — David Krakauer: He explains Shannon’s connection between information theory and statistical physics. "Intelligence is the thing that we do that ensures that the problem is very efficiently solved" — David Krakauer: He defines intelligence in terms of problem-solving efficiency and performance better than chance.

Implications: The conversation suggests that intelligence is not fixed or purely individual: culture, education, and cognitive artifacts can extend it, while bad institutions and misleading metaphors can diminish it. For research and policy, the message is to study cognition as an adaptive, information-rich system and to design environments that increase collective intelligence.