Darwin Among the Machines: The Evolution of Global Intelligence
Source: https://www.penguinrandomhouse.com/books/72962/darwin-among-the-machines-by-george-dyson/ ↗
Dyson traces the idea that machines might evolve intelligence from its seventeenth-century origins — Hobbes's Leviathan as artificial organism, Leibniz's calculus of reason — through Samuel Butler's 1863 essay that gave the book its title, to von Neumann's cellular automata and the global telecommunications network of the 1990s.
The argument is not that machines will become conscious but that evolutionary dynamics apply to technological systems whether we intend them to or not.
Dyson draws on deep historical research, connecting Turing, Barricelli's digital organisms at the IAS, and the architecture of the early internet into a single lineage.
The result has considerably more historical depth than most books on machine intelligence, treating three centuries of ideas about self-reproducing systems as a coherent intellectual tradition.
Written before the current AI wave, it reads now as remarkably prescient about the trajectory of distributed computation.
Central argument
Dyson argues that evolutionary dynamics — selection, variation, reproduction — apply to technological systems as an inherent property rather than a designed feature, tracing this idea from Hobbes's conception of the state as artificial organism and Leibniz's calculus of reason through Samuel Butler, Turing, Barricelli's digital organisms, and the architecture of the early internet. The central thesis is not that machines will achieve consciousness but that the global telecommunications network already constitutes a self-evolving system, producing emergent intelligence as a structural outcome of distributed computation. Three centuries of thinking about self-reproducing systems form a single coherent lineage that culminates, Dyson contends, in the internet itself.
Critique
The book's strength — its long historical arc — also limits its analytical precision: by treating Hobbes, Butler, von Neumann, and the internet as nodes in a single lineage, Dyson risks flattening genuine discontinuities in what 'intelligence' and 'evolution' mean across those contexts. The analogy between biological evolution and technological change is doing significant explanatory work, but it is never fully stress-tested; the mechanisms of variation and selection in software systems differ enough from genetic reproduction that the metaphor may obscure more than it reveals. A reader looking for a rigorous account of how exactly evolutionary pressure operates on digital infrastructure will find historical narrative where they need mechanism.
Why it matters for product
Dyson's core claim — that complex behaviours emerge from distributed systems regardless of designer intent — is a direct challenge to the assumption that product architecture can be fully governed top-down. For a CPO, this reframes questions of platform strategy: the features, integrations, and usage patterns that accrete around a digital product are not simply the result of a roadmap but of evolutionary pressures the organisation may not be consciously managing. Recognising this shifts the design problem from 'what should we build' toward 'what selection environment are we creating', which has concrete implications for how metrics, API policies, and ecosystem incentives are set.