Technological Variation

In the AI Story

Basalla identified multiple sources of technological variation. The craftsman's tinkering — the small modifications that accumulate through daily practice. The engineer's systematic experimentation — the deliberate search through possibility space guided by theoretical understanding. The accidental discovery — the unexpected result that reveals a new direction. The transfer of a technique from one domain to another — the cross-pollination that produces novel combinations. And what Basalla called fantasy invention — the imagined artifact that exists in fiction, mythology, or speculative thought long before the technical means to build it materialize.

The constraint on variation is what distinguishes Basalla's framework from the heroic-inventor mythology. The popular narrative treats the inventor as a figure of almost unlimited creative freedom, capable of producing novelty from the resources of individual genius. Basalla's framework treats the inventor as an explorer of a constrained landscape — a landscape whose contours are determined by the prior state of the art. The inventor's creativity is real, but it operates on materials that already exist, in directions that the landscape has already marked as possible.

The phenomenon of simultaneous invention provides the strongest empirical support for the constrained-variation view. When Alexander Graham Bell and Elisha Gray filed telephone patents on the same day in 1876, they were not experiencing cosmic coincidence. They were operating within the same variation landscape, and the landscape had reached a point where the next development was overdetermined. Darwin and Wallace arriving independently at natural selection. Newton and Leibniz developing calculus simultaneously. The pattern is not exceptional. It is structural. It is what happens when the variation landscape constrains the possible directions of novelty tightly enough that multiple independent explorers converge on the same territory.

Applied to artificial intelligence, the constrained-variation view dissolves the mythology of the founder-genius. The transformer architecture was developed by a team at Google in 2017, but the attention mechanism it built on had been circulating in the research community for years. Large language models emerged almost simultaneously at multiple labs because the variation landscape had reached the point where they were overdetermined by the accumulated state of the art in neural network research, computational infrastructure, and dataset availability. The specific founders who captured the narrative advantage are contingent. The emergence of the technology itself was, by 2022, effectively inevitable given the prior state of the landscape.

Origin

The concept is developed most fully in chapters 3 and 4 of The Evolution of Technology. Basalla drew on the sociology of invention, particularly the work of William Ogburn on simultaneous invention, and on the history of specific technological lineages — the steam engine, the automobile, the airplane — where the pattern of constrained variation is most clearly documented.

Key Ideas

Intentionality operates within constraint. Human choice drives variation, but the choice is constrained by what the variation landscape makes available.

Multiple sources produce variation. Tinkering, systematic experimentation, accidental discovery, cross-domain transfer, and fantasy invention all contribute to the variation pool.

Simultaneous invention is structural. When multiple minds converge on the same innovation, the convergence reveals the constraints of the variation landscape rather than cosmic coincidence.

The inventor is an explorer, not a creator. The variation landscape is the territory; the inventor is the traveler who happens to reach a particular clearing first.

Fantasy shapes direction. The imagined artifacts a culture accumulates mark the directions its inventors will attempt to explore.