Unnatural Selection

By Tom Standage, 28 July 1998

EVERYBODY KNOWS that predicting the future is a difficult and dangerous business: witness IBM chairman Thomas Watson's complete failure to foresee the computer revolution with his infamous 1943 suggestion that there was a world market for “about five” computers. Explaining the past ought, in comparison, to be relatively easy; after all, we know what happened. So why is there no scientific theory of human history?

As anyone who has read Guns, Germs, and Steel will recall, Jared Diamond poses this very question in his conclusion of his Pulitzer-Prize-winning book. “The histories of dinosaurs, nebulas and glaciers are generally acknowledged to belong to fields of science rather than to the humanities,” he writes. “But introspection gives us far more insight into the ways of other humans than into those of dinosaurs. I am thus optimistic that historical studies of human societies can be pursued as scientifically as studies of dinosaurs—and with profit to our own society today, by teaching us what shaped the modern world, and what might shape the future.”

But while a Grand Unified Theory of history is evidently still a long way off, when it comes to theorizing about the history of one of the strongest forces shaping the modern world—technology—several authors, including Diamond, have recently had a stab at it. Each of them borrows theory and terminology from different areas of science in attempt to explain the lurching, unpredictable nature of technological history. And yet, in their efforts to explain why things are the way they are, have these would-be historical scientists overlooked an obvious scientific cupboard to plunder? If you ask me, the most elegant way to explain technology history is already right under our noses—and on our walls, and our desktop patterns. I say technology history is fractal. The question is, how do all these rival theories stack up?

AS DIAMOND'S FANS WILL KNOW, his “geographical” theory of history looks at a number of elements—the availability of domesticable plant and animals, and the emergence of epidemic diseases, complex societies and new technologies—to explain how geographical factors, as opposed to inherent racial supremacy, led to the technological and political supremacy of Western European countries. When it comes to explaining the history of technology in particular, Diamond kicks off by deriding the now-exploded ”heroic” theory of invention, which states that breakthroughs are made by a gifted handful of genius inventors in response to social necessity (what might be termed a non-scientific “biographical” theory of history that was popular in the nineteenth century).

Instead, says Diamond, invention occurs spontaneously in all parts of the world—but the circumstances of the inventors, which are themselves largely the result of geography, determine what materials are available, and whether the invention is actually adopted or not. For example, wheeled vehicles never arose in the Americas except as toys—not because nobody ever thought of inventing them, but because there were no indigenous domesticable animals suitable for pulling them. Similarly, the relative abundance of domesticable crops within Eurasia encouraged the adoption of farming, and the establishment of complex settled societies capable of supporting specialist craftsmen. The result? Specialist technologies like guns and armor.

Diamond also lobs in a dollop of terminology from chemistry: technological advances, he says, are an example of an autocatalytic process, as the combination of simple technologies leads to more complex inventions. But his theory only goes so far: it explains why isolated peoples, left to their own devices, ended up with this or that technology, and how superior technology could give one group of people an advantage over another. He stops short of postulating Darwinian competition between technologies themselves: he is concerned only with competition between users of technology.

However, another recent book, The Soft Edge by Paul Levinson, takes this extra step by proposing a “natural history” of technology, borrowing from evolutionary theory rather than geography and chemistry. Levinson's approach draws on evolutionary epistemology, which draws analogies between the evolution of biological organisms and the evolution of human knowledge. In an earlier academic work, Human Replay, he advanced a Darwinian theory of media evolution, and his latest book extends this approach to information technology.

“We should not be surprised,” he writes, “to find in the history of information technology an evolutionary dynamic in many respects very much like that of the literally natural, organic world.” But hang on, you might say, aren’t new technologies the result of conscious invention by humans, rather than the blind, aimless march of natural evolution? No, says Levinson: the analogy with evolution is valid, because so many technologies end up being used in ways their inventors had not imagined. Thomas Edison, for instance, expected his newly-invented phonograph to be used primarily to record telephone conversations, not music. The difference, says Levinson, is that the environment that determines which inventions are ”fit”—and thus which ones survive—is an audience of sentient humans; so information technologies evolve in ways that replicate natural patterns of human communication.

Levinson uses this “anthropotropic” theory of technological evolution to explain, for example, why silent movies were replaced by the talkies, but radio survived the introduction of television. A technology that facilitated hearing without seeing, he says, was perfectly acceptable; listening to the radio is like talking in a darkened room. In comparison, seeing without hearing was unnatural, so silent movies stood little chance in competition with non-silent ones. Despite many such examples, Levinson's rather scattershot approach, with occasional forays into lit crit speak, is rather less convincing that Diamond's carefully reasoned argument.

In another new book, Media, History, and Technology, Brian Winston of the University of Westminster offers yet another theory, this time raiding a different cupboard altogether: linguistics. Like Levinson, Winston sets out to explain why some technologies succeed, while others are ignored. The fax machine, he points out, was originally invented in 1847, and the idea of television was patented in 1884. Winston postulates a “law of suppression of radical potential” that acts as a brake on the adoption of new technologies, slowing their progress until society is capable of assimilating them. The framework he uses to advance this theory is, surprisingly enough, drawn from the field of Saussurian linguistics.

Essentially, just as Saussure postulated that speech is “the surface expression of a deep-seated mental competence”, Winston suggests that technologies are expressions of underlying scientific competence. Using diagrams appropriated from Saussurian theory, he builds a six-step model in which technologies start out by going from ideation (where the possibility of a new technology is first identified) to prototypes (early versions of an invention that fail to get anywhere). After a while a “supervening social necessity” arises, and the creator of one of the prototypes is deemed to be the “inventor”. The law of suppression then kicks in, delaying the adoption of the technology, often for decades. The final step is diffusion, as the technology is widely adopted.

Winston starts off by fitting the invention of the electric telegraph into this model: the idea of sending messages by electricity had been around since the seventeenth century, and dozens of prototypes were created, but it was only with the appearance of a supervening social necessity—rapid signaling equipment for use on the railways, and later by the newspaper industry—that the electric telegraph was deemed to have been “invented” in the 1830s. An ensuing patent war, and a debate over whether the telegraphs should be publicly or privately controlled, delayed the widespread introduction of the telegraph for a few years, after which it took off. This six-step pattern is then applied (with varying degrees of convincingness) to the telephone, radio, television, computers, and wired and wireless networks. Unusually, Winston finishes off with a prediction: that holographic technology is currently stuck in the “suppression” phase, but that some day soon, holographic TV is going to be huge.

At any rate, it seems that bits of geography, chemistry, evolutionary theory, and linguistics can all be used to frame certain aspects of the history of technology. But while we're in the business of raiding other disciplines for models to explain history, haven’t we overlooked a rather obvious candidate?

GIVEN THEIR ENTHUSIASM for modeling everything from earthquakes to network traffic to river formation, it is amazing that the advocates of complexity theory—the related fields of chaos, self-organizing systems, and fractals—have not jumped at the chance to apply it to history. Because history, and the history of technology in particular, looks an awful lot like a fractal.

Benoit Mandelbrot, who coined the term fractal, defined it as a fragmented shape that can be subdivided in parts, each of which is approximately a reduced-size copy of the whole. This is also known as “self-similarity”. Fractals occur widely in nature, from snowflakes to coastlines to clouds to mountains to the branches of trees. Their fractal structure means that, for example, a twig looks like a miniature tree, and the edges of a rock pool just like a coastline viewed from the air. By inspection, the history of technology is self-similar across time, as successive generations apply new technologies to age-old problems (such as communicating at a distance, or reproducing pictures), and inventions are forgotten and rediscovered, (like the fax machine was). It is also self-similar in space, as things are invented simultaneously in different countries (such as the electric telegraph), and rival inventors and their inventions co-exist in the marketplace (competing brands of car, for instance).

Different technologies go through the same life cycle of invention, innovation, competition and diffusion, but on different scales and at different times, neatly conforming with Mandelbrot's definition. The Internet of today, for example, has many similarities with the electric telegraph of the nineteenth century: the same uses, but at different times.Battles over standards are sometimes fought by individual inventors (as in the case of the phonograph, for example), and sometimes by multinational companies (as in the current Web browser war): the same battle, but at different scales. And at the root of all these similarities is the unchanging nature of human curiosity and the desire to apply new discoveries to the same old problems: simple rules that lead to a complex pattern, another characteristic of a fractal.

Technology history also demonstrates so-called power law behavior, in which the occurrence of events is inversely proportional to their significance. The distribution of earthquakes, for example, follows a power law: there are a very small number of very large quakes, but a large number of small quakes. The shape of a tree is another example: it has a few large branches, but many small twigs. The same behavior is true of technologies: most inventions are relatively insignificant, but a handful are revolutionary. No doubt mathematicians will be aghast at my cavalier use of their terminology, but it's no worse than the way jargon from evolutionary theory or chemistry has been brutally pressed into service by historians. If it can be put on a proper footing (something I don't pretend to be qualified to do), a fractal theory of history would explain why history is so complicated and unpredictable. It would also neatly provide a theoretical basis for old truisms like “history repeats itself” and “there is nothing new under the sun”. There you have it, Mr. Diamond: a theory of human history that encompasses every buzzword going, and even proves the validity of folk wisdom. What more could you ask for?

Paul Stone
pas@MNSi.net
London, ON