Evolution is an extraordinary process. It is difficult to understate its role in creating the diversity of life on Earth. But the study of this process has forced scientists to conclude that evolution is not an exclusively biological phenomenon. Indeed, biology is just a special case.
Instead, evolution is a general process that plays a role in any system in which there is reproduction, variation, fitness testing, and iteration over many generations. The process of evolution can easily be reproduced in silico, leading to artificial life and to evolutionary algorithms that can solve a wide variety of problems.
Computer models have also captured the behavior of evolution and allowed researchers to predict its future, such as the diversity it creates. These models are powerful microscopes for studying and understanding evolution in the real world.
But while researchers have long studied the role of evolution in biology and computer scientists have long studied evolution in silico, social scientists and anthropologists have yet to get to grips with the role evolution in technological development. This is the way that cultural objects evolve over time, things like stone tools, metal weapons, and more modern objects such as cameras, computers, televisions, and so on.
The problem is that nobody agrees on how to measure change in these systems in which there is no obvious analogy with the familiar ideas of genetics and sexual reproduction. Indeed, various attempts to describe technological evolution have become bogged down in ways to describe diversity—how can you objectively categorize the differences between one generation of televisions and the next? All that means there is little understanding of the way technologies evolve.
Today, that looks set to change thanks to the work of Erik Gjesfjeld at the University of California, Los Angeles, and a few pals, who have found a way to analyze the evolution of American automobiles from their invention in the 19th century to the present day. Their method provides unprecedented insight into the forces at work in automobile evolution. And they say it can easily be applied to other technologies.
Their method gets around many of the standard problems in assessing technological evolution. One of the most difficult is measuring changes in diversity over time. Many researchers think of diversity as a property based on the variety of objects or organisms, the balance between them, and their disparity.
But measuring these properties in the technological world is difficult. Understanding the balance between technologies, for example, requires a detailed understanding of the abundance of technological products over time. And any calculation of disparity requires knowledge of product characteristics and how product classes can be distinguished.
None of that information is easy to come by, particularly for complex products that are based on many different lines of technological development. Automobiles, for example, are the result of advances in manufacturing technologies, design technologies, engine design, technologies associated with fuel, lubrication, aerodynamics, ergonomics, computing, and so on. Getting an overview of all these is an impossible task.
So Gjesfjeld and co have come up with an alternative approach. They ignore all the messy details about exactly what technologies—and their balance and disparity—contribute to each generation of vehicle. Instead, they count only the rate at which new models appear and die out, treating each model like a new species.
Then they assume that evolution is responsible for these changes and use computer analysis to fit the data to a particular, well understood model of evolution. Having fixed the parameters in this model, they can then use it to infer the environmental forces that caused car models to change and to make predictions about the future evolution of automobiles.
The details make for interesting reading. The team’s data comes from EBay Motors, which has an extensive database of the make, model, and production year of cars and trucks produced in the United States between 1896 and 2014. In total, the team gathered data on 3,575 different car models made by 172 unique manufacturers. This is the “fossil record.”
They note in particular the first and last year of production for each model. This is the date of the origin and extinction of each species. Plotting this data produces a curve showing the rate at which new species originate and die out over time.
The researchers’ key breakthrough, however, is the way they analyze this curve using a process called a birth-death Markov chain Monte Carlo algorithm. This simulates the process of evolution among individuals to generate a curve showing the rate of origin and extinction of species.
The trick, of course, is to generate a curve that matches the history of automobile production. And the team achieves this with the model running through 10 million generations of individuals. Having done this, they can see the major environmental changes that must have caused the automobile evolution to follow the same curve.
An interesting question is to ask what the simulated environmental changes correspond to in the real world. Gjesfjeld and co say their analysis suggest there have been two major changes in the rate at which new automobiles models appear, one in 1933 and another in 1984: “1933, corresponding to the Great Depression, and 1984, coincident with the reengineering of cars to meet fuel-efficiency standards,” they say.
Equally, they have identified two shifts in the extinction rate: “1935, once again corresponding to the Great Depression, and 1960, which marks the height of the ‘Big Three’ dominance in the U.S. automotive market,” they say.
The data also reveals a curious feature of automobile development since the 1980s. It turns out that since then, the extinction rate of American car models has been higher than the origination rate. “This indicates that over the last 30 years more American car models have been lost per year than gained,” they say.
At the same time, diversity has also decreased. “This means that although fewer new car models have been introduced toward the present, the life span of these models has tended to increase,” they say.
That provides important evidence in favor of certain theories of technological evolution. “Our research is consistent with the idea that once ‘dominant’ car models are established after World War II, short-lived “experimental” models become less common due to increased costs of producing these models relative to established designs, leading to an overall drop in net diversification,” they say.
But that’s only for gas-powered cars. In recent years, the automobile industry has witnessed the emergence of hybrid vehicles and electric cars, and these are following a different pattern of evolution. “We predict that electric and hybrid cars may be experiencing the early stages of a radiation event, with dramatic diversification expected in the next three to four decades,” they say.
That’s interesting work that has significant potential. By focusing only on the dates that new models originate and die out, Gjesfjeld and co can create a detailed understanding of their evolution and the forces that have influenced it.
And since this information is relatively easy to gather for other modern technologies, it should only be a matter of time before we see more comprehensive treatments of other technologies such as mobile phones, cameras, microwaves, and perhaps textiles, furniture, and so on.
All that can be pieced to together to provide a picture of cultural evolution the likes of which anthropologists have never had. And if that doesn’t herald the emergence of a golden era for social science, what does?
Ref: arxiv.org/abs/1604.00055 : Competition and Extinction Explain the Evolution of Diversity in American Automobiles
Geoffrey Hinton tells us why he’s now scared of the tech he helped build
“I have suddenly switched my views on whether these things are going to be more intelligent than us.”
ChatGPT is going to change education, not destroy it
The narrative around cheating students doesn’t tell the whole story. Meet the teachers who think generative AI could actually make learning better.
Meet the people who use Notion to plan their whole lives
The workplace tool’s appeal extends far beyond organizing work projects. Many users find it’s just as useful for managing their free time.
Learning to code isn’t enough
Historically, learn-to-code efforts have provided opportunities for the few, but new efforts are aiming to be inclusive.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.