Why Science Fiction Can’t Predict the Future (And Why That’s a Good Thing) - Reactor

We love imagining the future. Much of the news is taken up with guessing what will happen next: weather, sports, the election, the stock market… And betting on the future is the engine that drives Wall Street as well as Las Vegas. Science fiction, often described as “stories about the future,” is prone to being judged by the degree to which it succeeds in predicting what is yet to be, and mainstream commentators often praise science fiction writers for being prescient, as though they are modern Delphic oracles or Philip K. Dick’s precogs come to life. 

This is all very ridiculous to people who enjoy science fiction, since the genre has an abysmal record of making accurate predictions. Just look around you. The year 2000 has long come and gone, and we don’t have killer robots with humanoid skeletons roaming a post-nuclear hellscape, nor do we have manned missions to Jupiter supervised by a sentient AI nostalgic for its childhood. (To be sure, we do live with chatbots capable of writing haikus about Bitcoin—but one may well consider that the actual dystopian scenario). We don’t have flying cars, despite them being regularly predicted for a century and becoming de rigueur in science fiction milieus¹.

To be sure, many science fiction authors explicitly reject the idea that their work should be read as predictive. Ursula K. Le Guin, for instance, writes, “Science fiction is not predictive; it is descriptive.”² However, enough authors embrace the prophetic role that “science fiction tells us our future” remains the defining view of the genre. Moreover, despite the lack of accurate, concrete predictions, science fiction continues to exert a powerful influence on how the general public thinks about new technology. Concepts like “Big Brother,” “virtual reality,” “metaverse,” “cyberspace,” and so forth have entered the popular lexicon. They are used to describe contemporary technologies, even though the real-world versions are nothing like their original, science-fictional depictions.

One example of this phenomenon is the ubiquitous, constant state of surveillance we now live under in the West, accomplished mainly by individuals making rational, voluntary choices over the last few decades to progressively trade away their privacy for convenience. There are indeed cameras and recording devices everywhere at all times, but our Big Brother is not the Orwellian hyper-totalitarian state; instead, it’s a many-tentacled monster cobbled together from technology companies, advertisers, governments (at all levels), “agentic” bots that seek to make us their agents, bad laws, internet mobs and trolls, loneliness, laziness, the false promise of genuine connection via social media, and above all, our imperfect selves. When we say Orwell is “prescient,” we are really praising his skill at crafting an evocative metaphor, not his predictive ability. 

Why is science fiction so bad at predicting the future, even as we continue to use science fiction terms to describe reality?

The best way to work through this paradox is through a concrete case study. In 1899, as part of preparations for the 1900 Paris Exposition, a group of French artists led by Jean-Marc Cotê created a series of images imagining life in the year 2000. Intended as cigarette box inserts or postcards, the images were not widely distributed and largely disappeared from public awareness until Isaac Asimov rediscovered a set and published them in a book in 1986, Futuredays: A Nineteenth-Century Vision of the Year 2000. The images draw heavily on the science-fictional imagination of Fin de siècle writers like Jules Verne and H.G. Wells and thus constitute a kind of visual distillation of 19th-century futurism.

The retro-futuristic scenes depicted are full of whimsy and steampunk charm: multi-limbed robot barbers groom satisfied patrons while a human operator fiddles with levers and buttons nearby; a pair of firemen equipped with angelic bat wings fly to the upper stories of a burning building to save a baby and its mother, who’s posing like a medieval Madonna; a bespectacled scientist armed with a huge syringe faces off against dangerous microbes, who, magnified by a microscope, now resemble clawed dragons; a girl carefully places eggs into one end of a dial-festooned, smoke-belching machine, while baby chicks emerge from a slide on the other side of the machine and head for a pile of feed, which instantly bulks them up into ready-to-eat broilers; schoolchildren sit in neat rows with beatific smiles while a set of wired helmets pump knowledge into their skulls, which is drawn from a nearby machine—cranked by a fellow pupil under the supervision of the schoolmaster—that crushes books into information much the way LLMs pulverize human knowledge into³ …

It’s easy to critique the collection’s blind spots regarding social and cultural evolution—race, class, gender roles, colonialism, and so on—and its confident assumption that the Paris of 1899 had reached the very pinnacle of human fashion and style. But look beyond these obvious faults, and you begin to see the anxieties that drove their imagination persist as potent specters long past the year 2000. 

Take, as an example, this house-cleaning robot. Seeing the image today, our first instinct is to shout “Roomba.” But a real Roomba—sleek, autonomous, stuffed full of sensors and machine learning—is nothing like the gangly device pictured here, which apparently requires a tethered human operator to supply all the smarts so the cleaning can be done. There’s no need to belabor the point; clearly, Cotê’s team missed the mark.

Or did they? The messy secret behind much modern gadgetry and “labor-saving” devices is how little they can actually do on their own. Even after two decades of nonstop improvements, robotic vacuums and mops aren’t much smarter today than they were at their inception. They get stuck on the tiniest of thresholds between rooms, entangle themselves in dangling charging cords, lock themselves in bathrooms, crawl into nooks from which they cannot extricate themselves, and somehow always run out of battery before they can find their way home. Not only do they take longer to clean a room than if I simply did it myself, but I have to spend the entire time they’re working on babysitting duty, either rescuing the robots or anxiously waiting to be summoned for a rescue. (You could get around this by reconfiguring your home so that it’s robot-safe, but surely we can agree that conforming your life to the needs of robots is not the future we signed up for?) So, in a manner of speaking, I really am still tethered to the robot vacuums, just like the woman in the picture. The artists may not have anticipated the challenge of constructing an autonomous machine that navigates a home, but they somehow nailed the false promise of home robotics⁴.

There is a consistent throughline here. The images often tapped into some form of social anxiety—the need for more food, faster transportation, freedom from the tedium of household tasks—but the proposed solutions, often based on the prevailing technology vocabulary of the time, turned out to be off the mark. (The images usually do resonate metaphorically, which is also important, but more on this later.)

But why are the proposed solutions so wrong? If we examine the history of technology, things seem to flow logically. Consider the history of passenger vehicles: constant improvements in the construction of carriages and the art of horse breeding reached a technological plateau until the advent of the steam engine, which was then supplanted by the superior efficiency of the internal combustion engine, which is now in the process of being displaced by electrical motors as a result of our desperate need to reduce carbon emissions. Every step of this history follows the previous one logically, so why can’t the evolution of technology in the future be predicted according to a set of ironclad laws, much like the psychohistory of Asimov’s Foundation series?

This is because history, insofar as it’s misunderstood as delineating a sequence of logical syllogisms that explain the sequence of what has occurred as inevitable, is nothing more than a story, a mere fiction. Reality is messy and complicated. There is no ineluctable flow of logic, no intelligible prime mover guiding the past to the future. The past is just the path we left behind as we stumbled through a wilderness of potentials, passing by worlds that could have been.

Let’s reexamine the history of cars. If you were to travel to America around the time when Cotê’s artists painted their predictions, you would have found that about 40% of the cars ran on steam, 38% on electricity, and 22% on gasoline⁵. If you were to ask an observer of the time to predict the future of automobiles in America, how would she answer?

If history were a matter of inescapable logic, then here is the evidence before the observer. First, consider the steam car. Steam engines, after decades of refinement, were very quiet and could produce a wide range of torque without a transmission, resulting in fewer components and easier construction (compared to, say, the internal combustion engine). In addition, since steam was an old technology, responsible for the Industrial Revolution and turning America into a global power per techno-determinist accounts, there were many skilled engineers familiar with steam, which meant steam cars were easy and cheap to maintain and repair. It was true that steam cars required time (as long as half an hour) for the boiler to reach the desired operating temperature, and water freezing in winter was an issue, but people were working on solutions for these problems. All in all, steam seemed a very good technology bet for the industry.

Second, consider the electric car. Electricity was the buzzy technology of the time, with electrification being synonymous with modernization and the future. Electric cars started instantly and were quiet, clean, and easy to operate (again, no transmission, no gearshift, etc.), making them especially popular with new drivers. Electric cars were also powerful, holding some of the earliest speed and hill-climbing records. Although electric vehicles were limited in range, this was not considered an insurmountable problem. After all, Thomas Edison, the most well-known and trusted innovator of the time, was actively working on improving battery technology for electric cars. If you had been asked to invest your money in the growing automobile industry, electric vehicles definitely would have seemed a very good bet as well.

In last place, technologically speaking, was the gas car. This type of car seemed plagued with problems. Hand-cranking, required to start the engine, could injure the inexperienced. The need for a transmission not only made construction difficult but also increased operational complexity (and there was no standardization of interfaces among different gas car makers). The internal combustion engine, powered by a series of small explosions, was loud, dirty, and prone to stalling. Against these drawbacks, the only definite advantages of the gas car were a relatively long range (though even this was questionable, as early gas cars had similar ranges to electric vehicles) and the ease with which the exhausted power source could be “refreshed” by refilling the gas tank (as opposed to the slow process of recharging a battery). It was hard to see how this troublesome technology could beat out trusted steam and the new darling, electricity.

We all know what happened next—the “reasonable” predictions and extrapolations that picked electric or steam cars turned out to be science fiction. It’s fascinating to read accounts of why history turned out the way it did. Literally hundreds of different reasons have been proposed for the eventual triumph of the gas car: the discovery of cheap oil in Texas, the failed business model of renting electric vehicles to customers that ruined the largest EV maker, the American cultural preference for long drives and what we would now call “glamping,” the invention of the electric starter, the dogged determination of Henry Ford, the low quality of early roads, the slower-than-expected rate of urban electrification, the faster-than-expected rate of innovation among gas car makers … Every single account is plausible and justified by evidence; every explanation tells a perfectly logical story. 

This is always the pattern in technology, whether we’re talking about advances in sailing ships, automobiles, touch screen interfaces, home automation, the global food supply, or artificial intelligence. Confronted by a challenging problem, thousands of innovators hack away at it, trying out dozens of different solutions. All the solutions have advantages and disadvantages, and plausible paths to victory could be charted for each. 

A sequence of random events follows: new resources are discovered, a war breaks out, a tremendous new mind decides to work on the problem, a different political party comes into power, a company goes bankrupt, a technology wins out in a seemingly unrelated field, everyone suddenly decides one thing is cool while another is not… Somehow, the random sequence results in a breakthrough, where one particular solution achieves an insurmountable lead over competitors. People will never agree on how this happened; countless stories can be told, all incorporating some of the factors and leaving out others. All that matters is that suddenly, all the attention, resources, “oxygen” go to the winner. Winning begets more winning, leading to faster improvement, more innovation, higher profits, and lower costs, until all other potential solutions are forgotten and the winner seems obvious. But that is only because we have mistaken the consequences of winning for its causes.

Chance and fortune bounced our world this way and that, and humans strove and squabbled and lied and fought until we got to where we are. To justify everything, to make sense of the random walk that took us here, we must craft character arcs, identify heroes and villains, and fashion the world as it is into the conclusion of a morality play. We are here because this led to that which turned to this and then that—it all makes sense, right? So the winners must have always meant to be winners, and the losers must have always meant to be losers.

The prospective view, in that moment before the breakthrough, when all the potential solutions are vying for attention, is completely different from the retrospective view, long after the breakthrough technology has transformed the world and secured its own triumphalist narrative. Survivorship bias, confirmation bias, selection bias, hindsight, narrative fallacy, wishful thinking, arrogance… there are countless names for the cognitive biases humans exhibit when we try to tell the story of the past from our place in the present, and we must constantly remind ourselves that the way it is is not the way it has to be.

Finally, we have an answer to why science fiction is so bad at predicting solutions. Science fiction authors imagine the future based on present dreams and anxieties. But they’re confined to their moment, which lies before the breakthrough that will usher in the future. Even those who consider themselves rigorous futurists can only see a field of competing potential solutions to the anxieties of their age. No matter how much care they put into researching the strengths and weaknesses of each solution, they cannot know the random sequence of events that will ultimately decide the winner. All they can do is guess, pick a winner (or make up one), and then reimagine the whole world to justify that choice. Plausibility is then secured through a web woven from character arcs, heroes and villains, and a satisfying sense of moral resolution. Science fiction authors craft stories, and stories always turn effects into causes.

The future is not plausible (and the present scarcely more so; how many of us could have believed—before given incontrovertible evidence—the cruelty we are capable of to realize the dream of fast poultry in our industrial farms?). That is why science fiction can never predict the future. However, by crafting entertaining stories, authors invent powerful metaphors that shape how we imagine our technological future and understand our technological reality. These metaphors are why science fiction matters. Science fiction is, as many have long recognized, a myth-making literature, best understood as a young province in the ancient empire of fantasy. Like all fantasy, it draws from the collective unconscious, giving substance to beings that have no weight, voicing thoughts that cannot be put into words, depicting forms that cast no shadow. However, unlike other, older forms of fantasy, science fiction draws its metaphors from our scientific understanding of the universe. Thinking machines that interrogate their makers; immortality achieved by rewriting the code of life; new cities and ways of life on distant planets; strangers who don’t share our evolutionary history; yes, even immaculate houses with autonomous mops and instant poultry emerging from a thumping engine—these are all metaphors that allow us to make sense of a world in which the products of our imagination and craft, technology and invention, increasingly dominate not just our own evolutionary future, but the future of the planet as a whole. We live in a world in which the possibility field is growing ever grander, and new myths are needed to make sense of it.

Ursula K. Le Guin describes the myth-making function of science fiction in her 1976 essay, “Myth and Archetype in Science Fiction”:

On this level, science fiction deserves the title of a modern mythology⁶.

Read as prediction, Frankenstein is a failure. But prophecy was never the point. The power of myths does not depend on oracles.

A more recent example of science fictional mythmaking may be found in Silvia Park’s Luminous. In this near-future world of ubiquitous robotics and extensive body modification, the line between human and machine has blurred to the point of nonexistence. Society, long used to the idea of using robots to perform emotional labor and caretaking tasks, now considers robot children as a suitable replacement for (or even improvement on) human children. As we contemplate, delight in, or are revolted by the implications of this change on the role of children and the meaning of parenthood, we’re in fact engaging with another variation of the archetype of the Abdicating Parent. It matters little whether the scenario in Luminous will come to pass; just like the Titans and the Olympians, Milton’s God and Satan, Dr. Frankenstein and his Creature, Park’s human parents and robot children give us a new way to understand and talk about the fraught relationship between mortal beings who seek immortality via generation. Park, I suspect, cares no more that future parents will actually pick out robot children than I care that future artists will really use AI to help their audience to dream together—it is the archetype, the dream, the myth that matters, that will last.

In the fullness of time, when we look backward from the other side of technological revolutions, we can see that science fiction authors got all their predictions wrong. They failed to imagine all the intervening cultural shifts, inventions, economic twists and turns, war and peace, politics, the actions of little birds and the dreams of leviathans. We laugh at the gap between their vision and reality. However, in that wrongness, we also find solace and hope. So much of science fiction is dystopian because extrapolating anything to its logical end is terrifying and also because stories that scare us are more entertaining and serve as warnings of worlds we do not want. But science fiction is not future history. There are no laws dictating the future from the present. We make the future, and it won’t be dystopian unless we build it that way.

And long after the predictions have been forgotten, the metaphors the Promethean authors invented to make sense of a changing world remain true and full of power. Frankenstein’s monster endures, as do instant poultry and helpful machines that are helpless and needful. These modern myths become part of our vocabulary, the framework and tools with which we make sense of the impossible present and then construct the unimaginable future.

1. We can argue about what was the first example of a ‘flying car’ (in the sense of an everyday vehicle meant for the average commuter). My pick is Hugo Gernsback’s Helicar in 1923. ↩︎
2. Le Guin, Ursula K. Introduction to The Left Hand of Darkness, i-iv. New York: Ace Books, 1976. ↩︎
3.  Given the lighthearted execution in some of the images, one could argue that the intended audience wasn’t supposed to take these “predictions” any more seriously than we take the “predictions” in Futurama. Still, the amount of detail put into the inventions is more than necessary for a joke, and there is a compelling sense of sincerity in all the scenes. These were not just absurd scenarios devised for laughs, but entertaining attempts to imagine future solutions for real problems. ↩︎
4. Recent “advances” such as humanoid household robots that are, in fact, just shells operated via telepresence by exploited workers add another dark layer to this drama of lies, deceit, and failed promises. ↩︎
5.  George C. Cromer, Orville C. Cromer, Christopher G. Foster, and Ken W. Purdy. “automobile.” Encyclopedia Britannica, November 4, 2023. https://www.britannica.com/technology/automobile. ↩︎
6. Ursula K. Le Guin, “Myth and Archetype in Science Fiction,” Parabola, Vol. I, No. 4, (Fall 1976). ↩︎

Buy the Book

All That We See or Seem

Ken Liu

Book 1 in the Julia Z scifi thriller series

Buy Book

All That We See or Seem

Ken Liu

Book 1 in the Julia Z scifi thriller series

Book 1 in the Julia Z scifi thriller series

Buy this book from:

• Amazon
• Barnes and Noble
• iBooks
• IndieBound
• Target