20 times the smartest people in the room predicted the future and got it completely wrong

There is a specific quality to the predictions that turned out to be catastrophically wrong: they were not made by fools. The IBM $IBM chairman who said there was a world market for perhaps five computers was one of the most successful technology executives of his era. The Decca executive who passed on the Beatles had signed and successfully developed other acts. The physicist who said heavier-than-air flying machines were impossible was Lord Kelvin, one of the most distinguished scientists of the 19th century. The strategist who said oil in Pennsylvania would never amount to anything was a banker whose analysis of the existing oil industry was entirely accurate.

The predictions failed not because the people who made them were unintelligent or uninformed, but because they reasoned from the present to the future in a specific way that consistently fails: they assumed that the world would change incrementally, that existing constraints were permanent, that the new thing would compete on the terms of the existing thing, and that the primary function of an innovation would be the function that was most obvious from the current vantage point. The telephone, in its early years, was predicted to be primarily a tool for broadcasting music and lectures to subscribers — a prediction that was logical from the perspective of what broadcasting infrastructure could do, and that completely missed the conversational, personal communication function that made the telephone transformative.

The failure of expert prediction is not random noise. It has a specific structure: experts tend to be correct about the direction of change and systematically wrong about the pace, the specific form, the primary use cases, and the social consequences. The predictions in this list are selected not primarily for their entertainment value — though several are genuinely funny — but for what they reveal about the specific cognitive patterns that make intelligent people confidently wrong.

Understanding those patterns is the primary value of this exercise. The goal is not to mock people who could not know what we now know, but to identify the specific ways that present-tense expertise becomes a trap for thinking about the future — so that the same traps are at least recognizable the next time a credentialed person makes a very confident prediction.

Joao Fernandes / Pexels

In 1876, Western Union was offered the patent for Alexander Graham Bell's telephone for $100,000. The company's internal evaluation concluded that the telephone had "too many shortcomings to be seriously considered as a means of communication" — specifically that it could only transmit over short distances, required a dedicated wire between two specific points, and had no clear business use case that the telegraph didn't already serve more reliably.

Every specific technical criticism was correct as of 1876. The telephone of that year was limited in range, point-to-point in its architecture, and lacked any obvious business application the telegraph didn't serve better. The memo's failure was not its technical analysis but its assumption that current limitations were permanent — and that the telephone's value would always be compared to the existing telegraph rather than to an entirely new communications infrastructure it would create itself.

Western Union's president, William Orton, reportedly asked: "What use could this company make of an electrical toy?" The telephone generated approximately $1 billion in revenue within 20 years of Western Union's rejection.

Inge Wallumrød / Pexels

Lord Kelvin — widely regarded as one of the most distinguished physicists of the 19th century, the man who formulated the second law of thermodynamics and determined the value of absolute zero — stated in 1895 that "heavier-than-air flying machines are impossible." He repeated the sentiment in 1902, one year before the Wright brothers' first flight at Kitty Hawk.

He was not alone: the physicist Simon Newcomb published a mathematical argument in 1903, the same year as the Wright brothers' flight, demonstrating that heavier-than-air flight was physically impossible. His paper appeared in an academic journal approximately two months before December 17, 1903.

Both men correctly reasoned that no 1890s steam engine could power a heavier-than-air aircraft — steam engines were too heavy for their power output. They were wrong that this constraint was permanent. The lightweight internal combustion engine that made flight possible was not yet in the analysis.

Ruben Boekeloo / Pexels

Thomas Watson Sr., the chairman of IBM $IBM, is reported to have said in 1943 that he thought there was a world market for "maybe five computers." The quote is disputed in its exact form, but the underlying sentiment was widely shared by computing pioneers of the era, and IBM's own business planning in the 1940s and early 1950s did not anticipate mass market demand.

The prediction was reasonable from 1943's perspective. Computers of that era were room-sized machines costing millions of dollars, requiring specialized operators and climate-controlled rooms, performing calculations for government agencies and large research institutions. The concept of a personal computer was not imaginable from within the framework of what "computing" meant then.

The specific cognitive failure was assuming that the form factor and cost structure of existing technology represented permanent constraints rather than early-stage engineering problems. The transistor, which would make small, cheap, reliable computing possible, was invented at Bell Labs four years after Watson's supposed remark.

Wolfgang Weiser / Pexels

Lewis Strauss, chairman of the United States Atomic Energy Commission, predicted in 1954 that nuclear power would make electricity "too cheap to meter" — a phrase that became one of the most famous examples of technological optimism in history. The nuclear industry subsequently struggled with cost overruns, safety regulations, and waste disposal challenges that made nuclear power among the most expensive electricity sources of the late 20th century.

The prediction was made at the peak of post-war techno-optimism, in the context of the "Atoms for Peace" program redirecting nuclear technology from military to civilian use. Strauss was describing a vision rather than an engineering projection, but it was taken as a serious technical prediction and cited as a standard against which nuclear power's actual economics were measured for decades.

The specific failure was assuming the fundamental physics of nuclear fission would translate linearly into cheap electricity, without accounting for the engineering, safety, regulatory, and waste management costs that the physics doesn't determine.

United Press International / Wikimedia Commons

On January 1, 1962, the Beatles auditioned for Decca Records in London. The label's A&R executive, Dick Rowe, passed on signing them, reportedly saying that "guitar groups are on the way out" and that "the Beatles have no future in show business." Decca signed Brian Poole and the Tremeloes instead.

Rowe's assessment wasn't entirely groundless: guitar-based skiffle and rock and roll were genuinely declining in commercial popularity in early 1962, having peaked in the late 1950s. The Beatles themselves were about to undergo significant development — the lineup, songwriting, and sound that would define Beatlemania were not yet fully formed at the Decca audition.

The prediction's failure was assuming that the trajectory of a genre determined the trajectory of any particular act within it, and failing to detect the specific qualities — songwriting, chemistry, charisma — that would make the Beatles not merely participants in an existing genre but creators of a new one.

Marshal Yung / Pexels

Darryl Zanuck, head of 20th Century Fox and one of the most powerful studio executives in Hollywood history, predicted in 1946 that television would fail as a mass medium within six months: "People will soon get tired of staring at a plywood box every night." He reportedly made similar predictions through the early 1950s as television viewership grew.

Zanuck understood the quality and production value of theatrical film and could see that television couldn't match it. He failed to model the possibility that convenience — watching entertainment at home without purchasing a ticket — was a value proposition that could successfully compete with higher production values.

Convenience versus quality is a competition that convenience has won repeatedly throughout media history. Zanuck's failure to model this was the failure.

SHVETS production / Pexels

Robert Metcalfe — the inventor of Ethernet and co-founder of 3Com — predicted in a 1995 InfoWorld column that the internet would "soon go spectacularly supernova and in 1996 catastrophically collapse." His specific concern was that the internet's infrastructure wasn't scaling adequately to meet the explosive growth in users.

Metcalfe promised to eat his words if the prediction was wrong. At the 1997 Sixth International World Wide Web Conference, he blended his printed column into a smoothie and drank it.

The prediction was not technically ignorant — the scaling concerns were real. The failure was assuming the engineering community would not solve the scaling problems in time. In fact, both usage and infrastructure grew rapidly, and infrastructure kept pace with demand.

Ben Tran / Pexels

Clifford Stoll, a respected astronomer and cybersecurity pioneer, published a Newsweek essay in 1995 arguing that the internet would not transform commerce, education, or daily life. His specific claims: "no online database will replace your daily newspaper," "no computer network will change the way government works," and that the internet's limitations as a commercial medium were fatal.

Stoll's technical understanding was genuine — he had been working with networked computers since the 1980s. His error was reasoning from the limitations of the 1995 internet — slow modem speeds, clunky interfaces, limited content, no secure payment infrastructure — to permanent conclusions about what the internet could become.

He was evaluating a technology in its infancy from the perspective of its infancy, without accounting for the pace at which the limiting factors would be resolved. He has since acknowledged this with good grace.

Wang Jared / Pexels

When Edwin Drake was attempting to raise investment for the first oil well in Titusville, Pennsylvania in 1859, he encountered near-universal skepticism from the financial community. One assessment from a New York banking firm advised investors: "Waste no more money on this foolish project."

The specific error was reasoning that the existing uses for petroleum — primarily as a lighting fuel competing with whale oil — represented the ceiling of its commercial value. The internal combustion engine hadn't been invented; the automobile was decades away; the petrochemical industry didn't exist.

The prediction's failure is the failure of reasoning about a resource's future value entirely from its present uses, without modeling the possibility that new uses might be discovered or created. Within 50 years, petroleum was the most strategically important commodity in the world.

Alesia Kozik / Pexels

Yale economist Irving Fisher — one of the most distinguished economists of the early 20th century, whose work on interest rates and monetary theory remains foundational — announced in October 1929 that stock prices had "reached what looks like a permanently high plateau." He made this prediction approximately three weeks before the Black Tuesday crash that marked the beginning of the Great Depression.

Fisher's prediction was based on genuine analysis: corporate earnings were strong, the economy was growing, and the Federal Reserve had maintained monetary stability. His failure was assuming the market's price level reflected underlying fundamental value rather than speculative excess.

Fisher lost most of his personal fortune in the crash and spent the rest of his career developing economic models to explain what had happened. His post-crash work on debt deflation became foundational — but the prediction is the one for which he is most remembered.

Pixabay / Pexels

Herman Kahn — the RAND Corporation strategist and widely considered one of the most serious thinkers on nuclear strategy in the Cold War era — predicted throughout the 1960s that nuclear war between the United States and the Soviet Union was highly likely, potentially inevitable. John von Neumann reportedly said in the 1950s that nuclear war was "absolutely inevitable."

The prediction failed because it modeled rational state actors without fully accounting for the specific leadership decisions, institutional constraints, and individual acts of restraint — Vasili Arkhipov's refusal to authorize a nuclear torpedo, Stanislav Petrov's decision not to report a false alarm — that prevented the scenarios rational-actor models suggested were likely.

The Cold War's peaceful resolution was not inevitable; it was the product of specific decisions by specific people in specific moments, which strategic models don't easily capture.

Abed Ismail / Pexels

Throughout the 1950s and 1960s, popular science magazines regularly predicted that personal jetpacks would be widely available by the 1980s or 2000. The predictions were illustrated with optimistic artwork showing commuters arriving at office buildings by air, landing on rooftops and in parking lots.

The specific engineering problem consistently missed was the fuel-to-flight-time ratio: practical jetpacks could sustain flight for approximately 20 to 30 seconds. The problem was not that jetpack technology failed to develop — it did — but that energy density requirements for sustained personal flight are genuinely difficult.

The predictions also completely missed the regulatory, safety, and urban planning challenges of widespread personal flight — airspace management, noise, collision risk — that would exist independently of the engineering challenges.

wal_ 172619 / Pexels

At the turn of the 20th century, numerous credentialed observers predicted the automobile would remain a luxury novelty and would never replace the horse. A 1903 editorial in The Literary Digest argued that "the horseless carriage...will never, of course, come into as common use as the bicycle." The president of the Michigan Savings Bank advised Henry Ford $F's lawyer not to invest in Ford Motor Company: "the horse is here to stay, but the automobile is only a novelty — a fad."

The specific error was reasoning from the economics and infrastructure of 1900 — roads built for horses, cities designed around horse-based transport, fuel availability limited to urban centers — to permanent conclusions about the automobile's viability.

The prediction failed on the same terms as most in this list: it correctly analyzed current constraints and incorrectly assumed those constraints were permanent rather than early-stage problems that would be systematically solved.

Viktorya Sergeeva / Pexels

IBM $IBM's market research in 1980 predicted that the total addressable market for personal computers was approximately 500,000 units globally — a figure that informed their decision to enter the personal computer market cautiously. Within five years, IBM had shipped millions of personal computers and the market was growing faster than their initial estimate by a factor of hundreds.

IBM's error was not in direction (they correctly identified that personal computing would be significant) but in magnitude. Their research extrapolated from the existing business computing market — asking business customers whether they would purchase a personal computer — without modeling the consumer market that would emerge, the home use cases that would develop, or how lower prices would create demand that didn't exist at higher price points.

The specific lesson: market research among current customers of an existing product systematically underestimates demand for a new product serving different customers for different purposes at different price points.

Zelch Csaba / Pexels

The success of the Apollo program generated widespread predictions that human colonization of Mars would follow within decades. Wernher von Braun published plans in 1969 for a Mars mission by 1982. NASA planning documents outlined a Mars landing by the mid-1980s.

None of these predictions adequately accounted for the political conditions that funded the Apollo program — specifically the Cold War space race with the Soviet Union, which provided the political will to sustain the enormous expense. Once the US had beaten the Soviet Union to the Moon, the geopolitical rationale for sustaining the spending required for Mars evaporated.

The predictions also underestimated the specific technical challenges of Mars — radiation exposure during transit, entry and landing on a planet with a thin but significant atmosphere, life support requirements for a multi-year mission — that make Mars significantly harder than the Moon in ways not fully characterized in the 1960s.

Min An / Pexels

The rise of ebook readers and digital content generated repeated predictions through the 2000s and 2010s that the printed book was dying. Nicholas Negroponte predicted in 2010 that physical books would be dead within five years. Publishers and booksellers experienced genuine disruption; Borders went bankrupt in 2011.

The prediction was not fulfilled. Physical book sales stabilized and partially recovered through the 2010s. By the mid-2020s, print books remained the dominant format for book sales in every major market, with ebooks plateauing at approximately 20 to 30% of the market rather than continuing to grow.

The failure was the straight-line extrapolation of a trend that turned out to be S-curved. The specific advantages of physical books — the reading experience, the absence of screen fatigue, the gifting function, freedom from battery and software updates — proved to be durable preferences for a significant proportion of readers.

Brian Ramirez / Pexels

The Arab Spring generated predictions that social media had permanently altered the power balance between authoritarian states and their citizens. Commentators argued that collective action enabled by social media could not be contained by authoritarian censorship.

The prediction was not entirely wrong — social media did enable significant political mobilization in specific circumstances — but it was wrong about permanence and direction. Authoritarian governments proved capable of adapting faster than the optimistic predictions anticipated: through surveillance of social media activity, prosecution of users for online speech, deployment of bot networks for disinformation, and internet shutdowns during political crises.

By the mid-2020s, social media's effects on political systems were assessed as more likely destabilizing for democracies — through disinformation, polarization, and amplification of extreme voices — than for authoritarian regimes, which had learned to exploit the same tools that had initially threatened them.

Pavel Danilyuk / Pexels

When COVID-19 forced mass adoption of remote work in March 2020, the prevailing expectation among managers, economists, and organizational researchers — based on pre-pandemic research showing remote workers were less productive on certain measures — was that productivity would fall significantly.

The productivity data that emerged was more mixed than the pessimistic prediction. Several large studies found knowledge worker productivity was maintained or improved during remote work; others found declines in specific collaborative or creative work. The most consistent finding was that remote work's effect on productivity varied enormously by role type, individual circumstance, and organizational culture.

The prediction's failure was treating "remote work" as a uniform intervention with a predictable average effect, when it was a heterogeneous set of conditions whose effects varied enormously by context.

Allen Boguslavsky / Pexels

Elon Musk predicted in 2015 that fully autonomous self-driving cars would be available to consumers within two years. He repeated similar predictions annually through the following decade, consistently placing "full self-driving" capability approximately one to two years away. As of 2026, truly autonomous vehicles capable of operating without human supervision in all conditions have not achieved the commercial deployment predicted for years earlier.

The prediction's failure reflects the "last mile" problem: a technology reaches 90% of a capability threshold relatively quickly and then spends an unexpectedly long time on the final 10%. Handling edge cases, unusual road conditions, and the long tail of scenarios that don't appear in training data has proved significantly more difficult than the first 90% of capability.

The prediction also underestimated regulatory timelines: autonomous vehicle deployment is not only a technical problem but a regulatory and liability problem, and the pace of regulatory framework development consistently undershot the pace of technical development.

Mikita Kavaleuski / Pexels

The flying car has been "10 years away" for approximately 80 years. Predictions of imminent commercial flying car availability appeared in the 1940s, the 1950s, the 1970s, the 1990s, and repeatedly through the 2010s, when the electric vertical takeoff and landing (eVTOL) aircraft category reinvigorated the prediction. As of 2026, commercial air taxi services exist in limited pilot deployments in a small number of cities but have not achieved the scale that any decade's predictions anticipated.

The flying car prediction is distinguished from other failed predictions in this list by its persistence: unlike predictions made once and then forgotten, the flying car prediction is remade by each new generation of engineers and entrepreneurs who believe that this time the technical and regulatory obstacles will fall. The prediction has survived being wrong for eight decades because each iteration of the technology is genuinely different from the previous one, even if the core constraint — the difficulty of making personal flight safe, quiet, cheap, and practical enough for mass adoption — has proven durable across all of them.