In February 1967, Star Trek aired an episode called “Court Martial.” Kirk is on trial. The ship’s computer has produced a visual record showing that he ejected a pod during a yellow alert and killed a crewman. The record is clear. Kirk’s own memory contradicts it. The court believes the machine. While the trial proceeds on the starbase below, Spock sits in a briefing room playing three dimensional chess against the ship’s computer. McCoy walks in and calls him cold blooded for playing games while the captain’s career burns down. Spock replies that he has just won four games in a row. McCoy stops cold. That is impossible. Spock programmed the chess computer himself, using his own knowledge of the game. The most he should ever be able to achieve against it is a draw. In the episode Spock puts it plainly: the computer cannot make an error, and assuming he does not either, the outcome should be stalemate after stalemate. That is the clue that breaks the case. Someone tampered with the computer. The evidence against Kirk was fabricated. It is a well built scene. It also rests on a premise the writers considered too obvious to defend. In the 23rd century, aboard a starship crossing light years, a person beating a machine at chess is proof that the machine has been sabotaged. Thirty years later, on May 11, 1997, Garry Kasparov resigned after nineteen moves in the sixth game of his rematch against IBM’s Deep Blue. The match went to the machine, 3.5 to 2.5. It was the first time a computer had taken a full match from a reigning world champion under tournament conditions. Kasparov said afterward that he had lost his fighting spirit. The Enterprise’s chess computer was beaten by a person. IBM’s chess computer beat the reigning world champion. Star Trek’s 23rd century was overtaken in 1997, and the show never saw it coming. The shape of the error I wrote a while back about science fiction as a forecasting instrument, and about how often it lands specific inventions decades ahead of the engineering. Star Trek is the standard example, and it deserves the reputation. The show has a real hit rate on hardware. What interests me now is the shape of its misses. Forecasting errors are usually treated as noise. Somebody guessed, somebody was wrong, move on. But Star Trek’s errors are not scattered randomly. They point in one direction, consistently, across three decades and four series. The show was generous with physics. It handed out faster than light travel, matter transport, artificial gravity, and energy shields without much hesitation. It was stingy with computation. Every time the plot required a judgment call, a hunch, a moment of insight, or a game that mattered, the machine came up short and a person filled the gap. Sixty years later, we have none of the physics and considerably more than the computation. We cannot transport a paperclip across a room. We can hand a stranger a device that translates their speech into forty languages, writes a legal brief, and identifies a skin lesion. That inversion is the useful part. If you want to get better at reading the future, the interesting question is not which predictions were wrong. It is why the wrong ones cluster where they do. The hits, briefly Give the show its due first, because the record is genuinely good. The communicator. Kirk flips open a palm sized device and talks to a ship in orbit. The Motorola StarTAC, the first mass market flip phone, arrived in 1996 with the same gesture built into it. Martin Cooper, who made the first handheld mobile call in 1973, spent decades being credited with taking the idea from Star Trek, and for years he accepted the credit. He later corrected the record. The real influence was Dick Tracy’s wrist radio, which he read as a boy, and Motorola had been working on handheld cellular since the late 1950s. He has called the Star Trek attribution one of his own mistakes. The myth is now better known than the correction, which is its own lesson about how influence stories get written. The PADD. Officers walk around carrying flat glass slabs they read reports on. That one landed almost exactly, roughly two decades early. The talking computer. “Computer, locate Commander Riker.” Jeff Bezos has said the long term vision for Alexa was to become the Star Trek computer, something you could ask anything and converse with naturally. He grew up playing Star Trek in Houston, where one kid always had to play the computer. Amazon later added “Computer” as a selectable wake word on Echo devices. The reference is not subtle. The universal translator. Live speech to speech translation now runs on a phone. This one arrived early and quietly, and almost nobody treats it as science fiction anymore. The medical tricorder. In 2012 the Qualcomm Foundation and XPRIZE launched a competition named after the device, asking teams to build a portable unit that could diagnose thirteen conditions and monitor five vital signs without a clinician. Three hundred teams entered. In April 2017 the top prize went to Final Frontier Medical Devices, a family led team out of Pennsylvania headed by an emergency physician and his brother, for a system called DxtER. The prize existed because the fictional device was specific enough to be an engineering target. The replicator. The partial credit case. Star Trek imagined a wall unit that assembles physical objects from stored patterns, including hot food and Earl Grey tea. We have additive manufacturing that prints titanium brackets, dental aligners, houses, and, in laboratory settings, food. The direction is right and the gap is enormous. The show also treated replicator technology as a diplomatic instrument, handed to civilizations that needed it, which is a reasonable read on how manufacturing capability actually moves between economies. The rest. Video calls to Starfleet Command. Wall sized displays. The earpiece. Biobeds showing vitals continuously above a patient. Doors that open when you approach them, which the show pulled off in 1966 with a stagehand yanking a rope offscreen. That is a strong record for a series with a rope operated door. Where it went the other way Now the misses. They are more interesting. Chess, and every game after it Start where we started. “Court Martial” treats machine chess supremacy as impossible, and that impossibility is load bearing for the plot. The underlying model is that the computer is a recording of Spock. It holds his knowledge of the game, played back without error, and it cannot exceed him because there is nothing in it that did not come from him. A filing cabinet with a voice. Reasonable for 1967. Every decade since has dismantled it. Strategema Twenty two years later, in July 1989, The Next Generation aired “Peak Performance.” A Zakdorn master strategist named Kolrami comes aboard. Riker challenges him at Strategema and loses in about twenty three moves. Then Data plays him. Data is an android with a positronic brain, the only one of his kind in Starfleet, and Kolrami takes him apart in a matter of seconds. Data has something close to a breakdown over it and questions whether he can be trusted at his post. Eventually he plays Kolrami again with a different premise. He plays to prevent a conclusion, declining every opening and simply blocking. The game runs past thirty five thousand moves each. Kolrami throws down his controls and storms out. Data allows that he did not technically win, though he did, in his phrase, bust him up. Read that back with 2026 eyes. An artificial mind cannot beat an organic strategist at a board game, and its workaround is to be more patient. In 1989 that was a satisfying resolution about the limits of computation. In 2016 AlphaGo beat Lee Sedol at Go, a game whose branching factor was supposed to make it the wall that brute force could not climb. The following generation of the system learned it from scratch, with no human games as input, and surpassed every prior version. The premise of “Peak Performance” did not merely age badly. It reversed. Velocity In May 1998, the Voyager finale “Hope and Fear” opens with Janeway and Seven of Nine playing Velocity on the holodeck. Janeway wins six rounds out of ten. Seven is frustrated. She has Borg enhanced visual acuity and stamina and should, by her own accounting, win every round. Janeway tells her there is more to the game than stamina, and names intuition. Seven calls intuition a human fallacy. That scene states the show’s whole position on machine intelligence in about ninety seconds. The captain has a capability the enhanced being does not. It is unnamed, unquantified, and located precisely where measurement fails. That is the writing’s requirement rather than an accident of it. Whatever the machine cannot yet do gets promoted to the definition of the human. Then the machine does it, and the definition moves. Since 1967 that boundary has moved in one direction. Chess. Checkers, solved outright. Go. Heads up poker, including the bluffing, which was supposed to be the human preserve inside the human preserve. Protein folding. Language. Images. Code. Competition mathematics. Each time, the thing that fell was described beforehand as the thing that required judgment, feel, or intuition. Each time, after it fell, the description was quietly reassigned to whatever was still standing. I am not claiming the boundary vanishes. I am claiming that anyone whose position depends on where the boundary sits today is renting that position on a short lease. One android in a galaxy of ships Here is the structural miss underneath all of the others. The Federation spans thousands of worlds. It has replicators, warp drive, and a fleet. It has one Data. Machine intelligence in Star Trek is artisanal. It is a singular achievement by a lone genius, Doctor Noonien Soong, who dies without leaving reproducible instructions. Data is unique, precious, legally ambiguous, and irreplaceable.
