The first video game was made in 1947, when Thomas T. Goldsmith Jr. and Estle Ray Mann filed a US patent for a cathode-ray tube amusement device. It used analog circuitry to steer a dot across a screen and simulate firing at targets. If you count only games running on stored-program computers, the answer shifts to 1950 or 1958, depending on which definition you accept.

That gap of eleven years is not trivia. It exposes a definitional fault line that runs through every history of the medium: what counts as a video game, who gets credit, and why the popular answer (often "Pong, 1972") is roughly a quarter-century late.

Mechanism: how a 1947 game drew a dot on a CRT

Goldsmith and Mann's device had no computer, no microprocessor, and no memory in the modern sense. It connected to an oscilloscope-style cathode-ray tube and let a player adjust two knobs. One knob moved a dot horizontally; the other moved it vertically. The player then pressed a button to "fire" at a target drawn on a transparent overlay fitted over the screen.

The electronics were pure analog. Potentiometers controlled the deflection voltages fed to the CRT's horizontal and vertical plates, which steered the electron beam. The target was not rendered by the machine at all. It was a physical plastic sheet, so the "graphics" were part overlay, part glowing dot. [Citation: US Patent 2,455,992, filed January 25, 1947, granted December 14, 1948.]

Why this matters mechanically: the device computed nothing. It shaped voltages. That distinction is exactly why historians split over whether it is a video game or an interactive display. The screen responded to player input in real time, which satisfies most working definitions of a video game. But no program ran, so nothing was "computed" in the Turing sense.

Quantified context: dated milestones, one per definition

Different definitions produce different "firsts." The table below lays them out so you can see exactly which threshold each claimant clears.

Claimant Year Runs on a computer? Programmable? Visual output Typical verdict
CRT amusement device (Goldsmith/Mann) 1947 No No Analog dot on CRT First patented video game
Bertie the Brain (Joseph Kates) 1950 Custom tube logic No Light bulbs, not raster First electronic game with a display
Nimrod (Ferranti) 1951 Dedicated computer No Lamp panel First computer built to play a game
Tennis for Two (William Higinbotham) 1958 Analog computer No Oscilloscope trace First computer game with a real screen
Spacewar! (MIT) 1962 PDP-1 Yes Vector CRT First widely shared digital game
Pong (Atari) 1972 Arcade hardware Yes Raster CRT First commercial hit

Each row clears a different bar. Bertie the Brain, built by Kates for the 1950 Canadian National Exhibition, played tic-tac-toe on a 4-by-4 grid using a custom array of vacuum tubes. It stood roughly 4 meters tall and drew crowds for two weeks. It had no screen; the board lit through bulbs. So it is electronic, and a game, but arguably not video.

Nimrod, unveiled at the 1951 Festival of Britain, ran on a Ferranti computer and played Nim. It used a lamp panel rather than a CRT. Again: computer game, yes; video game, contested.

Tennis for Two is the claimant most often called "the first video game" in museum labels and textbook captions. Higinbotham built it at Brookhaven National Laboratory in about three weeks using an analog computer and an oscilloscope. Two players turned knobs to hit a bouncing ball over a net line. Visitors played it at an open house in October 1958. It was never patented and never sold.

Edge cases: why the 1950s keep getting re-dated

The 1950s produced at least four distinct "firsts," and each one is real within its own frame. The confusion is not sloppy history; it is unresolved taxonomy.

The CRT device is the earliest documented interactive electronic game. The patent filing date is January 25, 1947, and the grant date is December 14, 1948. Those are hard dates, verifiable in the US patent record. Everything earlier is a claim without a filing.

Bertie the Brain (1950) is the earliest game a public could actually play, if you accept light-bulb output. It predates Nimrod by a year and Tennis for Two by eight.

Nimrod (1951) is the earliest game running on a general-purpose stored-program computer, though it displayed through lamps rather than a screen.

Tennis for Two (1958) is the earliest game with a true electronic display that simulated a physical sport, and the earliest built purely for entertainment rather than demonstration.

Spacewar! (1962) is the earliest game that spread between institutions. Students at MIT wrote it on a DEC PDP-1, and copies traveled to other PDP-1 sites on paper tape. That distribution is why it, not Tennis for Two, anchors most academic histories.

A counterintuitive detail worth holding onto: the patent for the first video game was granted before anyone had built a commercial computer game, and the inventors never marketed it. Goldsmith and Mann's device was never manufactured at scale. It exists as a patent, a prototype, and a legal claim, not a product. The first video game was also, in commercial terms, the first vaporware.

Misconceptions: Pong, the "first game," and the patent problem

The single most common error is dating the first video game to Pong in 1972. Pong was the first arcade video game to achieve mass commercial success, and it turned Atari into a company. But it arrived 25 years after the CRT patent and 14 years after Tennis for Two. Calling Pong "the first video game" confuses market adoption with invention.

The second error is assuming "first video game" and "first computer game" are the same question. They are not. The distinction hinges on whether output must be a raster or vector display driven by a video signal. Under strict definitions, Bertie the Brain and Nimrod fail and Tennis for Two passes. Under loose definitions, all three count.

The third error is treating the 1958 date as settled. It is the most cited answer, and it is defensible, but only if you add "running on a computer with a screen." Without that qualifier, the 1947 device wins.

Red flag: sources citing 1958 without mentioning the 1947 patent are using a narrower definition, often without saying so. Red flag: "Pong was the first video game" is a commercial-history claim, not an invention claim. Red flag: dates for Bertie the Brain and Nimrod are exhibition dates, not filing dates; verify which one a source means. Red flag: the CRT device was never mass-produced, so "first to market" belongs to a later machine. Red flag: emulator footage of 1950s games is a reconstruction, not the original hardware.

Timeline: from patent to arcade in six steps

Year Event What changed
1947 Goldsmith and Mann file the CRT amusement device patent First documented interactive electronic game concept
1948 Patent granted (US 2,455,992) Legal record established; no product built
1950 Bertie the Brain shown at the Canadian National Exhibition First playable electronic game in public
1951 Nimrod demonstrated at the Festival of Britain First game on a general-purpose computer
1958 Tennis for Two played at Brookhaven First game on a true electronic display
1962 Spacewar! written on the PDP-1 at MIT First digital game distributed to other sites
1972 Pong released by Atari First mass-market arcade video game

Read down the "what changed" column and a pattern appears. Each step adds one capability: interactivity, then public play, then computation, then display, then distribution, then commerce. No single machine did all six. That is why a single "first" date keeps slipping depending on which capability you prioritize.

Practical actions: how to answer the question for yourself

If you need a defensible answer, pick your definition before you pick your date. The steps below force that discipline.

Define the medium. Decide whether a video game requires a video display, a stored program, or both. Write the criterion down before reading any history.

Trace the patent record. The 1947 filing is the earliest documented, verifiable interactive game claim. Search US Patent 2,455,992 to confirm dates and claims directly.

Separate invention from adoption. Pong's 1972 date marks commercial success, not invention. Use it only when the question is about market history.

Check the hardware behind each claim. Bertie the Brain used bulbs, Nimrod used lamps, Tennis for Two used an oscilloscope. Only the last is a video display in the strict sense.

Cross-check museum and academic sources. When two dates conflict, look for the qualifier ("first with a screen," "first patented") that reconciles them.

State your threshold. A one-sentence caveat prevents most disputes: "Earliest documented: 1947; earliest on a computer with a screen: 1958."

A practical threshold that resolves most arguments: if the output device is a CRT or video monitor driven by a signal the player controls, the 1947 device qualifies; if the game must be computed by a program, the earliest defensible date is 1950 for tube-logic machines or 1958 for analog computers with displays.

Limitations: what the record cannot settle

The historical record is thin in specific places. No surviving film shows the 1947 device running, and no production unit is known. Bertie the Brain was dismantled after the exhibition; photographs exist, but no working hardware. Nimrod survives only in fragments and photographs. Tennis for Two was rebuilt in 1997 from Higinbotham's notes and period parts, which means the 1958 original cannot be tested today.

Those gaps matter because they limit what any "first" claim can prove. A patent proves conception and legal priority, not playability. An exhibition photo proves public display, not sustained operation. A reconstruction proves feasibility, not original performance.

The honest state of the evidence: 1947 is the earliest documented, verifiable interactive electronic game concept. 1950 is the earliest public playable electronic game. 1958 is the earliest game on a true video display running on a computer. 1972 is the earliest mass-market commercial video game. All four are correct within their definitions, and none of them is the whole answer.

One professional caveat worth carrying away: historians increasingly avoid the phrase "first video game" altogether. They say "earliest known" and attach the criterion. That is not pedantry; it is the only way to keep four legitimate firsts from collapsing into one misleading date.

The next time you see a single year attached to the question, check which threshold it silently assumes. The date is usually right. The missing qualifier is the problem.