Spacewar! leaves the PDP-1

After Table 1 in the Kinephanos study (2015). Expanded with new additions recovered by Berry and Marino in 2026. The PDP-8 source (Evan Suits's LAB-8 version, "Interplanetary death and destruction on your LAB-8") is archived in this project's sources. Ralph Gorin's PDP-10 version ran up to five ships and was the one played at the 1972 "Intergalactic Spacewar! Olympics" at the Stanford AI Lab, reported in Stewart Brand's celebrated Rolling Stone dispatch ("HOW MANY SHIPS? MAXIMUM IS 5").

Five ports, read closely

Some of these source code ports survived, in a source listing, a magazine write-up, or a recovered printout, but others are harder to find. Each is a piece of source code history and expands our understanding of how Spacewar! was diffused across the world.

• Cambridge: Duel on the PDP-7 · Peterson and Viner's "viscous space", 1968, and the Margaret Masterman connection.
• Minnesota: invisibility instead of luck · Kuhfeld's CDC 3100 version and its 1971 Analog write-up.
• Stanford: Gorin's five-ship game · the PDP-10 version played at the 1972 Olympics, and Alan Kay's "build your own".
• MIT-AI: the self-documenting SPCWAR · the ITS lineage that wrote its own version history into its comments.
• PDP-11: Bryant, Seiler and the GT40 · the 1974 DECUS version lost and recovered from a printout.

See also the deep dive into the PDP-1 itself, the machine the game was written for, the versions of the MIT game, and the restorations that keep it runnable.

← The MIT versions   Timeline →

What grew out of it, and how far it travelled

This section is tentative and part of ongoing research: the catalogue of commercial versions and the game's diffusion are still being verified, and the findings below are provisional.

Beyond the labs that re-implemented Spacewar! for their own machines, the game seeded the first commercial video games:

• Galaxy Game (Bill Pitts & Hugh Tuck, Stanford, 1971): the first coin-operated video game, a faithful PDP-11 re-implementation of the Stanford PDP-10 version, on an HP 1300A display.
• Computer Space (Nolan Bushnell & Ted Dabney, Nutting Associates, 1971): the first commercially sold coin-op video game, directly inspired by Spacewar! but rebuilt on a raster television display. Half a century later Norbert Landsteiner closed the loop: his Ironic Computer Space Simulator (masswerk.at, 2016) is Computer Space rewritten as new PDP-1 Macro assembly and run on the same emulated machine as the original Spacewar!, returning the raster descendant to the point-plotting display it had left behind, and pointedly imagining the coin-op as if it had been written for the computer that inspired it.
• Space Wars (Larry Rosenthal, Cinematronics, 1977): the first widely released vector arcade game, based on Spacewar!. Rosenthal, an MIT graduate who had seen the original (though by his own account never played it), built his own Vectorbeam hardware, adapting a commercial raster CRT with custom circuitry to trace vectors, and Cinematronics productised it. It sold on the order of 10,000 cabinets (accounts vary, from 8,000 to 30,000) and was the highest-earning US arcade game of 1978, a striking commercial turn for a freely circulating, public-domain game.¹ The vector display sometimes credited as Rosenthal's improvement was really a return to the point-plotting screen the PDP-1 had used all along.

That commercial line is also the only firmly documented route by which the game travelled abroad. The original on its native PDP-1 stayed essentially North American; the documented foreign diffusion in 1961–1980 is Japanese and commercial, one step further down the lineage: Taito's licensed Space Wars (Japan, 1978) and Sega's Space Ship (1978), both descending from the Cinematronics game. The one early European re-implementation is Cambridge's DUEL on a PDP-7 (c.1968). No Soviet original-game version is documented in the Russian-language histories so far surveyed, but do let us know if you have more information on this.

The game has also kept being rewritten by individuals. Bill Seiler, who co-wrote the 1974 PDP-11 version, later wrote a DOS/CGA SpaceWar (v1.72, 1985), distributed as shareware; in 2026 Pete Clarke produced a faithful Python/Pygame recreation of it, porting the physics and game logic 1:1 from Seiler's x86 assembly source.

One port survives only as memory. Peter Samson has recalled working later on a classified computer system for an unnamed "three-letter agency", an eight-screen installation he thought resembled the SAGE air-defence machine: when the hardware ran late, the team threw in some software for free, including a port of Spacewar! that ran four games at once across all eight screens, with random matchmaking so players did not know whom they were facing. By his account the system's intended purpose was "not too far removed" from playing Spacewar!. We only have the transcript of this interview as the source for this claim.

That military thread did not stay anecdotal. The early games industry was, as Kline, Dyer-Witheford and de Peuter argue, "heartily subsidized by military research": Spacewar!, the ARPANET that carried it, networked multiplayer systems and the three-dimensional navigation of virtual environments all "emerged from a context of Pentagon-funded research". The traffic ran both ways, and kept running. From ARPANET and Spacewar! to Atari, and then back from the arcade game Battlezone to the US Army's SIMNET networked tank simulator and on to BattleTech, military research and commercial gaming settled into what they call a "circular, self-reinforcing, synergistic dynamic": a relationship that began as a civilian spin-off from war preparation but became a way of getting the entertainment sector to help underwrite the costs of military innovation and training. Spacewar!'s free circulation inside a Cold-War research budget is the first turn of that loop.

Drawn from the project's working diffusion catalogue, compiled for the monograph; individual entries carry their own sources and confidence ratings. On the commercial history of Space Wars and the question of who could own a public-domain design, see All in Color for a Quarter; on the Cinematronics / Vectorbeam lineage and the vector-versus-raster question, see Thomas H. Rousse, "Reconstructing WARRIOR" (Kinephanos, 2015). The commercial and foreign findings are provisional pending further research.

¹ This raises a question the project intends to pursue rather than settle here: how a communally written, freely circulating, public-domain program became the basis of a copyrighted, top-grossing commercial product, and on what any claim to own such a design could rest.