Skip to main content Skip to main navigation menu Skip to site footer
Tennis for Two, Spacewar!, Pong, Computer Space, Graphics, coin-op, software studies, platform studies

Gravity in Computer Space

Noah Wardrip-Fruin (University of California, Santa Cruz)


Every field has its “just so” stories, which are easy to repeat uncritically. We are only prompted to stop once the stories are appropriately examined. Sometimes a new examination can be the result of uncovering new evidence, but it can also be the result of looking at familiar evidence through new lenses. This article examines the familiar stories about a group of early video games: Tennis for Two, Spacewar!, Computer Space, and Pong. It looks at them through the lens provided by two concepts not previously used: playable models and operational logics. In doing so, it reveals that a common story in game history—of Computer Space flopping due to its complication, while Pong succeeded due to its simplicity—is at the very least in need of amendment, and more likely should be abandoned. In the process, this article also clarifies the importance of gravity in the two game design spaces explored by these early games. Through this, it underscores the importance of understanding that some early video games were created without general-purpose computation, and that the affordances and limits of television technology are central to this era of the field’s history.


The word video appears in video games for a reason. Even though text-only games have a fond place in my heart, and certainly sound-only games exist, there is no denying the central place of video screens. It is on video screens, and with other elements of television technology, that video games came to prominence in our culture. The public space of the arcade and the private space of the television-connected console were where a new generation of games reached a broad audience—and where designers did the most work to bring new experiences to that audience.

David Sudnow’s Pilgrim in the Microworld is an early book about getting deeply connected to game screens.1 In describing this connection, Sudnow—an academic phenomenologist and passionate jazz pianist—brings his skills of keen observation, critical thinking, and dedicated practice to the world created by the video screen, its computationally driven objects, physical game controls, and his own actions. As Sudnow discusses, that world is not inside the screen. Rather, as he learned to play, and as each person learns, “we traverse the wired gap with motions that make us nonetheless feel in a balanced extending touch with things.”2

Figure 1

In Breakout for the Atari VCS/2600 (the version Sudnow wrote about), the player controls a paddle moving across the bottom of the screen, bouncing a ball against a wall at the top of the screen. Each bounce removes a brick from the wall. Each color-coded layer of the wall, when reached, makes the ball bounce faster than the layer before. Eventually, successful play requires that paddle positioning be done by reflex, operating the controller and the spatial model through the “system of bodily spaces” Sudnow describes, rather than through conscious planning. (Image of Breakout taken by the author during play on the Stella emulator, version 5.0.2)

What happens when we traverse that gap? Sudnow portrays the virtual paddle controlled by Breakout players (fig. 1) as like a part of a player’s body, akin to her own knuckle or the fully physical extension of a baseball bat:3

When you’ve got the paddle properly in hand it’s a different kind of thing, not really a thing at all, but an extension of your fingers. You bring the so-called “second section” of the paddle beneath the ball in the same way you can move the back of your hand toward a cup in front of you so the knuckle of your index finger touches it. You don’t have to look at that knuckle. … When a paddle or a bat is incorporated by the body, becoming a continuation of ourselves into and through which we realize an aim in a certain direction, such implements lose all existence as things in the world with the sorts of dimensions you measure on rulers. They become incorporated within a system of bodily spaces that can never be spoken of in the objective terms with which we speak of objects outside ourselves.4

Sudnow is describing the experience of having deeply learned the model of space in a game. Once it is learned, we cease to think about it, instead considering the actions we wish to take within it. Further, much of our ability to play within it is transferred from that game to similar games, just as most of our ability to drive transfers from car to car.

Once we can play within a game’s model, that becomes the foundation of further experiences. Most books about game design focus on introducing challenges and goals to push players past what they can easily do within models. My forthcoming book, currently titled How Pac-Man Eats,5 describes how additional meanings have been, and can be, layered atop the experience of virtual space.6

This article looks beneath the surface, an approach I pursue with three goals. First, I aim to trace the development of continuous, two-dimensional spatial models in video games—not to engage in a fetishization of firsts but for what their development can expose about the importance of aspects of these games that have commonly received less attention.7 The tracing reveals the importance of gravity, an aspect of these models that is seldom discussed. Second, through this process, I aim to emphasize something too-often forgotten: the work of game design innovation is not the exclusive province of sketchbooks and software, but also wiring diagrams and electric relays, and how the latter’s affordances and limitations have shaped our history. In particular, I recount two attempts to develop video game designs for wide audience-oriented hardware (comparatively inexpensive television technology), versions of which designs had previously existed in research labs (where they could employ computers far more expensive than could be deployed to wide audiences). Finally, I aim to specifically cast doubt on one of the most repeated “just so” stories in our field:8 that the reason Computer Space was not the first commercial smash hit video game (rather than Pong) was the complexity of its controls.9

But first, as background, I will explain the kind of model I’m discussing (using Michael Mateas’s concept of the “playable model”) and the foundations of such models in the fundamental elements of games (which I have termed “operational logics”).

Playable Models and Operational Logics

Consider a Disney-themed chess set, perhaps with Minnie and Mickey as queen and king, Daisy and Donald as knights, a bevy of princesses as pawns, and so on. Now consider a checkers board, minus its original pieces, being played instead with washers for one player and coins for the other. In the first case, it might be taken as critique to play a game in which young women are Disney’s pawns. But in neither case does anything change about the gameplay. Both Disney chess and household-object checkers are simply re-skins of an underlying game that remains the same.

Now consider the various games that licensed the DOOM engine from id Software in the mid-1990s.10 We could call these games re-skins of DOOM, but in most cases that would be misleading. Not only are the graphics changed but also generally the geography of the levels, the behavior of enemies, the types of weapons, and so on.

One view would be that the games use a set of technologies inherited from DOOM. From this perspective, the DOOM engine becomes a platform on which other games are constructed. This approach can provide valuable insights, as many works of platform studies reveal.

But it gives us no way to account for other games that appear quite similar. Consider games created using game developer Bungie’s engine. Like Heretic,11 Strife,12 and Chex Quest13(which were all built on DOOM’s platform), Bungie’s Pathways into Darkness14 and Marathon15 are about moving through constrained spaces and shooting at enemies. The code on which they rely is quite different from the DOOM engine, as is the hardware on which the code is designed to run (Macintosh computers of the 1990s, rather than Windows machines), but the playable models constructed through the code, running on the hardware, are deeply similar.

Playable models are a kind of “procedural representation”—they don’t provide a fixed representation (like a painting or sculpture), or one that unfolds in a fixed manner (like a film or linearly read novel), but one that is generated via processes, and for this reason may differ with each presentation, or even each moment of its presentation. Playable models is Mateas’s term for procedural representations that are designed to have the particular features of being “learnable” and “actionable,” as Mateas explains in this text he contributed to the report from a 2010 workshop:

[For a game] to be learnable, a player must be able to make inferences about a game’s state and build up a mental image or model of the underlying system as they interact with the rules of the game. This may not mean that they are able to completely reverse-engineer the system. Rather, a learnable computational model is one supporting the incremental development of simplified and partial mental models that successfully provide guidance for future exploration and interaction within the game rule system. This exploration is afforded through mechanisms of engagement, that is, a means for a player to affect the state of the game in a manner consistent with his or her desires. To meet this requirement, the game must also be actionable. Defining games as systems that employ such playable models distinguishes them from traditional systems and computational models in other disciplines such as physics or engineering, where cognitive properties of [being] learnable and actionable are not factors. 16

Pathways into Darkness and Marathon share with DOOM playable models of space and combat that are learnable and actionable in highly similar manners, producing very similar procedural representations when players take analogous actions in analogous environments. Such playable models can be implemented on widely different platforms and with widely different specific scenarios, but their experiences are fundamentally aligned.17 This is why our understanding transfers so well from game to game, even on platforms with no overlapping hardware or software.

The foundation of any playable model is a set of operational logics.18 For example, the spatial models of games like DOOM have logics such as navigation, movement physics, and collision as a foundation. Similarly, the primary foundation of Monopoly’s playable model of real-estate investmentis its resource logics, though a pattern-matching logic determines which properties can be further developed.

Joseph C. Osborn explains the relationship between playable models and operational logics in this way: “How can a designer construct a playable model for some real-world system? How can they ensure that they produce a system which both models a concept and makes that model playable? Simulations are built out of rules, and these rules are enacted by computation; but playability requires that these rules are communicated in a way that helps players understand the operations of the system. From an authorial standpoint, operational logics provide the building blocks of readable rules, and for the player their literacies in operational logics give them a ladder which they can climb to understand the model.”19

Operational logics provide such a ladder by cutting across a divide in how we often talk about games. Much game scholarship discusses separately what a game purports to communicate (its skin or theme or fiction) and how it operates and enables play (its rules or mechanics or models). But the foundational logics of games are both at the same time. We can only understand a logic such as collision by seeing it as something that communicates (e.g., virtual objects can touch) and something that operates algorithmically (e.g., when two entities overlap, do something). Specifically, I argue that operational logics are combinations of abstract processes with their communicative roles in the game, connected through an ongoing game state presentation and supporting a gameplay experience. These aspects will be explored further, below, in the context of particular logics.

From one perspective, operational logics are a site of remarkable flexibility and creativity. On a technical level, much as with playable models, a logic such as collision can be implemented in many ways—from the 2D hardware implementation of the Atari VCS/2600 to the 3D quadtrees or raycasting that might be found in current software. And on a communicative level, collision detection might be used to represent a wall we can’t pass through, a bullet hitting a body, or something more unexpected. Game designers have found a remarkable range of activities to communicate with this logic, from Pac-Man’s eating to the deflecting of hate speech found in Anna Anthropy’s Dys4ia.20

But from another perspective, the very fact that logics communicate something in particular (e.g., the virtual touch of collision detection) means that they can’t arbitrarily communicate anything. At most, they can communicate things we understand metaphorically, such as the deflection of a statement.

Continuous Spaces and Graphical Logics

Of course, games like DOOM, and the sort Sudnow describes, are not the only ones that are played in a virtual space. The very earliest video games, decades before public coin-op machines and private home consoles—such as Christopher Strachey’s 1951 MUC Draughts21—presented spaces of some sort. But like the board game Strachey emulated (known in the US as checkers), the spaces in these games were discrete.They were divided up into nonoverlapping spaces, and each game action involved moving a piece from one discrete space to another, with no meaningful in-between position available.

The works that introduced games as a cultural force, instead, had models of space that felt continuous,like DOOM’s. The first hit video game in the computing subculture—Spacewar!22created the experience of smoothly flying a ship through space.23 The first broadly known game, Pong,24 allowed players to smoothly move paddles to intercept a ball that, itself, moved smoothly through the space.

This feeling of continuity requires not only many potential positions in the virtual space (so many that moving between them creates a feeling of continuousness) but also an ongoing sense of time.25 Games like Strachey’s, on the other hand, also operate in terms of discrete time—one player moves, then the other moves. And this sort of spatial model, with discrete space and time, is not only common in video games that emulate board and card games. It is also commonly seen in genres such as strategy games, simulation games, and puzzle games.

For us to understand the core mechanics of games—the key actions that players take—we have to understand the logics and models that are their foundation. For example, jumping is a core mechanic in many games. But the possibilities for player action and game response make jumping fundamentally different in a game like Strachey’s (with a discrete spatial model, built using appropriate versions of logics) and a game like David Crane’s Pitfall!26 (with a continuous spatial model and different versions of logics).27 Though Crane’s game came three decades after Strachey’s, the importance of this difference made it possible for Pitfall to become a highly influential early entry in the platformer genre.

This brings us to the next goal of this article. Three key families of logics, mentioned above, are often called graphical logics: collision, movement physics, and navigation (the last being the common name for control logics used in spatial models). Each of these could be implemented in many ways, including as part of discrete spatial models. But I (and others) use the phrase graphical logics to refer to these logics when they form the foundation of continuous spatial models, presented visually. This is now usually done with the tools of computer graphics; such models are, in a sense, computer graphics made playable. (However, as this article will describe, they have not always been implemented this way.)

These logics are key to many of the experiences that defined video games as we know them. In particular, they undergird the continuous, two-dimensional spatial models upon which video games grew into an important cultural and economic force, as with coin-op games of the 1970s like Pong. And they continue an essential role today, as with mobile phone games such as Angry Birds,28 which have made video game play ubiquitous.

Collision, Movement, and Physics: Tennis for Two

The history of computing is a funny thing. Innovative hardware and software are made up of reusable parts, or made for machines with other uses, and then may be disassembled or overwritten without leaving a trace. What history we know of video games, like other areas of computing, is the result of the happenstance of preservation as much as deliberate archiving and investigation.

Raiford Guins is probably the person who has spent the most time investigating the history of the game now commonly called Tennis for Two.29 As Guins discusses in his 2014 book Game After, it’s not even clear how the game got that name—which seems to have appeared about four decades after the last time anyone played the original.30 Luckily, the game was also investigated in the 1980s, as part of a long series of video game lawsuits initiated by Magnavox, which itself created greater interest, leading to the preservation of documents and conducting of interviews that have helped us understand some of its history. As Guins reports, these documents include diagrams of the original plans for the circuits, as well as notes indicating that the plans were revised in practice, but with no documentation of the revisions.31 So the original game cannot be entirely recovered. Still, we know about it, both from documents of the time and from attempts to recreate it since.

Tennis for Two is certainly not the first video game.32 I currently believe that honor belongs to Strachey’s MUC Draughts, discussed above(which I have given that name because I believe its lack of a name is part of why it has not been credited more widely).33 But I also believe that Tennis for Two holds an important place in the history of video games: it is the first known video game with a continuous model of space, and it introduced versions of the key operational logics of collision and movement physicsthat are appropriate for building such models. (Here I will discuss Tennis for Two as introducing a “collision detection” logic, focusing on the moment of touch, rather than “collision handling,” which includes threading the collision event through more elements of the model. But both are within the larger family of collision logics.)

Tennis for Two was created by William Higinbotham, Robert V. Dvorak, and David Potter as a demonstration for the 1958 visitors’ day at Brookhaven National Laboratory (BNL).34 It presented a simplified side view of a tennis court on an oscilloscope screen, with a visible ball, net, and groundline, but invisible rackets. Two players used simple controllers, each with two inputs: a button for determining when to hit the ball and a dial for determining the angle. In footage from the documentary When Games Went Click35—taken of a re-creation led by Peter Takacs, Gene Von Achen, Paul O’Connor, and Scott Coburn36—when the ball hit the net or ground, it would bounce back (figs. 2 and 3).

Figure 2

Scott Coburn, Peter Takacs, and Gene Von Achen who, together with Paul O’Connor, reconstructed the retrospectively named Tennis for Two for its fiftieth anniversary in 2008. As Diane Greenberg reports, “Using Higinbotham’s original plans, Takacs and his colleagues rebuilt the game with vintage parts, including mechanical relays and germanium transistors that first became commercially available in the 1950s. The original 1950s model analog computer that was made from vacuum tube circuits had to be simulated using modern integrated circuit chips.” See Diane Greenberg, “Celebrating ‘Tennis for Two’ with a Video Game Extravaganza,” The Bulletin, Brookhaven National Laboratory, October 31, 2008, (Image courtesy Brookhaven National Laboratory)

Figure 3

In a recreation of Tennis for Two , the ball, net, and ground are visible, but the rackets are not. Recreated for the New York Historical Society Museum’s exhibition, “Silicon City” (2014–15). See Raiford Guins, “Tennis For Two at the NY Historical Society Museum,” November 19, 2015, Re-creation consisted of a simulation of an oscilloscope displaying an emulation (by Ben Johnson and Peter Takacs), reconstructed controllers (by Adelle Lin), and exhibition design that “enlarges” the original oscilloscope (by Jeanne Angel, working with the show’s exhibition and graphic designers, John Esposito and Kira Hwang), with consultation from Raiford Guins. (Image courtesy Raiford Guins)

This bouncing is particularly important because it is here that we see the collision detection logic at work. All the elements of a logic are in place. The game state is presented on the oscilloscope, showing the ball colliding and responding to that collision. The display is controlled by analog computation, carrying out the abstract process, “When the boundaries of two virtual objects intersect, declare the intersection,” enabling the response.37 The gameplay experience involves being able to drive the game’s collision detection over and over, learning how it operates in the spatial model of Tennis for Two. Taken together, all parts of the operational logic clearly fulfill the communicative role, “Virtual objects can touch, and these touches can have consequences.”

Interestingly, as Nathan Altice pointed out to me in conversation,38 it is less clear that collision detection supports the key mechanic (or gameplay “verb”) that it will in future tennis/table tennis video games.39 This mechanic is returning the ball when it is served, or returned, by the other player. Because the paddles are invisible, it is hard to argue that collisions with them are communicated, so the logic is unfulfilled. It is the unfulfilled collision logic that most strongly gives the sense that Tennis for Two is not yet a mature version of a tennis/table tennis video game.40

The other logic introduced by the game (in a version appropriate for playable models of continuous space) is movement physics. This logic communicates that virtual objects move according to physical laws, sometimes in ways beyond the direct control of players. The abstract process evaluates the rules that apply to determine future positions of virtual objects. Of course, in the case of Tennis for Two, these rules were not written down mathematically, or as software, but rather assembled in physical components. As I aim to illustrate in this article, for those interested in understanding the logics of video games deeply, looking at early games is helpful in part because it broadens our thinking about how logics can be implemented. Electric components can be where an innovative game designer does her work, not just lines of code.

One interesting aspect of Tennis for Two is that it doesn’t have objects that players can move around the visual space. Rather, every movement shows players an example of what happens when the ball moves with particular force at a particular angle. The game state presentation is always showing how the forces of physics are shaping the arcs, bounces, and remaining momentum of the ball. And this is key to the player learning the spatial model and developing the ability to effectively use the game’s two key mechanics: serving the ball and returning the ball.41

Tennis for Two also illustrates another important idea about operational logics: they can be implemented in different ways, which shapes the resulting gameplay experience and playable model. For example, the movement physics logic in Tennis for Two is implemented in a way that includes gravityand wind resistance, which are often absent from later games (like Pong). The implementation of gravity and wind resistance threads through all aspects of the logic. For example, in Tennis for Two (and many other games) the implementation of gravity adds an abstract process: all unfixed elements not attached to a fixed element, or not on the opposite side of a fixed element, have attraction toward a point or region of the space, of which there may be more than one.42 The communicative role is that one or more elements (the bottom of the screen in Tennis for Two, the central star in Spacewar) is a source of gravity. This issue will become key in our discussions of Spacewar, Computer Space, and Pong.

Tennis for Two’s innovation was not in using analog computation to present images on the screen, or even in using collision detection and physics. As BNL’s account tells us, “The computer's instruction book described how to generate various curves on the cathode-ray tube of an oscilloscope, using resistors, capacitors and relays. Among the examples given in the book were the trajectories of a bullet, missile, and bouncing ball, all of which were subject to gravity and wind resistance.”43 But using these elements to construct video games’ first playable model of continuous space and designing the first example of a tennis/table tennis video game (which would go on to great commercial success) were certainly very significant innovations.

Higinbotham and his collaborators also employed an inventive approach to implementation—particularly, implementation of the visual display. Versions of this approach would later become very important to game developers in other resource-constrained contexts, such as the Atari VCS/2600 (as described by Nick Montfort and Ian Bogost in Racing the Beam44). This approach involves using a single visual output to draw multiple objects on screen, by changing its position quickly. BNL employee Peter Takacs, who led the re-creation of Tennis for Two, describes this: “Higinbotham used the transistors to build a fast-switching circuit that would take the three outputs from the computer and display them alternately on the oscilloscope screen at a ‘blazing’ fast speed of 36 Hertz. At that display rate, the eye sees the ball, the net, and the court as one image, rather than as three separate images.”45

Tennis for Two’s gameplay was novel and compelling enough that visitors lined up to experience it. However, it seems unlikely that serious thought was given to opening up the experience to a wider group. The game was composed of a large, heavy amount of hardware, driven by an analog computer.46 This was before the spread of digital computation made such experiences much easier to distribute—because they could travel as software for machines already configured with display screens, rather than as schematics for assembling and configuring an analog computer and additional hardware. That moment of software distribution for a landmark game came, instead, with Spacewar—the game that introduced the fundamental combination of a continuous spatial model, a combat model, and scenario/level design to the computing world.

Navigation, Combat, and Scenario Design: Spacewar!

In 1961, Digital Equipment Corporation (DEC) occupied an unusual place in the computing landscape. Though it was a sizable company selling very expensive machines, it was also seen as a radical upstart, challenging the computer orthodoxy overseen by the dominant International Business Machines (IBM).47 DEC was selling a vision of interactive computing, or working directly on a computer, that now seems everyday—but was an abrupt break with the tradition of delivering programs to computer operators (perhaps as trays of punch cards) and then waiting patiently for a reply after the scheduled run of one’s program. What DEC was selling wasn’t brand new, however. It was an experience pioneered in universities, government labs, and other research organizations—such as the MIT labs that were home to the legendary Whirlwind and TX-0 computers.

In the summer of 1961, the three people who conceived of Spacewar—Steve “Slug” Russell, J. Martin Graetz, and Wayne Wiitanen—were all working in the IBM computing paradigm (at Harvard University's Littauer Statistical Laboratory).48 Then Graetz was hired at MIT by electrical engineering professor Jack Dennis, a recent PhD who was in charge of the department’s computers, and Russell returned to MIT’s Artificial Intelligence Group.49 Dennis gave students and staff remarkable latitude in using the computers,50 including the TX-0 and, particularly, the PDP-1 computer donated by DEC in September 1961.51 Russell, Graetz, and Wiitanen came up with the concept for Spacewar while brainstorming what to do with the Type 30 Precision CRT display that was scheduled to be installed a couple months after the PDP-1. They wanted to outdo the existing demonstration programs for the Whirlwind and TX-0, which included a bouncing ball, a mouse that would traverse user-constructed mazes, a tic-tac-toe game, and Marvin Minsky’s generative animation program Tri-Pos (better known as the Minskytron).52

The first version of Spacewar was completed by Russell in February 1962. Two visually distinct ships could rotate and thrust to move across the screen in a manner governed by relatively realistic physics (momentum is key to playing Spacewar). In creating this much, Russell had already introduced another key logic for playable spatial models: navigation. Further, the two ships could fire projectiles at each other, and destroy each other when there was a collision between a projectile and a ship—introducing video games’ first mechanics of combat in continuous time. The wedding of these two experiences—of continuous spatial movement and combat in continuous time—became an extremely important, almost dominant, approach as video games developed further.53

Focusing on Spacewar’s introduction of the navigation logic, we can see what an important difference it makes that the rackets in Tennis for Two are invisible. For a model of any domain to be playable, it must support interaction.54 That is, there must be a control logic available to players that changes the state of the model in a way that is presented to players. For continuous spatial models, the key control logic is navigation. Even a game as simple as Pong gives players the ability to navigate paddles up and down the screen. But because of the nature of Tennis for Two’s paddles, the introduction of navigation to continuous space video games fell to Spacewar.55

The abstract process for navigation, in this type of model, could be stated as moving one (or more) virtual object(s)—and/or the rest of the virtual space apart from the object(s)—so that all the elements of one are progressively offset relative to the other, based on a player-controlled input, for as long as that input is in effect. The communicative role could be stated as the player controlling the type, and/or direction, and/or timing of the movement of a virtual object with which she identifies. (Both the abstract process and the communicative role would be different for a navigation logic in a different sort of spatial model—such as the first-person 3D spatial model of DOOM or the link-based, textually represented model of the William Crowther and Donald Woods Adventure.)

The implementation of navigation requires that some process is “listening” for input from the player and that some process can make a representation of something under the player’s control change position in the space of the model. Of course, the player’s control can be partially indirect, as when the player acts together with a physics logic. This is the case in Spacewar, where the player’s primary mechanic for movement is thrusting. The player holds down the thrust control for differing amounts of time, and the ship is rotated to different angles to provide input to the physics logic that (together with the other forces already in effect) determines where her ship will navigate. Indirect control is also present in other types of games. For example, in platforming games players can often let go of (or walk off of) ledges, with the expectation that physics will take them down in the model’s space.

Learning to perform navigation, both directly and indirectly, begins with players identifying what represents them in the game state presentation—then experimenting with interface controls to learn how the navigation mechanics operate. As players become more fluent in navigation, they are increasingly able to choose actions to take in the game’s spatial model (ones that they desire, and ones that the game supports) and anticipate the potential results. Player understanding may initially be considered and deliberate. But as Sudnow and others discuss, in most cases it quickly becomes a bodily understanding, with the things that represent us in continuous space games (and the interfaces that bridge our actions and theirs) coming to feel like extensions of our own bodies and decisions about how to act made at a preconscious level.56 If we fail to make this transition, many games (including Spacewar with a competent opponent) move far too quickly for us to succeed. Brendan Keogh—building on the work of Sudnow, N. Katherine Hayles, Henri Lefebvre, James Ash, and others—writes of this as “embodied literacy.”57

At this point in Spacewar’s development, it might have been considered complete—except for one glaring problem. Over time, it wasn’t a very compelling experience. It was an exciting demo, but not something with depth, something that would pay off players’ investments of time and consideration. As Graetz writes: “Up to this point, Spacewar! was heavily biased towards motor skills and fast reflexes, with strategy counting for very little. Games tended to become nothing more than wild shootouts, which was exciting but ultimately unrewarding.”58

Spacewar lacked an interesting gameplay scenario, what is often referred to as level design (though games such as Spacewar obviously are not divided into levels). It didn’t lack primarily in terms of the logics it implemented for play, though the further development of one of these was key to its next step. Rather, Spacewar lacked an interesting situation for play. It lacked a selection and arrangement of elements that would matter in terms of its models and mechanics, shaping patterns of play—opening new strategic possibilities and new skills needed to exercise them.

The solution was twofold. First, Dan Edwards expanded the physics logic of Spacewar to include gravity. Second, a large star was introduced in the middle of the play area, with a gravity well encompassing the entire area.59 Colliding with the star, like collision with a projectile, caused immediate death (which would happen relatively quickly if players applied no thrust to their ships at the start of a game). As Graetz explains: “The star did two things. It introduced a player-independent element that the game needed; when speeds were high and space was filled with missiles, it was often sheer luck that kept one from crashing into the star. It also brought the other elements of the game into focus by demanding strategy. In the presence of gravity both ships were affected by something beyond their control, but which a skillful player could use to advantage.”60

The first fruit of this new scenario was a maneuver called the CBS opening—because the paths of two ships executing it resembled the television network’s eye-shaped logo (fig. 4). Skilled players, rather than immediately thrusting away from the star, would fire short blasts directing their ships to fall into the gravity well and whip around the star. This proved to be the fastest way to build up all-important momentum.61

Figure 4

Whipping around Spacewar ’s central star, through the use of momentum and gravity, creates a shape reminiscent of the U.S. CBS network’s logo in this time-lapse image. (Image from J. M. Graetz, “Spacewar! Real-Time Capability of the PDP-1,” in DECUS Proceedings 1962: Papers and Presentations of the Digital Equipment Computer Users Society , 1962, 38,

Spacewar was complete by the end of April 1962 and was first shown publicly at MIT’s Science Open House in May. Its creators expected a crowd and got one. By summer the group that completed it was drifting away from MIT, but the game had taken on a life of its own. As Graetz reports, “Program tapes were already showing up all over the country, not only on PDP-1s but on just about any research computer that had a programmable CRT.”62 Unlike Tennis for Two, which could not be distributed as software, Spacewar was on its way to becoming the world’s first hit video game.

But Spacewar’s rapid spread wasn’t simply due to labs sharing tapes with each other. As Henry Lowood writes, “the game was shipped by DEC with PDP computers as a test program to verify their operation after new installations. Spacewar became a fixture in university and industrial laboratories of the 1960s and 1970s.”63 Spacewar was a perfect demonstration of the vision of personal, interactive computing that DEC was selling—so DEC not only distributed it, but selected it to be one of the first things seen after each successful DEC computer installation.

Further, as a testament to the depth of strategy and skill that could be brought to the completed Spacewar, Stewart Brand (of Whole Earth Catalog fame) organized a “Spacewar Olympics” at the Stanford Artificial Intelligence Laboratory (SAIL) in 1972. The event was sponsored by Rolling Stone, and Brand published his account both in that magazine and in his later book, II Cybernetic Frontiers.64 This not only served to document the game’s popularity after a decade of play but also brought that popularity to another level of public consciousness.65

Yet there appeared to be no easy way to bring Spacewar to that public. The very fact that made it almost omnipresent in the computer culture was what blocked it from the wider culture: it was software for digital computers. Such computers were simply too expensive for any sort of commercial public deployment of games, even a decade after Spacewar’s creation.

The Importance of Implementation: Computer Space and Pong

Lowood’s excellent article “Videogames in Computer Space: The Complex History of Pong” tells the story of what happened after Spacewar’s unprecedented spread through the 1960s computer culture.66 He traces the converging paths of commercial video game pioneers Nolan Bushnell and Ralph Baer, and in doing so tells the story of how certain approaches to continuous spatial models—and the game designs they supported—became dominant in the early video game coin-op machines and home consoles.

Bushnell’s story is the more famous. He graduated in 1968 from the University of Utah, where he later claimed to have played Spacewar, and moved to California to work for the technology company Ampex. This was not far from Stanford University, the future site of SAIL’s Spacewar Olympics, and the likely site of Bushnell’s actual first exposure to the game.67 Working with fellow Ampex employee Ted Dabney, Bushnell sought a way to commercialize Spacewar—but the great expense of even stripped-down digital computers (such as the Data General Nova) made it seemingly impractical. Two others actually did this, but only in a limited manner, as Lowood reports: “[A] recently graduated SAIL student, Bill Pitts, and his friend Hugh Tuck built a coin-operated (coin-op) computer game, The Galaxy Game, for the newly released PDP-11/20, DEC’s first 16-bit computer. … [They] converted the PDP-10 version of Spacewar for this machine.”68 With a total expense of $20,000 (including computer, display, cabinet, and controllers), it was clearly not a route to reaching the broader public.

Bushnell and Dabney came to realize that the only practical route was to attempt to implement the key operational logics in hardware—using television technology—rather than in software for general-purpose digital computers. In pursuing this route, they were able to reproduce a good number of Spacewar’s key elements in innovative new implementations. For example, as Lowood writes: “[A] small number of diode arrays connected to logic gates produced the rotating images of rockets seen on the screen; the rocket images were clearly visible even in the pattern of diodes on one of the PC boards. … Bushnell’s rockets were essentially hardwired bitmaps that could be moved around the screen independently of the background, a crucial innovation that made it possible to produce screen images efficiently.”69

This enabled the logic of navigation. However, other elements had to be stripped out, most notably reducing the physics logic to remove gravity—and with it Spacewar’s central star scenario design (fig. 5). The result, Computer Space, was released in August 1971 by Nutting Associates. Computer Space failed to become a hit, while Pong was a great success the following year. This is one of the most discussed pairs of events in video game history, and the outcome is generally attributed to the complexity of Computer Space as compared with Pong’s simplicity.

For example, Mark J. P. Wolf writes, “The first arcade video game, Nolan Bushnell's Computer Space (1971), failed commercially, because players found its controls difficult to understand and use, and it was not until Bushnell's second effort, PONG (1972), which had a single control and simplified graphics, that the videogame as a commercial entity finally found success.”70 John A. Price offers a similar comparison, writing that Bushnell and Nutting Associates “released the world’s first commercial video game under the title of Computer Space in 1970. Only 1,500 of the machines were sold, in part because the game was too complex and too difficult to play. Determined to come out with a game that was not too complex, in 1972 Nolan Bushnell started the Atari Company. Their first game was a table tennis game called Pong for the arcades. Pong was highly successful and it was widely imitated.”71 Nick Montfort and Ian Bogost, in the book that initiated platform studies, tell the story in similar terms—observing that Bushnell sought a way to bring an experience like Spacewar to the public, and the result “was Computer Space, which arcade game manufacturer Nutting Associates released in 1971 to very limited commercial success. Complexity of play was part of the problem—the general public wasn’t accustomed to arcade games … . Pong solved the problem that plagued Computer Space—ease of use—partly by being based on the familiar game table tennis and partly thanks to the simplicity of its gameplay instructions.”72

Interestingly, the common version of this story offers no source for the idea that being overcomplicated led to Computer Space not becoming the first hit video game. Those that do cite a source indicate that Bushnell may be a contributor to (or even the originator of) the idea. In another book Wolf notes: “Bushnell attributed the lack of success of Computer Space to be partly due to the complexity of the game’s instructions and its gameplay mechanics.”73 Lowood quotes Bushnell as saying that “‘the typical guy in the bar’ was completely baffled.”74 Steven L. Kent writes in The Ultimate History of Video Games, “Bushnell admits that the instructions were too complex: ‘Nobody wants to read an encyclopedia to play a game.’ He also blames Nutting for marketing the game badly.”75 Tom Sito offers the same Bushnell quote, in his Moving Innovation: A History of Computer Animation, “the instructions were too complicated. ‘Nobody wants to read an encyclopedia to play a game,’ Bushnell confessed.”76 Sito writes this despite noting, on the previous page, that the more complicated Galaxy Game (discussed earlier in this article) was being played by the “coffeehouse habitués” on the Stanford campus, only a few miles away from where Pong would become a hit.

There are, of course, many other reasons one might imagine for Computer Spacenot catching on as Pongdid. Perhaps it was the fact that it was a single-player game at first, with a two-player version not introduced until 1973. Maybe the issue was cabinet design—as Guins notes, the wood grain design of Pong (as opposed to the fiberglass cabinet of Computer Space) was such that it would fit into many more contexts.77 Maybe it was a game scenario that, while compelling in the computer subculture, didn’t resonate with the broader public.

Figure 5

Computer Space had the dueling spaceships of Spacewar —both in the original single-player version and a later two-player release—but without gravity and the central star. This was previously found to be a less-than-compelling experience when the original Spacewar was being developed at MIT. Unfortunately, Bushnell and Dabney do not appear to have known this, and Bushnell later blamed the failure of Computer Space on its complexity . Image of Computer Space from a cabinet repaired by Ed Fries. See Ed Fries, “Fixing Computer Space,” EDFRIES: The Game Is Not Over (blog), March 13, 2015, (Image courtesy Ed Fries)

But a different explanation is suggested by the facts already discussed in this article. The design of Computer Space—essentially Spacewar without gravity and the central star—was already known to lack compelling gameplay in February 1962, when the game was still in initial development at MIT. While its complexity, or (initially) single-player approach, or cabinet design, or subject matter, or marketing may have been an additional issue, it seems likely that the key reason for the failure of Computer Space is that the versions of its operational logics that could be implemented in commercially viable technology could not support a compelling version of Spacewar-style game design. As former Atari employee Jerry Jessop commented to the New York Times, “The game play was horrible.”78

Matt Barton’s account, in Vintage Games 2.0, is the only oneI have found that directly discusses the gap between Bushnell’s preferred story and the realities of Computer Space gameplay:

Bushnell concluded that the game was too complicated for gamers accustomed to pinball. “The lesson was,” said Bushnell, “keep it simple.”

However, Computer Space’s failure is probably owed more to its dull gameplay than dull-witted players—as later, more successful efforts shows[sic].79

As Barton suggests, should we require further evidence that a game of Spacewar’s complexity could be a success outside the lab, we need only look at the story of Space Wars,80 a 1977 version developed by Larry Rosenthal for Cinematronics. As Barton and Bill Loguidice write, “Rosenthal’s key innovation was developing a special processor, which was cheap to make yet still sophisticated enough to run the full version of Spacewar!, complete with the gravity well and two-player dog fighting that made the original so compelling.” Rather than turning off audiences with its complexity, “it was a wonderful adaptation of Spacewar! and earned rich profits for Rosenthal and Cinematronics,” Barton and Loguidice conclude.81

Looking again at the hardware for these early games may help clarify why Computer Space’s development resulted in its particular design. Lacking Rosenthal’s specialized processor, it appears that the implementable spatial models for Bushnell and Dabney, and the gameplay they could support, were significantly determined by available television technology. Baer’s story, running in parallel with Bushnell’s and Dabney’s, helps illustrate this.

As Baer reports in Videogames: In The Beginning, in 1966 he was working at Sanders Associates—and while waiting for a bus, returned to previous musings about using home televisions as game devices.82 After some unofficial work by himself and Bob Tremblay, Baer got approval to pursue the project further, bringing in Bob Solomon, Bill Harrison, Bill Rush, and others. The team developed technology prototypes and games at the same time, including games with player-controlled objects chasing each other, so-called pumping games in which players tried to raise or lower color levels (for example, one player trying to raise blue water behind a screen overlay of a burning house, while the other tried to lower red fire from the top of the screen), and multiple-choice games. But nothing seemed compelling enough to justify a home unit’s likely cost.

When Rush suggested a third, machine-controlled object (in addition to the two player-controlled ones) the Sanders team quickly found themselves exploring the types of tennis/table tennis video games that Higinbotham and his collaborators had pioneered (though Baer’s team was likely unaware of Tennis for Two). This might seem like a doomed project, given the discussion above. After all, gravity was as central to the design of Tennis for Two as it was to Spacewar, and maybe even moreso, because the fundamental gameplay of Tennis for Two was lobbing the ball back and forth, then choosing the right moment in its parabola to hit the button and return the ball to the other player along another gravity-controlled arc.

The Sanders team didn’t invent video games (as Baer and others have suggested). But they did come to a crucial realization: that a compelling tennis/table tennis video game could be created without gravity implemented in its physics logic. What this required was (literally) a different perspective from that pursued by the BNL team. Rather than the side view of Tennis for Two (as though sitting near the umpire, looking down the line of the net), the Sanders team adopted a top-down view (as though looking at the game from above, with the net stretched from top to bottom of the screen). When seen in this way, the gravity-defined arcs of the balls in such games would become much less salient, such that they could be abstracted away while still having the video game be interpretable as a procedural representation of the physical world gameplay.

This sort of gameplay turned out to be compelling. And without the need for gravity, it could be supported by relatively low-cost television technology.This meant the Sanders team had an appealing game using components that could potentially be sold in a home consumer unit. This led to a licensing deal with television manufacturer Magnavox—who then produced the groundbreaking Odyssey, the first home video game system, released in 1972.83

The Odyssey system was first demonstrated in spring, but not released until summer—a gap that proved important. Bushnell attended an early demonstration and, shortly afterward, Bushnell and Dabney separated from Nutting and founded the company that would become Atari. They were soon joined by another ex-Ampex engineer, Al Alcorn, who Bushnell assigned to create a simple tennis/table tennis video game—which he characterized as the simplest game he could think of. However, as Lowood writes, “When he tasked Alcorn with a ball-and-paddle game, his suggestion must have been influenced by what he had seen from Magnavox.”84

Alcorn’s top-down-view game, christened Pong, was launched in a tavern later in 1972 and—as noted in multiple accounts quoted above—became an immediate hit, initiating the coin-op era of video games. It presumably even helped drive sales of the Odyssey, which for a time was the only way to play a tennis/table tennis video game at home. In becoming such a great success, Pong demonstrated to the world that top-down tennis/table tennis video games are a design sweet spot for the kinds of simple, continuous spatial models that could be implemented with television technology. Many similar games from competing companies were launched in both the coin-op and home markets in the coming years. And it was Magnavox’s decision to sue these companies, Atari, and others that led to the rediscovery of Tennis for Two.


With the launch of top-down tennis/table tennis video games, the initial invention of playable models of continuous space (in continuous time) for video games was completed. Of course, significant inventive work remained to be done and would shape the future of games when it arrived. For example, the rise of the first-person shooter (FPS) genre, particularly with Wolfenstein 3D and DOOM, would only take place after the invention of the implementations that would allow for real-time 3D spatial models on consumer-level hardware. It is also important to note that the underlying technologies and design practices would benefit from long-term military investment—some of it indirect—as seen in this article with the prominence of BNL and MIT’s TX-0 lab, both of which depended on military funding.

My main interest here is not simply to retell the story of these games, which is often told. Rather, it is to tell it with attention to the specifics of the versions of the operational logics that were actually implemented for these early games, how they were implemented, and what playable models and gameplay scenarios these enabled. Doing this here has revealed that one of the most widespread game history stories—that complexity kept Computer Space from Pong-like success—is at the very least partial, and more likely simply mistaken. By engaging the limitations of the television technology with which Computer Space was implemented, we can understand why its physics logic lacked gravity. By looking at the importance of adding gravity to the physics logic of Spacewar,and how it only achieved compelling gameplay after this addition, we can understand that it was likely a lack of compelling gameplay that led to the fate of Computer Space. We can’t know for certain how the story of Computer Space’s overcomplication became so widely accepted, but we can speculate that Bushnell, the source most often cited, would have been happier to have Computer Space known for overestimating its audience than for being uninteresting at its core.

At the same time, this article’s tracing of gravity’s role in early spatial models for games has revealed something further. While gravity was part of the physics logic of the earliest known tennis/table tennis video game (Tennis for Two), the Sanders team established that such designs could succeed without gravity, through a change in gameplay perspective. Pong, the first hit video game, lifted this changed perspective from the in-process Magnavox Odyssey. In short, the story of Computer Space versus Pong is actually of one game design known to fail without gravity as part of its physics logic versus another game design known to succeed under such circumstances—the circumstances that faced all parties at the launch of the coin-op game era.

As a field, we need to start telling these stories differently. And it is my hope that we can also begin to use attention to the specifics of logics and models—and their implementations, even when these stretch beyond the comfort zone of software implementations—to reconsider other widespread stories about game history.


1. ^ David Sudnow, Pilgrim in the Microworld (New York: Warner Books, 1983).

2. ^ Sudnow, Pilgrim in the Microworld, 37.

3. ^ As Henry Lowood points out, Sudnow’s discussions of video games (and game controllers) not only echo earlier discussions from phenomenologists but also key ideas for computational media’s development in the 1960s. These include technological ideas, such as Doug Engelbart’s goals for the mouse and other human-computer interaction devices. But they also include more theoretical ideas, such as Marshall McLuhan’s characterization of media as extensions of the human senses. Henry Lowood, email to author, October 5, 2019.

4. ^ Sudnow, Pilgrim in the Microworld, 121–22.

5. ^ Noah Wardrip-Fruin, How Pac-Man Eats (Cambridge: MIT Press, forthcoming).

6. ^ This article is itself adapted from in-process writing toward How Pac-Man Eats.

7. ^ This is in contrast to other ways that the topic of space in games might be approached, without a focus on logics and models. To further specify what this article will discuss, it focuses on the development of two-dimensional spatial models, rather than three-dimensional ones. It also focuses exclusively on space as represented within the game, rather than wider spatial contexts (such as where players are located in the everyday world). And, finally, it also focuses on the specifics of how space is used in the space combat and tennis/Ping-Pong designs of the games it discusses. For treatment of other issues, there is a wide history of work on game space, much of which has focused on 3D space—though it also includes some treatment of special issues such as glitch spaces and the spaces of connection games. Some places to begin include: Janet H. Murray, Hamlet on the Holodeck: The Future of Narrative in Cyberspace (New York: Free Press, 1997); Espen J. Aarseth, “Allegories of Space: The Question of Spatiality in Computer Games,” in Cybertext Yearbook 2000, ed. Markku Eskelinen and Raine Koskimaa (Jyväskylä, Finland: Publications of the Research Centre for Contemporary Culture, University of Jyväskylä, 2001), 152–71; Mark J. P. Wolf, The Medium of the Video Game (Austin: University of Texas Press, 2001); Friedrich von Borries, Steffen P. Walz, and Matthias Böttger, Space Time Play: Computer Games, Architecture and Urbanism: The Next Level (Basel, Switzerland: Walter de Gruyter GmbH, 2007); Michael Nitsche, Video Game Spaces: Image, Play, and Structure in 3D Worlds (Cambridge: MIT Press, 2008); Celia Pearce, “Spatial Literacy: Reading (and Writing) Game Space,” in Proceedings of Future and Reality of Gaming (FROG) (Vienna, Austria, 2008); Jesper Juul, A Casual Revolution: Reinventing Video Games and Their Players (Cambridge: MIT Press, 2009); William Humberto Huber, “Epic Spatialities: The Production of Space in Final Fantasy Games,” in Third Person: Authoring and Exploring Vast Narratives, ed. Pat Harrigan and Noah Wardrip-Fruin (Cambridge: MIT Press, 2009); Dan Pinchbeck, Doom: Scarydarkfast (Ann Arbor: University of Michigan Press, 2013); and Nathan Altice, I Am Error: The Nintendo Family Computer / Entertainment System Platform (Cambridge: MIT Press, 2015).

8. ^ I am using the phrase “just so” story to describe a story that appears to explain what we observe but attributes it to causes that are of unprovable, partial, or false validity.

9. ^ While Computer Space is often referred to as a failure, this is only true in comparison to the massive success of Pong. As Benj Edwards writes, “The game sold fairly well for the first commercial video game—estimates range from 500 to 1,000 units—but it was no blockbuster.” Benj Edwards, “Computer Space and the Dawn of the Arcade Video Game,” Technologizer, December 11, 2011, 3,

10. ^ id Software, Inc., DOOM, MS-DOS, John Romero, John Carmack, Sandy Petersen, Tom Hall, Dave Taylor, et al. (id Software Inc., 1993).

11. ^ Raven Software, Heretic, MS-DOS, Brian Raffel, Ben Gokey, Chris Rhinehart, Shane Gurno, Steve Raffel, Brian Pelletier, James Sumwalt, Michael Raymond-Judy, Eric C. Biessman, Timothy Moore, Kevin Schilder, et al. (id Software, GT Interactive, 1994).

12. ^ Rogue Entertainment, Strife, MS-DOS, Susan G. McBride, Jim Molinets, Sean Patten, Nicholas Earl, Gary Lake-Schaal, Tim Willits, Michael Kaplan, John Sanborn, James Monroe, Peter Mack, Rich Fleider, Steven Maines, et al. (Velocity Inc., 1996).

13. ^ Digital Café, Chex Quest, Microsoft Windows 3.1, Dean Hyers, Mike Koenigs, Scott Holman, Davis Brus, Charles Jacobi, Josh Storms, Andrew Benson, Mary Bregi, et al. (Ralston-Purina, 1996).

14. ^ Bungie Studios, Pathways into Darkness, Apple System 6, Jason Jones, Alexander Seropian, and Colin Brent (Bungie Studios, 1993).

15. ^ Bungie Software Products Corporation, Marathon, Apple System 6, Jason Jones, Alexander Seropian, Ryan Martell, Alain Roy, J. Reginald Dujour, Greg Kirkpatrick, Colin Brent, et al. (Bungie Software Products Corporation, 1994).

16. ^ While the workshop took place in 2010, the report was published in 2012. See Tom Boellstorff et al., “The Future of Research in Computer Games and Virtual World Environments: Workshop Report,” ed. Walt Scacchi (Irvine, CA: Institute for Software Research, University of California, Irvine, July 2012),

17. ^ It is also possible to create playable models that are quite different, but are understood to procedurally represent a similar domain, as long as the key “communicative obligations” for a model of that domain are met (to use language Joseph C. Osborn developed, working with Mateas and me). See Joseph C. Osborn et al., “Combat in Games,” in Proceedings of the 10th International Conference on the Foundations of Digital Games (Asilomar, CA, 2015); and Joseph C. Osborn, “Operationalizing Operational Logics” (PhD diss., University of California Santa Cruz, 2018),

18. ^ While others may use similar phrasing (for example, Sarah T. Roberts’s discussion of the “operating logic of opacity” in “Digital Detritus”), I use the term operational logics to refer to a particular idea that I have been developing since 2005, largely in collaboration with colleagues at UC Santa Cruz. Some key publications in this development are: Noah Wardrip-Fruin, “Playable Media and Textual Instruments,” Dichtung Digital, 2005,; Michael Mateas and Noah Wardrip-Fruin, “Defining Operational Logics,” Proceedings of the Digital Games Research Association (Brunel University, London, 2009); Mike Treanor, Michael Mateas, and Noah Wardrip-Fruin. “Kaboom! Is a Many-Splendored Thing: An Interpretation and Design Methodology for Message-Driven Games Using Graphical Logics,” Proceedings of the Fifth International Conference on the Foundations of Digital Games (Asilomar,CA, 2010); Joseph C. Osborn, Dylan Lederle-Ensign, Noah Wardrip-Fruin, and Michael Mateas, “Combat in Games,” Proceedings of the 10th International Conference on the Foundations of Digital Games; Joseph C. Osborn, Noah Wardrip-Fruin, and Michael Mateas, “Refining Operational Logics,” Proceedings of the 12th International Conference on the Foundations of Digital Games (Cape Cod, MA, 2017); and Noah Wardrip-Fruin, “Beyond Shooting and Eating: Passage, Dys4ia, and the Meanings of Collision,” Critical Inquiry 45, no. 1 (September 1, 2018): 137–67. For more on Roberts’s concept, see Sarah T. Roberts, “Digital Detritus: ‘Error’ and the Logic of Opacity in Social Media Content Moderation, First Monday 23,no. 3 (2018),

19. ^ Osborn, “Operationalizing Operational Logics.”

20. ^ Anna Anthropy [Auntie Pixelante] and Liz Ryerson, Dys4ia, Adobe Flash, March 9, 2012, (no longer available).

21. ^ Strachey described this work in multiple publications, including at the 1952 ACM National Conference. But to my knowledge he never gave a name to the resulting game, which may contribute to its relative obscurity. I have named it MUC Draughts, a name I welcome others to use. See Christopher Strachey, “Logical or Non-Mathematical Programmes,” in Proceedings of the 1952 ACM National Meeting (Toronto) (New York: ACM, 1952), 46–49; and Christopher Strachey, “The ‘Thinking’ Machine,” Encounter 3, no. 4 (October 1954): 25–31.

22. ^ Stephen Russell et al., Spacewar!, Digital Equipment Corporation PDP-1 (Massachusetts Institute of Technology, 1962).

23. ^ For purposes of clarity, titles ending in an exclamation point (such as Spacewar!, Pitfall!, and Kaboom!) are given without their exclamation points after their first use.

24. ^ Atari, Inc., Pong, coin-op, Alan Alcorn (1972).

25. ^ We experience these models as continuous though, for modern games, at the level of implementation this is illusory, just as the continuity of movement in film is an illusion created by the use of many discrete frames. This distinction between discrete and continuous spatial models is different from the distinction between discrete screens of game content and continuous, scrolling screens of game content, as discussed by Clara Fernández-Vara, José Pablo Zagal, and Michael Mateas in “Evolution of Spatial Configurations in Videogames,” and Mark J. P. Wolf in The Medium of the Video Game. The importance of this distinction is certainly not unique to video games. For example, as Jon Peterson describes in Playing at the World, differing spatial models is one of the primary distinguishing factors between miniatures-based war simulation games (continuous space) and those based on boards and counters (discrete space). The distinction is also not unique to spatial models. For example, Ernest Adams and Joris Dormans, in their discussion of discrete and continuous mechanics, note that physics and timing mechanics are often continuous, while internal economies (considered broadly) are generally discrete. You generally cannot pick up half a power-up. See Clara Fernández-Vara, José P Zagal, and Michael Mateas, “Evolution of Spatial Configurations in Videogames,” in Proceedings of the 2005 DiGRA International Conference: Changing Views: Worlds in Play (Digital Games Research Association, 2005); Wolf, The Medium of the Video Game, 55–61; Jon Peterson, Playing at the World: A History of Simulating Wars, People and Fantastic Adventures, from Chess to Role-Playing Games (San Diego: Unreason Press, 2012); and Ernest Adams and Joris Dormans, Game Mechanics: Advanced Game Design (Thousand Oaks, CA: New Riders Publishing, 2012).

26. ^ Activision Inc., Pitfall!, Atari 2600, David Crane (1982).

27. ^ Of course, playable models of space need not fall neatly into one of these two categories. In fact, spatial models need not even be presented as images. For example, they can be organized as graphs and presented as text, as in the original Adventure and the tradition of interactive fiction. In such games players type commands to move between spaces (which may be connected to an arbitrary number of other spaces) and to take actions within spaces. In terms of the core spatial model, in such games jumping (the example mechanic discussed with the other games above) is an action performed in the undifferentiated space within a node of the graph.

28. ^ Rovio Mobile Ltd., Angry Birds, Apple iOS, Niklas Hed, Mikael Hed, Raine Mäki, Harro Grönberg, Jaakko Iisalo, Tuomo Lehtinen, Tuomas Erikoinen, Ari Pulkkinen, et al. (Clickgamer Technologies Ltd., 2009).

29. ^ William Higinbotham, Robert V. Dvorak, and David Potter, Tennis for Two, custom hardware, (Brookhaven National Laboratory, 1958), retrospectively named.

30. ^ Raiford Guins, Game After: A Cultural Study of Video Game Afterlife (Cambridge: MIT Press, 2014), 310.

31. ^ Guins, Game After, 97–98.

32. ^ Any identification of a first video game is predicated on some definition of video game. I prefer a definition that includes all games in which an electronic process (CPU-driven or otherwise) displays elements of the play environment (e.g., not simply a score number) to a screen (of any kind) in response to human play. Guins points out that this is far from being a universally accepted definition: “Many—those who prefer a technologically determined approach to history—argue that [Tennis for Two] wasn’t a ‘video’ game due to the lack of a CRT in the oscilloscope. Higinbotham discusses the oscilloscope tech in his deposition. To make matters even more confusing, the 1959 iteration of [Tennis for Two] used a CRT oscilloscope! I told Baer this during a Skype conference but he dismissed it outright” (Guins, email to author, October 6, 2019). The same objection could be applied to Strachey’s game.

33. ^ In this case, MUC stands for Manchester University Computer. Strachey’s love letter generator, which I believe is the first piece of electronic literature (and digital art of any kind) signed its creations “M. U. C.” See Noah Wardrip-Fruin, “Digital Media Archaeology: Interpreting Computational Processes,” in Media Archaeologies, ed. Erkki Huhtamo and Jussi Parikka (Berkeley: University of California Press, 2011); and Noah Wardrip-Fruin, “Christopher Strachey: The First Digital Artist?,” Grand Text Auto (blog), August 1, 2005,

34. ^ Higinbotham is generally credited with the game’s design and is also notable for being first chair of the Federation of American Scientists and a lobbyist for nuclear nonproliferation. He had previously done work (at the MIT Radiation Laboratory) on radar displays—the kinds of interactive screens that also inspired computing pioneers such as Douglas Engelbart. See Brookhaven National Laboratory, “BNL History: The First Video Game?,”

35. ^ The Vladar Company, When Games Went Click: The Story of Tennis for Two, concept and grant by Raiford Guins, Kristen J. Nyitray and Peter Takacs, script by Raiford Guins and Laine Nooney, May 24, 2013, Youtube video, 17:57,

36. ^ Guins, Game After, 272.

37. ^ The logics in this section were first discussed in my “Playable Media and Textual Instruments,” which did not formally name or define them. The full definitions were first introduced in “Defining Operational Logics,” by Michael Mateas and myself, but have since been revised. One key revision is the removal of the notion of a coordinate space when discussing collision. As Nathan Altice pointed out, many early games and platforms that clearly implemented these logics did not do so using a coordinate space. Other changes include the collapsing of continuous movement and physics into one logic, the categorization of navigation as a type of control logic, and the removal of the notion of hierarchical logics. All of these changes are inspired (for Mateas and me) by work with Joseph C. Osborn. See Osborn, Wardrip-Fruin, and Mateas, “Refining Operational Logics.”

38. ^ Nathan Altice, personal communication, June 30, 2017.

39. ^ In personal communication, Henry Lowood amplifies the importance of this point, writing, “This of course was the point argued over continuously during the various Sanders/Magnavox lawsuits. The little twist in Baer's argument (for the primacy of his patent) was the combination of collision detection and directional change of the object” (October 5, 2019).

40. ^ To be clear, I mean to indicate that Tennis for Two marks a still-incomplete version of this game design. I do not mean to suggest, however, that later games such as Pong were directly inspired by Tennis for Two, which was only rediscovered after the work of the Sanders team.

41. ^ These mechanics depend on a general control logic, but neither provides the specific experience we associate with the navigation version of this logic.

42. ^ Of course, gravity itself can be implemented at many levels of complexity and with varying relationships to everyday world phenomena. A more complex approach to gravity might use an abstract process in which all physical bodies attract one another. More complex approaches are uncommon, though not unheard of, in the spatial models of video games.

43. ^ Brookhaven National Laboratory, “BNL History: The First Video Game?.”

44. ^ Nick Montfort and Ian Bogost, Racing the Beam: The Atari Video Computer System (Cambridge: MIT Press, 2009).

45. ^ Brookhaven National Laboratory, “BNL History: The First Video Game?.”

46. ^ Further, as Guins points out, “this hardware was federal property and too expensive to keep in the configuration of the game. So after [19]59 it was disassembled and the various pieces of tech were used on other projects at the lab” (email to author, October 6, 2019).

47. ^ Theodor Holm Nelson, Computer Lib/Dream Machines (self-pub., 1974).

48. ^ J. M. Graetz, “The Origin of Spacewar,” Creative Computing, August 1981, 56.

49. ^ Computer History Museum, “Jack Dennis,” PDP-1 Restoration Project,

50. ^ On the TX-0, for example, interactive computing pioneer Ivan Sutherland had just been getting the feel of working with a display and light pen the previous winter—setting the stage for his groundbreaking Sketchpad system. As Sutherland puts it, while working on the TX-0’s display “the idea began to grow in my mind that application of computers to making line drawings would be exciting and might prove fruitful.” See Ivan Edward Sutherland, “Sketchpad, a Man-Machine Graphical Communication System” (PhD diss., Massachusetts Institute of Technology, 1963), 24,

51. ^ Digital Equipment Corporation, “PDP-1 Story,” Documenting DIGITAL, 1995–1998, (CD-ROM preserved by Gordon Bell); and Steven Levy, Hackers: Heroes of the Computer Revolution (New York: Anchor Press/Doubleday, 1984).

52. ^ Graetz, “The Origin of Spacewar,” 61.

53. ^ In video games that include combat, the combat models are most commonly designed to work with continuous spatial models (supported by graphical logics) and resource logics. For example, in Spacewar shooting torpedoes obviously requires movement physics and collision detection logics (together with a control logic for determining when to fire). Less obviously, there is also a resource logic involved. Though Spacewar does not treat the amount of damage that each ship can take as a resource except in the most basic sense (a single collision with a torpedo destroys the ship), the number of torpedoes a ship can fire is a time-limited resource. As Graetz writes, “There was a fixed delay between shots ‘to allow the torp tubes to cool.’” But this sort of observation does not give us a broader sense of how combat models operate in video games. And in fact such a broader sense is not often put into words, even if we know it when we see it—as Joseph C. Osborn, Dylan Lederle-Ensign, Michael Mateas, and I discovered when we explored this question. While it is generally taken as a given that combat is modeled by many games, we were able to find no discussions—in game design, game studies, or elsewhere—that tried to be specific about what this means across game genres. So we began with a close examination of how particular games implement combat, especially Super Street Fighter II and Final Fantasy. We were able to look at how each game implements and composes operational logics to form a model of activity that players interpret as combat, resulting in the paper, “Combat in Games.” See Graetz, “The Origin of Spacewar,” 62; and Osborn et al., “Combat in Games.”

54. ^ Supporting interaction is necessary, but not sufficient, for supporting playability.

55. ^ Spacewar is also notable for most likely being the first game to include all three of the key elements that Steve Swink has identified for “game feel.” Like Tennis for Two, it has Swink’s “spatial simulation.” The addition of the navigation logic contributes Swink’s “real-time control.” Finally, Spacewar has what Swink calls “polish”: “any effect that artificially enhances interaction without changing the underlying simulation.” We see polish, for example, when flames emerge from the back of the ships during thrusting. This is just a signal to the player and an enhancement of the visual presentation, as opposed to “interactions such as collisions, which feed back into the underlying simulation.” You can’t attack another player using the flames from your thruster. See Steve Swink, Game Feel (Boston: Routledge, 2008), 5–6.

56. ^ When it can’t do what we want, however, the identification tends to break down. I say things like, “I went into the room and tried to open the connecting door, without a key, but you can’t do that” (rather than “but I couldn’t,” which would have been the natural continuation if identification was maintained). Sudnow also describes how this identification does not exist before we learn how to play: “Then, with no sense that I was playing a game, not knowing who ‘I’ was among the various moving objects on screen, not even sure ‘I’ was there—without the slightest idea whether, why, or when whatever was happening would end, it was all over.” See Sudnow, Pilgrim in the Microworld, 5.

57. ^ Brendan Keogh, A Play of Bodies: How We Perceive Videogames (Cambridge: MIT Press, 2018), 14. For treatment of wider questions of embodiment, cognition, computational media, and the arts, see Simon Penny, Making Sense: Cognition, Computing, Art, and Embodiment (Cambridge: MIT Press, 2017).

58. ^ Graetz, “The Origin of Spacewar,” 64.

59. ^ For more on the shape of Spacewar’s gravity well (and its shifts between versions), see Norbert Landsteiner’s remarkable “Inside Spacewar!,” especially “Part 6: Fatal Attraction—Gravity,” 2014–2015,

60. ^ Graetz, “The Origin of Spacewar,” 64.

61. ^ In addition to Edwards’s contribution, Spacewar also benefited from Graetz adding a hyperspace capability (jumping players to a random location), Peter Samson adding a set of stars showing a section of the night sky as seen from earth, and Alan Kotok and Robert A. Saunders creating custom controllers.

62. ^ Graetz, “The Origin of Spacewar,” 66.

63. ^ Henry Lowood, “Videogames in Computer Space: The Complex History of Pong,” IEEE Annals of the History of Computing 31, no. 3 (2009): 7.

64. ^ Stewart Brand, “Spacewar: Fanatic Life and Symbolic Death among the Computer Bums,” Rolling Stone, December 7, 1972; and Stewart Brand, II Cybernetic Frontiers (New York: Random House/Bookworks, 1974),

65. ^ Brand’s article also describes some of the results of those who had created modifications and extensions to the original Spacewar in its first decade of release.

66. ^ Lowood, “Videogames in Computer Space.”

67. ^ While many accounts suggest Bushnell first encountered Spacewar at the University of Utah, Alexander Smith argues that this is highly unlikely: “So when did Nolan Bushnell first see the Spacewar! game? According to my own interview with Bushnell, when he relocated to the San Francisco area, he began attending several go clubs, as he had recently become fascinated by the game in his later years at the University of Utah. At the Stanford University go club, Bushnell met Jim Stein, who worked at the Artificial Intelligence Laboratory. In both our interview and the book High Score, Bushnell recounted how one day in 1969 Stein told him about the cool games available at the lab, where as we saw previously, Spacewar! was an incredibly popular pastime. Bushnell states that he told his friend that he already knew of Spacewar!, but would love to play it again. Note how this recollection so closely mirrors the story in his deposition that a friend with the first name Jim with whom he played chess told him about all the cool games in the Utah computer center. I believe there is a high degree of likelihood that Bushnell took the true story of how he was introduced to the game at Stanford and tweaked it to take place earlier at Utah instead in order to show that his ideas predated those of Ralph Baer.” Smith’s conclusion is in part based on research by Martin Goldberg, which Goldberg discusses at length in a blog post and also writes about briefly in note 13 of an article with Devin Monnens. See Alexander Smith, “The Book of Nolan,” They Create Worlds (blog), December 19, 2014,; Martin Goldberg, “Nolan Bushnell and Digging up Spacewar!,” Atari Inc.—Business Is Fun, January 11, 2014, (currently available at:; Devin Monnens and Martin Goldberg, “Space Odyssey: The Long Journey of Spacewar! From MIT to Computer Labs around the World,” Kinephanos: Journal of Media Studies and Popular Culture, June 2015, 124–47.

68. ^ Lowood, “Videogames in Computer Space,” 9.

69. ^ Lowood, 12.

70. ^ Mark J. P. Wolf, “Abstraction in the Video Game,” in The Video Game Theory Reader, ed. Mark J. P. Wolf and Bernard Perron (New York: Routledge, 2003), 49.

71. ^ John A. Price, “Social Science Research on Video Games,” Journal of Popular Culture 18, no. 4 (Spring 1985): 114.

72. ^ Montfort and Bogost, Racing the Beam, 7–9.

73. ^ Mark J. P. Wolf, Encyclopedia of Video Games: The Culture, Technology, and Art of Gaming (Westport, CT: ABC-CLIO, 2012), 81.

74. ^ Lowood, “Videogames in Computer Space,” 10.

75. ^ Steve L. Kent, The Ultimate History of Video Games: From Pong to Pokémon and Beyond: The Story behind the Craze That Touched Our Lives and Changed the World (Roseville, CA: Prima, 2001), 34.

76. ^ Tom Sito, Moving Innovation: A History of Computer Animation (Cambridge: MIT Press, 2013), 107–8.

77. ^ Raiford Guins, Atari Design: Impressions on a Cultural Form, 1972–1979, Cultural History of Design (New York: Bloomsbury, 2020).

78. ^ Mark Glaser, “Before the Big Bang: The Space Age Game That Set the Stage,” New York Times, August 9, 2001,

79. ^ Matt Barton, Vintage Games 2.0: An Insider Look at the Most Influential Games of All Time (Boca Raton, FL; Taylor & Francis, 2017), 9.

80. ^ Cinematronics Inc. and Vectorbeam Inc., Space Wars, coin-op, Larry Rosenthal (Amutech Ltd., 1977).

81. ^ Matt Barton and Bill Loguidice, “The History of Spacewar! The Best Waste of Time in the History of the Universe,” Gamasutra, June 10, 2009, 3,

82. ^ Ralph H. Baer, Videogames: In the Beginning (Springfield, NJ: Rolenta Press, 2005), 18,

83. ^ The Odyssey is a fascinating machine, with different games selected by the activation (or not) of different components on the board. In one version manufactured, these components can also be physically removed from the board allowing (as Nathan Altice demonstrated during a talk at an Expressive Intelligence Studio meeting) the games to be altered during play. For example, the removal of a particular component removes a game’s collision logic. The games are also designed to be both physical and electronic, with physical overlays meant to be placed on the television set and many games employing objects such as tokens, money, cards, and so on.

84. ^ Lowood, “Videogames in Computer Space,” 16.