> older writing > essays

Combat in Context

Nick Montfort

Form, Culture, and Video Game Criticsm
Princeton University
6 March 2004

Note: This talk was published with a few changes as "Combat in Context" in Game Studies 6:1.
I am restoring this text to the Web since I originally made it available at this location,
but you should read and cite the Game Studies version, which is both definitive and illustrated!
(nm 2008-04-20)

What I offer here is a critical consideration of Combat, the cartridge originally included with the Atari Video Computer System. Atari introduced the VCS in late 1977; it retailed for about $200 (in today's dollars, roughly $650). The console, model number CX2600 [1], came with two joystick controllers, two paddle controllers, a TV/game switch box, and a cartridge bearing product number CX2601 [2]. This was Combat; a "game program" with 27 games, according to the included manual: tank, tank-pong, invisible tank, invisible tank-pong, bi-plane, and jet (Atari 1977). It is an important cartridge to consider for several reasons:

Contemporary Combat

At E3 in May 2000, Harmon Leon of the now-defunct gaming site set up a booth that featured an Atari VCS with a Combat cartridge plugged into it. The unit was wrapped in aluminum foil. Leon announced through his megaphone and using a hand-lettered sign that this system was actually a new game called CyberBattle 2000; he enticed several conference attendees to play it (Leon 2000). A humorous review of CyberBattle 2000 appeared in after the conference, which read, in part:

[I]t's refreshing to see a videogame that pares down creativity, revealing the very essence of gameplay. The graphics engine, designed by ex-Rare and id programmers, manages to recreate the feeling of a battlefield, without overwhelming the player with unnecessary distractions. The glowing battlecraft reveal a subtle design ethic unmatched by any similar title. ... Musical influences have been borrowed from many sources, most notably Leonard Cohen and Philip Glass. The result is an ambient simplicity with a Mooglike analog vibe. ... that ... brings us to the plot. — a plot so delicately strung that absolute attention must be paid as it unwinds delicate threads of intrigue and suspense. The twist at the end of the game is stunning. Cyberbattle 2000 revels in the fact that no other game will ever achieve this perfect balance between simplicity and style. (O'Connor 2000)

Today, many remember Combat fondly and some still play it once in a while, even when not preparing academic talks about it. While the review was mockery (mainly of video game reviewing), there are plenty of sincere contemporary reviews that are quite positive. One reviewer calls the cartridge a "true classic" and notes that "While the graphics can only be described as grotesque, Combat has great gameplay. ... Besides Pong, Combat may be the ultimate two player game" (Langberg 2003). Another writes "This is the Charlie Chaplin of the gameworld ... Looks crap but is really, really good." (Bolton 2000). There is dissent — one reviewer writes, "I can't honestly say [it] has stood the test of time too well, not even as a two-player game" (Flying Omelette 2003) — but most of those who have written about Combat recently have praised it.

Oddly, Combat has barely been discussed at all by critics or historians of the video game. Scott Cohen's book, a history of Atari, does not mention Combat at all (Cohen 1984). Supercade: A Visual History of the Video Game Age, 1971-1984, includes only a one-inch-tall screenshot and the note that it "featured several variations of Atari's popular arcade game, Tank" (Herman 2001). I hope my analysis will begin to fill the gap in the current literature.

Levels of Combat, Contexts of Combat

To analyze Combat, I introduce a model which distinguishes five levels surrounded by cultural and social context. This model is directly based on the seven-layer model introduced by Lars Konzack (Konzack 2002), but I believe it refines Konzack's model in a few useful ways. I'll emphasize that I see the cultural and social context (which accounts for Konzack's seventh layer) as surrounding and interacting with a game at all five of these levels, not just at the top. From bottom to top, I consider:

  1. Platform, the system on which a game runs. I mean basically the same thing here as does Konzack when he identifies "hardware" as his first level. However, a game that runs on Windows XP would have that operating system, not just Intel-compatible hardware, as its platform.
  2. Game code, the computer program that realizes a game [3].
  3. Game form, the simulated world and rules of the game. Two different pieces of source code that have different compiled representations could both realize games that function and play identically. This level and the next draw a different distinction than do Konzack's layers of "functionality" and "gameplay"; those two are distinguishable in some ways, but are much more difficult to abstract apart and seem less interesting to consider separately.
  4. Interface, which sits between the player and the game form, connecting them. Chess games with identical rules and identical simulated opponents might present different interfaces: one a speech interface, one a 2D view of the board, and another a 3D view [4].
  5. Reception and Operation, accounting for what the player does in interpreting the meaning of the game; connecting the experience of the game to other experiences of gaming, art, literature, or life; and actually playing — doing something to interact with the game via the interface. This level corresponds approximately to what Konzack distinguishes as "meaning" and "referentiality," two concepts that really cannot be abstracted apart in the way that an operating system and an application program can.

Earlier levels enable and constrain what happens in later levels. Players can only assign meanings to images if there are images presented in the interface, which can only happen if the game form affords something to present using images, which can only happen if that game form is realized in code, which can only happen if the code runs on a platform — specifically, a platform that supports graphics.

Creating Context: For Two Players

"27 video games" is a notable phrase on the cartridge label, but so is "for two players." Combat requires two players, like the Pong games for home computers before it, like the arcade Pong, like Spacewar as developed at MIT in 1962, and like William Higginbotham's analog computer tennis game shown in October 1958. Of course you can run Combat with one player, or with no players, for that matter, but this hardly argues against it being a two-player game. The cartridge continued a social and competitive tradition in video gaming, and by being bundled with the VCS originally it helped to assert that play involving several people was the ordinary use of this system. (Atari expressed this idea in other ways, for instance, by running a TV commercial that portrayed additional generations of a family joining in to play on the VCS, eventually resulting in a huge family crowd watching grandma and grandpa play.) Today, video game consoles are sold with a single controller. This is no doubt done for revenue-enhancing reasons, but it certainly signals that something is different about the context of play — perhaps that the two-player capability no longer needs to be overtly indicated, or perhaps that two-player games really are options, and not the norm.

It is very difficult to decide what genre Combat fits into, probably because any genre assignment would be anachronistic. Video games of different sorts did exist in 1977, although home video gaming was pretty much an all-Pong experience at that point. The genre distinctions we see today developed later, however, as game forms became refined and conventionalized and as greater visual representational power was developed. But I'll focus on Combat rather than speculating here on which of these two factors was more important in the development of genres.

Platform: Racing the Beam

Since Combat ran originally on the Atari VCS I'll consider this platform, recognizing that we can also play it today on compatible systems such as the Atari 2600 Jr. or Atari 7800, or for that matter in an emulator. The VCS (pictured in the handout) was designed to be situated on or near a television; the wood-grain plastic front of the case makes it more like a piece of furniture or a stereo component than like some piece of hobbyist electronics, such as a HAM radio or one of the Altair microcomputers that had already been released by 1977. The two joystick controllers, each with one button, are used in Combat. The original VCS has six switches in front, including left and right difficulty and a game select switch, used to choose from the 27 games. In Combat, the difficulty switches change the range of fire and, in the plane games, change the speed of flight. Some configuration — but not much — was required to switch the TV over to VCS input, to insert a cartridge, to turn the VCS on, and to select a game to play. The system was ready for play fairly quickly — there were no long load times, as with some CD-ROM based console games. Playing games on the VCS was an activity that could be easily mixed among other living-room experiences such as watching TV and playing board games.

The main processor of the VCS is the Mostek (MOS Technologies) 6507, a low-cost version of the already low-cost 6502, the chip that powers the Commodore PET. This processor runs at 1.19 MhZ, as does the custom graphics and sound chip, the Television Interface Adapter that was designed by Atari's Jay Miner. The TIA provides for five moveable objects, control over the background color, and stationary graphics. The VCS could directly address no more than 4KB of cartridge ROM, although bank-switching allowed larger cartridges to be developed. The system has 128 bytes of RAM — not 128KB, but 128 bytes, due to the extraordinary cost of RAM at the time. Keep in mind, though, that games ran from the cartridge ROM; Combat certainly was not loaded entire into this 128 bytes of RAM. This storage was only used for variables reflecting the game state, such as the players' score and the current bearing and velocity of their warcraft.

The VCS is considered an extremely difficult platform to program for, not only because of the austere system resources I have described, but also because of the amount of video RAM that is available: none. Programming the Atari VCS means "racing the beam." A program running on the VCS cannot write to a screen buffer, as was the case, for instance, on all home computers of the era, but must proceed in sync with the drawing of the image on the television, generating each scan line as the electron beam crosses the screen to draw the current one. One modern-day VCS programmer writes: "I've never heard of an Atari VCS game that crashed and now I know why. There simply isn't room for the kind of sloppiness that allows a crash-level bug to slip through. If you're that sloppy, it won't even draw the screen." (Williams 2001) One other feature of the platform is notable: The background is defined by a mere 20 bits that can either be duplicated or mirrored for the second half of the 40-pixel line. This is what accounts for the horizontal symmetry of the playing field in Combat.

Game Code: Assembling Combat

Combat was developed from a core tank game (similar to the 1974 Kee Games Tank for the arcade, but without the mines that were part of that game) into a game program that offered advanced tank variations as well as bi-plane and jet games. The program was written in assembly language, as was always the case on the Atari VCS. The early work was done by Joe Decuir as he was working on the hardware design of the Atari VCS with Miner; development was completed by lead cartridge developer Larry Wagner (Decuir 2003). The cartridge designers of early VCS games were never even identified by Atari, and certainly were not prominently credited. Few people today, even among 8-bit programmers and Atari VCS fans, know who the programmers of Combat are.

Critics have sometimes shied away from making comments on the code level because they do not have access to the source code of the game or because they lack programming expertise in the language the game was written in, but in such cases much can still be said about how programming practices, tools, and languages influence the development of a game. Combat was fashioned to fit in 2KB of cartridge ROM. To give you an idea of how much information this is, I have included the entire Combat program, in machine code, on the handout. To give you some idea of what the source code was like, I have provided, on the other side, a snippet of a disassembled and commented version of this machine code, which should give you an idea of what Wagner was looking at and working on as he finished Combat.

Although the VCS was exceedingly difficult to program, understanding a program like Combat — particularly a thoroughly commented program like this disassembly of Combat — is actually not difficult, and does not require a computer science degree or a college education of any sort. Assembly language for the 6507 provides a very small number of primitive operations, and programs mostly load values from memory, store values to memory, and do some simple arithmetic. I've really included the machine code and a sample of the assembly code on the handout not to bewilder or impress you, but to show that there is not a tremendous amount to understand here, and that the complete game code can be fully comprehended by dedicated scholars.

The code and its context suggests the following: The extreme difficulty in devising a good Atari VCS program is probably part of what led programmers to first create an enjoyable game with simple elements, and then to build on this by allowing 27 substantial but straightforward variations on these elements. Simple as Combat may be, the most basic structures of the game (shared by all variants) occupy a good bit of the cartridge's 2KB by themselves. Creating additional variants was a more feasible option, once the timing for a single variant had been worked out, than was enriching the experience through additional visual detail, by making subtle rule changes that are only seen in special cases, or by creating some extreme variation in game elements.

Although a similar pattern is seen in other early cartridges such as Decuir's Video Olympics, it wasn't the rule in later Atari VCS game programs. After programming tools had been further refined, more memory was available to cartridge programmers via bank-switching, and a host of tricks and techniques were available in existing examples of code, visual detail rather than game variations would be added as a way of enhancing some games — it can be seen in Off the Wall and Road Runner, for instance. So, the approach of Combat is not inherent to the platform, but results from the nature of the platform in the early context of VCS programming.

Game Form: 27 Video Games

You can see from the chart provided on the handout (an ASCII representation of a chart in the manual) what the 27 games on the Combat cartridge were. In a very offhand way, I could justify calling these different games by observing that some people are skilled at certain ones but can easily be defeated when playing others. There is also a formal justification. As the chart indicates, important formal elements of these games are varied: the type of fire is either a straight missile, guided missile, or machine gun; hits can be direct, off the walls, or either; the battlefield can be open, one of two mazes, or a wrap-around sky with clouds; and, reading down, the nature of the vehicle (including its visibility and steady-state velocity) can vary as well. To further increase interest, five of the plane games provide an asymmetrical contest in which one player has multiple planes that fly locked together; they fire multiple missiles but can also be easier to hit.

Just as Espen Aarseth identifies Quake II as being a way of playing games such as capture the flag and deathmatch, rather than being a single game itself (Aarseth 2001), Combat is a game program that offers 27 games. These are arranged thoughtfully, in categories and also with a simple variant first. Game 1 is fun and a good introduction to Combat, but games after it advertise themselves visually as more complex and worth playing. Combat provides a good way for game critics to see how difficulty and enjoyability can vary when only one element, or a small number of elements, change.

Interface: Direct Connection

I'll say very little about the interface. Giving or withholding information from the player is part of game form (as the invisible tank games demonstrate); the interface only deals with the different ways that the same information may be presented, and the different ways that the same type of control information may be input by the player. But there were interface choices to be made, as a hack of the Combat cartridge demonstrates. Paul Slocum's Combat Rock cartridge functions identically to Combat visually and in terms of game form, but all the sound effects in the original cartridge are replaced by a soundtrack; the cartridge repeatedly plays "Rock the Casbah" by The Clash. This does highlight something interesting about Combat: that the interface provided is almost purely devoted to conveying information about the game form in as direct a way as is possible, without decoration.

Reception and Operation: Death and Stories in Combat

At the top level, that of reception and operation, I'll consider some questions that might sound amusing: Is Combat violent? Is Combat narrative, or dramatic? What story does it tell? These last two are the sorts of questions that were asked by the reviewer, but I will at least attempt to take all three of these questions seriously.

Combat is violent; it depicts warfare and death, albeit in a very abstract way. The platform, the Atari VCS, did seem capable of representing violence that is offensive to some people, as the 1983 Wizard Games title The Texas Chainsaw Massacre demonstrated. But the violence in Combat is not only very stylized visually, it is also formally impermanent: your tank or plane is only disabled for a moment when it is hit, in a way that resembles a game of tag more than it does warfare. Your vehicle is not even destroyed and regenerated elsewhere on the screen, a less extreme form of resilience. This may go some of the way toward explaining why people generally look and act like they are having fun as they play Combat, and do not look or act like they are murdering anyone. Interestingly, bystanders in Grand Theft Auto: Vice City recover from being killed much as do the tanks and planes in Combat, although this is seldom pointed out in discussions of that game's violence.

I don't think it will prove very useful to view Combat from a narrative standpoint, although the program does depict events that are connected by time and causality. I've managed to avoid referring to movies and novels so far, and it wasn't very difficult — on the contrary, it would have been rather hard to relate Combat to Tank Girl or the silent film Wings in any useful way. Combat might somehow be tortured into confessing a story, but what it mainly does is facilitate a particular type of competition, providing an apparatus and context for play, a computer-mediated way of interacting with another player, rather than a digital vehicle for a message. It is difficult to see the game program as related to a war movie — the illustration on the cover of the manual might be related, but that is part of a paratext. On the other hand it is easy, and helpful, to see a relationship to billiards in the tank-pong games (particularly the ones in which an indirect hit is required) and to see the relationship to games like target shooting, catch, and tag.

It's also useful to recognize that Combat is related to early recreational computing, including playful programs that aren't video games, and to programs written for systems with very limited resources. Just as the extreme, formal difficulty of composition suggests some connection between the Oulipo and the troubadours, there is a link between Combat and the programs developed on the stripped-down version of the TX-0, the first fully transistorized computer, programmed in MIT's computing department in the late 1950s.

New Questions

Aarseth writes that aesthetic studies of games may have become possible because video games "unlike traditional games or sports, consist of non-ephemeral, artistic content (stored words, sounds, and images) which places the games in question much closer to the ideal object of the Humanities, the work of art" (Aarseth 2003). It's difficult to imagine that the pure representational ability of a cartridge like Combat is any greater than that of a deck of playing cards or an intricately carved chess set. What is different is not the content of Combat, but the fact that it is an interactive computer program, allowing new sorts of complex simulation and play, and this is, or should be, the critical factor that enables new sorts of detailed study and analysis of video games.

I will close with what I hope are more useful questions to ask of Combat and of other video games: What other games, game forms, and game elements do they draw on? What player skills from other games will transfer to this game? How exactly is the game played, with how many people, doing what sorts of things? Also, what programming language was the game written in? How many people were involved in designing and producing it, and what earlier code was re-used? How did it play with or against the platform on which it was implemented and the history of programming on that platform?

There are certainly games with cinematic connections (e.g., Karateka, House of the Dead), dramatic connections (e.g., Fašade), and literary connections (e.g., The Hobbit, Varicella). These may yield to related sorts of analysis, and the most interesting games, if they are to be fully understood, will certainly require such analysis. Still, when we consider video games overall, the two things that they all have in common stand out: they are games, and they are implemented on computers. The first and most essential connections we seek, and the most important forms and contexts that we consider, should be those related to gaming — on and off the computer — and to computing.


[1] The unit came to be known as the "Atari 2600" only after the release of the Atari 5200 in 1982.

[2] The cartridge was not was not included with the VCS throughout the entire retail lifetime of the console, but it was made available through other channels, for instance, as a packaged retail cartridge from Atari. The Sears Telegames packaging of the game named it Tank-Plus. Zellers, a Canadian department store, sold copies of Combat called Frontline.

[3] Special-purpose hardware that ships with a cartridge game — as was the case with the keypad that came with the Atari VCS Star Raiders — poses a problem. It should probably not be included at any point beneath this level, since it is specific to the game rather than generic to the platform. But different hardware that looks the same to the computer might be used in different editions of a game, without changing the game code.

[4] The way a game is packaged and distributed (in a cabinet in an arcade, or online, or in a shrinkwrapped box) might also be considered at this level, although these aspects can of course be varied independently of the programmed interface; these aspects could be treated as part of the context, but are probably better understood as being paratextual. Some of the difficulty here and with special-purpose hardware comes from dividing the levels of analysis along formal lines, which is certain to give short shrift to important material aspects like these. Perhaps a similar multi-level model of material analysis could be developed to address this.


— nm, 2004-03-06