The Unit Is in the Eye of the Beholder

The Unit Is in the Eye of the Beholder

Emily Short
Unit Operations: An Approach to Videogame Criticism
Ian Bogost
Cambridge: MIT P, 2006

Emily Short interrogates Ian Bogost’s Unit Operations and finds his approach to videogame criticism too capacious in its attempt to account for a variety of expressive media, and too narrow in its focus on low-order choices in videogames.

Rob Swigart:

We see this point in action in the comments on the game Budget Hero she mentions in the previous paragraph.

Rob Swigart:

Not to be confused with my interactive novel Portal (1986) from Activision.


Ian Bogost seeks to define a general category of “unit operation,” distinct from “system operation”: that which applies to or is performed by separate objects or entities. The idea of “unit operations” draws heavily on concepts from object oriented programming (where each object is responsible for encapsulating its own behaviors) and the study of cellular automata (where the simple behavior of individual objects produces complex emergent patterns). He goes further and also situates “unit operations” and the study thereof with respect to a range of philosophical outlooks on the nature of the world (such as realism versus nominalism) and literary-critical approaches.

This interdisciplinary approach is both attractive and slippery. A unit operation, it seems, can be almost anything done by or to an entity: it can be the effect of a gene in a biological system, or the action a player is allowed to perform in a game, or the act of waiting for something that is not likely to occur, or a mode of classifying a chance encounter in the street. Unit operations can represent our ways of dealing with society around us; thus, to articulate a new unit operation can be to introduce into culture a productive new form of behavior. Bogost even uses the idea of unit operations to try to address how game studies should be positioned in academic disciplines, agitating for interaction between the study of videogames and the study of other creative disciplines. This attempt to situate game studies among the humanities is in reaction to those schools of videogame study which are functionalist (thus preoccupied chiefly with the way games work rather than with their expressive potential) and exclusive. While Bogost may have a point here, trying to tell academic departments how to organize themselves is a little like showing cats a seating chart.

Any concept that can be applied to so many disciplines is either extremely powerful or too general to be of any practical use. Unit operations seem sometimes to be one, sometimes the other.

Definitions and Agendas

At the beginning of Unit Operations, Bogost writes that he intends to explain his terminology and his inspirations in a way that will be accessible to the (relative) layperson, but not stultifying to experts in the diverse fields of comparative literature, computing theory, and videogame studies.

To the extent that I can judge, he mostly succeeds, but I am not an expert on many of the philosophers he cites and will leave it to other reviewers to address how accurately Bogost characterizes their theories. Even in an intentionally summarized and simplified form, however, the material he has to present is diverse and challenging, and I doubt there are many people who would find this book a light read.

There were a few moments when his choice of references gave me pause, even within the limitations of my background. Bogost several times mentions the ideas of Stephen Wolfram in a way that implies uncritical acceptance of Wolfram’s findings on simple systems and cellular automata as the basis of a wide range of scientific results. Considering the controversy surrounding Wolfram’s work, I might have expected these references either to be defended explicitly or to be presented with something of a cautionary note.

Specific examples do not always clarify Bogost’s theoretical distinctions. For instance, he identifies structuralist theory as a prime example of system operations. Considering Saussure’s belief in signs defined, not independently, but by relation to one another, structuralism does indeed appear to focus on the relationships between things and in a holistic concept rather than in the discrete elements. Structuralism therefore seems to deserve the name of “system operation” as much as any critical theory can. When Bogost applies his theory of unit operations to interpretations of such texts as Grand Theft Auto III, however, he arrives at results that would not be out of place in a structuralist reading: the emphasis on the duality between legal actions and illegal actions, as articulated through a variety of different events in the game, is reminiscent of J. P. Vernant’s essentially structuralist reading of the Pandora myth, in which the actions of the major characters are aligned around principles of giving and receiving, deception and openness. The approach to the whole and the approach to the components turn out to be less distant than we might want to imagine.

Nonetheless, Bogost has an important point that he wants to illustrate. The humanities and the technical sciences have much to learn from one another, and they do not engage one another enough. By looking at ideas from multiple disciplines, we can find ways for humanists to address technical issues, and technical scientists to think about humanistic questions.

Videogames, being one of the most procedural forms of art yet developed, are an obvious meeting point.

Narrative Progression

I could not help feeling that some of critical passages only partly proved the point that Bogost wanted to make. Invoking Stephen Spielberg or Charles Bukowski sometimes supported the legitimacy of his approach across media, but also sometimes showed how the theory of unit operations was not quite sufficient for understanding such products as films and poetry. His reading of Spielberg’s The Terminal explains that the narrative of the film is unsuccessful because the plot is not the point; Bogost writes:

Analyzing an artifact like The Terminal as a unit-operational film about themes of waiting rather than a system-operational film about the story of a handful of developed characters thus demands a novel critical framework. In my unit analysis of the film, the story serves as the glue for a configurative work about specific modes of uncorroborated waiting. This approach is quite different from the inverse, an analysis of the story of Viktor, Amelia, Dixon, and others with common touch points in the common theme of waiting. Such a distinction is core to the critical process of unit analysis, which privileges discrete components of meaning over global narrative progression. (19)

If we take theories of criticism as a way to find valuable meaning in apparently opaque texts, then the unit operations approach to The Terminal is a success. Certainly it is possible that some works - whether films or poems - make the most sense when understood through “discrete components of meaning.” But, conversely, some other works are very much reliant on narrative progression: what of them? Does unit operations theory claim that narrative progression is not (or should not be) desirable or relevant? Or is Bogost simply admitting the limitations of this critical approach?

My second objection has more to do with the rhetoric than the substance of Bogost’s idea. Advancing a plot-blind approach as the most productive route to videogame criticism looks like an admission that games cannot contain structured narrative to the same degree that other media can.

This may in fact be a just conclusion. Personally, though, I hope it is not true, and I do not consider that conclusion proven. Bogost does not argue against the story potential of games - indeed, in his remarks on the narratology vs. ludology debate (71), he suggests that games might indeed be able to produce stories. Nonetheless, the unit operations approach implicitly de-emphasizes narrative, and perhaps makes some claims about what we can and should expect from games.


Despite my reservations (about the clarity of the definition, about his intellectual influences, about the risks of dismissing “global narrative progression”) Bogost’s theory passes the most important of tests: it is useful.

Bogost uses the idea of unit operations to talk about the persuasive power and essentially biased nature of all simulations, whether they are intended as games or as serious models of the world, as for instance the budget models used by the Congressional Budget Office. (These categories can in fact sometimes collide: consider the recently released Budget Hero, which uses CBO predictions in order to direct a simple budget-policy simulation.)

Here Bogost applies “unit operations” not just to listing what the player is allowed to do, but to describing the sorts of events that the simulation accounts for; he argues, again valuably, that a simulation’s bias is often characterized as much by what it leaves out and chooses not to simulate, as by what it includes.

I found Bogost’s arguments about simulation and persuasion especially compelling. I would have liked, in fact, to read more about how he understands the persuasive functions of simulation, but I assume that for this I would have to turn to his other book, Persuasive Games, which I have not yet had the opportunity to read.

Social Communication

Bogost’s analysis of Grand Theft Auto III points out how the game’s apparent freedoms are nonetheless constrained by the fact that all actions are aligned according to their legality or criminality. The player is permitted to choose any of a wide set of activities, but the activities are nonetheless constrained to a certain range (primarily things that can be done with or to vehicles) and - perhaps more importantly - interpreted according to the set of oppositions relevant to the game.

Here Bogost makes what seems to me a very important observation. We can use unit operations to describe the various things that the player of a videogame is allowed to do (e.g., to steal a car, to fire a gun). The greater the number of available options, the greater the impression of freedom for the player. But it is the interpretive framework the game assigns to the player’s actions that generates the sensation of agency.

Many critiques of videogames, and many approaches to game design, have failed to recognize the distinction between what the player is allowed to do (namely, the specific actions he is allowed to perform using the interface of the game) and what he is allowed to mean (that is, the framework of significance the game places on those actions).

Chris Crawford points out the importance of the player’s operations with his emphatic question in Chris Crawford on Interactive Storytelling, “What does the player do?” No game design project can be approached, and no sensible critique of a game offered, unless we look at the actions - Bogost might say unit operations - that are available to the player. But Crawford tends to equate the number of verbs available to the player with the scope of interactivity in the game as a whole, making no distinction between the discrete actions and their narratively-framed meaning. The assumption in his Erasmatron/Storytron work seems to be that, in order for the player to have meaningful social interactions with other characters - in order for the player to express social intentions towards other characters - he must be provided with an equally wide range of social verbs:

The verb is the core of all interaction. Any piece of software has verbs in it… The classic verbs in a shooter are turn right, turn left, move forward, move back, and fire. That’s five basic verbs, and you’ll find those five basic verbs in every shooter, and then there will be another dozen ancillary verbs and things like move faster, run, jump, aim high and low, that kind of thing. But the whole trick in all of these games is to reduce the verb set. For example one example in the great majority of shooters is the joining of the verb “pick up” with the verb “go.” In other words, you don’t stop and pick something up, all you do is step over it and that picks it up… So the verb “go” ends up handling an awful lot of other verbs implicitly. So you end up mapping a lot of verbs into a kind of spatial reasoning, and that in turn keeps the verb count low and game designers like that, game players like that. The problem is, with social interaction, you just can’t get away with a tiny verb set, you need hundreds of verbs for social interaction. – Gamasutra interview with Chris Crawford

For Crawford, the relation between social actions and social meaning appears to be roughly 1:1. This is partly tied to the fact that Crawford wants to avoid imposing a specific narrative framework on the player any more than he has to: he wants a story and relations to emerge from a larger story world. All the meaning, for Crawford, has to come from the player’s unit operations, and that requires that they be very numerous and the simulation to handle them very complex.

Andrew Stern and Michael Mateas’ Façade goes even further, assuming that in order to give the player the feeling of genuine freedom in a social situation, the game must interpret all the gestures and words the player might choose to make. In fact, of course, Façade cheats: it maps the infinity of possible input texts to a finite range of social actions, then interprets these actions in terms of the narrative context. But it does its best to conceal the discreteness of player actions from the player himself. Here the mapping of possible acts (typing in any dialogue you like) to possible meanings (support, disagreement, flirtation, insult, etc.) is many to one. Because Stern and Mateas do make as much use as possible of the story context, the result does still sort of work - but a common complaint is that, despite the infinite range of accepted input, the story/game feels hard to steer, unpredictably responsive to player input.

These approaches might at first seem to be necessarily right. After all, how can the player have more expressive capacity than he has verbs (or actions or unit operations) available?

But there are games in which precisely this inversion occurs. Adam Cadre’s peculiar text-adventure-cum-golf-game, Textfire Golf, allows the player to play a game of skill against his golfing buddies, who are also his work colleagues. The game then extrapolates social implications from how the protagonist does relative to the other characters. Gamelab’s Miss Management hints at significant relationship implications behind the player’s choice of how to allot her time and which characters to favor in a time management game. The actions available to the player here are, respectively, swinging at a golf ball, and choosing to assign tasks to specific characters in an office-management simulation. The meanings of these actions, however - both for the player and for the game - have to do with interpersonal politics and power plays. The same idea translates into the realm of board games. One player’s order to move troops in Diplomacy might - depending on the evolution of alliances and promises up to that point - be a stalling tactic, a surrender, a betrayal, a warning, a feint, a deception, the first act of aggression in a long-expected war. The mechanics of action in Diplomacy are few and simple. The context for those actions is left to the players to devise as freely as they want, and therefore can become extremely complicated.

The examples here are mine, but the observation about complexity arising from simple operations is one of Bogost’s central tenets. Unit Operations offers a critical approach that encourages us to address the range of player actions separately from the expressive function those actions assume.

In Saussurian fashion, Bogost sees the meaning of player action as determined through the contrast with other actions that are also available in the game but which the player chooses not to take:

In GTA, every decision both includes and excludes another possibility, and thus choosing to drive an ambulance instead of bludgeoning a passerby for some cash to buy a new handgun develops fluid meanings that signify in relation to other possible unit operations. (168)

There are other strategies for assigning meaning to a player’s actions: thematic framing (as seen in GTA’s focus on the legality and police animosity) and narrative circumstance (whether he is currently performing a mission requiring a given act, for instance). These are ideas that Bogost touches, in varying ways, at a number of times during the book: his analysis of the morality of Deus Ex imposed on the operations of a pre-existing engine (63) is also useful in this regard.

Now it might be said that Bogost’s argument is merely a variation of the observation that games allow limited action sets that take on different significance depending on the tactical and strategic circumstance: to place a black Go stone on the board, or to move one’s queen to a certain chess square, are also actions which derive meaning from the context in which they occur. But Bogost is addressing cases in which there is no strategic gain to one move over another - where the player’s choice is not about which action is currently most likely to advance him towards a winning outcome, but about which action currently best expresses what he wishes to express. Instead of the complexity of the abstract game, we have the complexity of theme, the complexity of narrative.

The implications of this narrative complexity are valuable not only for the study of games but also for their development. I often see designers (often, but not always, amateur designers) who want to give the player expressive power by offering him as many operations as possible. In practice, these experiments are rarely as satisfying as the designers hope they will be. Players find the excessive number of options confusing, even numbing. For the authors, the task of treating all the many possible actions as meaningful becomes overwhelming, demanding excessive content production.

Tauter and more effective design arises from the disciplined use of a smaller set of acts. Even GTA III’s famous sandbox freedom actually arises, as Bogost points out, from a very restricted set of possible operations; it’s just that the player is given a large number of objects to act on, and an interpretive framework in which the same activity (such as taking a vehicle) has very different meanings at different times.

The trick is that we find it very difficult to reduce social communication to a small set of moves in the same way that we have learned to reduce running and shooting and picking things up. The successful experiments along this line have mostly been the ones that do not try to tackle the problem in abstract terms. These ask the question: in a given context, how do we convey (and modify) relationship status? By intentionally losing to the boss at golf? By making time to talk to another character during a busy office day? It is in the narrative context that we discover the unit operations necessary for social communication.

Final Thoughts

Ultimately, I found myself sufficiently persuaded by Bogost to wish he had offered many more readings of videogames based on unit operations - not only because such readings are productive in themselves, but because they might have exposed areas where his theory bears further elaboration.

Bogost makes the point early on that unit operations can occur at different levels: that, for instance, the unit operations of a frog’s body all go to make up the frog system, but the frog itself might be considered a unit in the larger system of a pond ecology.

When it comes to videogames, the specific actions available to the player (which Bogost is primarily concerned with in his videogame analyses) would then be only one kind of unit operation: in Grand Theft Auto, taking a car; in Portal, shooting an orange or blue portal, moving, pressing buttons, or picking up and dropping objects.

Bogost spends little time on the possible higher levels of unit operation in videogames. Portal is a particularly convenient example because it goes so far as to label these unit operations and explicitly teach them to the player: flinging oneself across space, dropping objects onto distant enemies, putting things into an incinerator, etc. These operations are made up of the smaller acts (shoot, move, press, take, drop), and much of the puzzle fun of the game is working out which of the higher-level unit operations is currently useful and then discovering how to execute it as a sequence of the smaller operations. Portal is thus made up of (at least) two levels of player unit-operation, and this is part of what makes it both challenging and satisfying.

The matter is further complicated because Portal is also arguably about a unit operation. In the way that The Terminal is about waiting, Portal is about one intelligent being disposing of another. This unit operation is strongly associated with the operation of putting things into an incinerator, but it is not identical, since there are times when the intelligence-disposal operation is mentioned, threatened, or hinted at without reference to object-incineration, and since the player is not always the agent.

What might unit operations theory tell us about the aesthetics not just of the lowest level actions (shoot, etc.) - which Bogost argues should be as expressive as possible - but of the higher ones? What might we say about the parallels between levels, as between “drop something into the incinerator” and “dispose of intelligent being”? Can we develop an aesthetics that might help us distinguish between a good/effective/expressive unit operation and an inferior one, or is that question even meaningful?

There is more that could be said here. To my mind, exploring these questions further - especially via specific critiques of a wider range of videogames - would have bolstered the argument for which Bogost laid such a wide-ranging theoretical foundation.