Why Gamers Are Not Performers

I argue that even if video games are interactive artworks, typical video games are not works for performance and players of video games do not perform these games in the sense in which a musician performs a musical composition (or actors a play, dancers a ballet, and so on). Even expert playings of video games for an audience fail to qualify as performances of those works. Some exemplary playings may qualify as independent “performance‐works,” but this tells us nothing about the ontology of video games or playings of them. The argument proceeds by clarifying the concepts of interactivity and work‐performance, drawing particularly on recent work by Dominic Lopes, Berys Gaut, and David Davies.

There has been significant recent interest in video games as an aesthetic or artistic phenomenon. Grant Tavinor's view of their art status is clear from the title of his groundbreaking book, The Art of Videogames; Dominic McIver Lopes ends his recent book, A Philosophy of Computer Art (), with a chapter suggesting that we might think of video games as one genre of computer art; and in A Philosophy of Cinematic Art (), Berys Gaut essentially argues that (many) video games are interactive films and thus have as good a claim to art status as ordinary films.1 I take no position here on whether video games are art.2 My interest is rather in what kind of artistic activity video‐game players are engaged in and what kind of artworks video games are, if video games are artworks at all.3 In particular, I argue that typical video games are not works for performance; players of video games do not perform these games in the sense in which a musician performs a musical composition (or actors a play, dancers a ballet, and so on).

One difficulty that arises in discussing video games is the sheer diversity of things that go by that name. For instance, I do not mean to deny that there could be, or even are, things reasonably called video games that are works for performance or that the players of those games are performers, a point to which I will return at the end of the article. What I deny is, first, the quite general claim, defended by some, that merely in virtue of their interactivity, video games are works for performance and gamers performers and, second, that typical video games meet whatever further conditions are necessary for something's being a work for performance. Obviously, this requires getting clear on the two central concepts of interactivity and work‐performance. But do we also need to get clear on the concept of video‐gamehood? I hope not. My general approach will be to focus on artifacts with certain properties paradigmatic of video games (notably computer‐based interactivity), keeping in mind contemporary examples such as the Legend of Zelda, Grand Theft Auto, Call of Duty, World of Warcraft, Portal, and Bioshock series. For simplicity, I also focus on single‐player games, though at the end of the article I briefly argue that my conclusions apply equally well to multiplayer games.

I. INTERACTIVITY

Video games are clearly ontologically multiple in the way that musical, theatrical, or dance works are. There is the game itself, but it can only be appreciated through an “instance” of the game, what I will call a playthrough, of which there are many.4 Moreover, each playthrough may differ in artistically relevant ways, for example, as to the fictional content of the playing (who dies, whether a particular mission is accomplished, and so on). One of the most obvious differences between traditional artworks (even ontologically multiple artworks) and video games is that video games are interactive. What does this mean? According to Dominic McIver Lopes: “A work of art is interactive to the degree that the actions of its users help generate its display (in prescribed ways)” (, 37). “Display” is Lopes's term for what an audience immediately attends to in appreciating a work (a paint‐covered canvas, the sounds of a musical performance, and so on). With regard to a video game, this is something like the sights and sounds of a single playthrough.5 “Users” of video games “generate” displays in prescribed ways when they play the game, just as a musician generates a display in prescribed ways when she plays from a score. For instance, you might generate a display in which your avatar turns left rather than right at a fork in the road by pressing a button on a controller. If, instead, you hack the game to generate a display in which all the zombies lie down, enabling you to waltz through every level unchallenged, you would thereby violate the prescribed way of generating the display; in other words, you would no longer be playing the game. The musical analogue of this situation might be that of an unskilled violinist who “performs” a Bach sonata for solo violin by playing it extremely slowly and transposing all the high notes down an octave. Such a performer is not playing Bach's work, though that work might be recognizable in his performance, since he does not follow Bach's explicit and implicit prescriptions (S. Davies , 1–197, summary at 196–197).

Berys Gaut criticizes Lopes's account of interactivity for being too inclusive (Gaut , 141–144). He points out that by this measure most traditional works for performance are interactive. “Users” (that is, performers) of Caryl Churchill's Top Girls, Martha Graham's Appalachian Spring, or Clara Schumann's piano trio all help generate these works’ “displays” (that is, their performances) in prescribed ways: they follow the instructions inscribed in a score or script or passed down orally, fleshing them out in the course of a performative interpretation. But these are traditional, not interactive, works.

To solve this problem, both Gaut and Lopes refine the concept of a user of an interactive work and its relationship to various other roles people play when engaging with different kinds of works. However, they refine the concept in significantly different ways. Gaut argues that “the audience role in the interactive case is to appreciate the work by instantiating it; merely watching the work while someone else instantiates it does not count as fully appreciating it” (, 143). This nicely covers both interactive innovations in the traditional performing arts, such as audience‐involving theatrical performances, and what Lopes calls computer art, arguably including some video games. Lopes, by contrast, does not require that a user is also the audience for the display she generates. Instead, he explains the role of user in terms of a distinctive mode of generating the work's display:

A person plays the role of user in generating a display of a work only if he or she (1) generates the display, (2) exploring the work, so that (3) an audience attends to the work partly by attending to his or her doing (1) and (2). (Lopes , 82)

Lopes does say that “[q]uite often the roles of audience and user are played by the same person” (, 83), but this misses Gaut's insight that such a combination of roles is essential to the kind of interactivity both philosophers are trying to capture.

Perhaps the kinds of cases that lead Lopes to stop short of necessity are those where, for instance, we appreciate a work by observing someone else interacting with it, for example the interactive displays in airports where the movements of passing travelers set images of piles of leaves in motion. But such cases do not falsify the claim that interactivity requires the identity of user and audience. First, if there is a user in this situation, it is the person we observe; were it not for the role that this person plays, we would not consider the work interactive. Second, though we may appreciate the work to some extent by observing another person interact with it, this cannot count as fully proper appreciation of the work insofar as it is interactive. (It might be a hybrid case that can be appreciated for its interactive aspects in addition to other, noninteractive aspects, just as a multimedia work might be appreciated for both its performance and nonperformance aspects.) Third, if a work were designed so that only a noninteracting audience could properly appreciate it through observing a third party exploratorily generating a display, it would be incorrectly classified as interactive. For the third party, while “interacting” with the work in some sense, would not be properly appreciating the work whose display he is generating (he would be a kind of artistic stooge), and the audience, while properly appreciating the display and its generation, would not be interacting with the work in any way that goes beyond the activities of audiences for traditional artworks. Theatrical examples of this kind of thing include magic and comedy routines in which a member of the audience is brought on stage and unwittingly humiliated in various ways. Though still a member of the audience in an institutional sense (for example, she has a ticket), such a person is no longer playing the audience role for this particular trick or sketch.6 In sum, Gaut captures the essence of interactivity better than Lopes.7

II. PERFORMANCE

However, Gaut thinks that while his account prevents traditional works for performance from being classified as interactive, it implies that interactive computer artworks, such as video games, are works for performance. That is, when you play a video game, according to Gaut, you are performing a work (the game), just as when a musician sings a song. He says that his definition of the role of user of an interactive work “states that users’ actions instantiate the work, i.e., that users have the key capacity of performers” (, 143) so that “in interactive works audience members take on the role of the performer as well as of the audience” (, 145).8

The features of interactivity that Gaut identifies, however, are not sufficient for performance. Workers at the foundry casting a bronze from an artist's mold instantiate the work, but they are not thereby performers. The problem is not just that they do not make decisions about artistically relevant features of the work or instance; an artist printing her own photographs may make just such decisions, but she is thereby no more a performer of her photograph than the foundry workers are performers of the sculpture.9 The obvious thing to say is that these are not the relevant kinds of instance (or work) because they are not performances (or works for performance). But now we need an account of the performance of a work for performance.

David Davies plausibly argues that “the performer differs from the mere agent whose behavior is subject to evaluation in that she intends for her actions to be appreciated and evaluated, and thus is consciously guided in what she does by the expected eye or ear of an intended qualified audience” (, 6; original emphasis). We might thus say that a child, a teacher, or a professional athlete gives a performance, meaning merely that their actions are guided by the appreciation and evaluation they are intended to receive (D. Davies , 4–7). A video‐game player can certainly perform in this sense, for herself or others. But what of the more demanding concept of performance of a work?

Lopes offers the following account of the role of work‐performer in order to distinguish interaction from work‐performance:

A person plays the role of performer in generating a display of a work only if he or she (1) generates the display (2) as a result of knowing what features it must have in order for it to be a display of that work and (3) with an intention to generate a display which has those features, so that (4) an audience attends to the work partly by attending to his or her doing (1), (2), and (3). (, 79)

Though we have moved on from that topic, Lopes presents this account in the hope of avoiding Gaut's objection that his account of interactivity is too broad. He points out that users of interactive computer works (including gamers) do not satisfy conditions (2) through (4), since the computational process of a computer work ensures that each of its playings will have the work's necessary features. (Note that conditions (2) and (3) here represent the central activity in producing displays as a performer, replacing the second, exploration condition in Lopes's account of the role of a user.) But if Lopes's intention in developing these concepts of performer and user is to make them mutually exclusive, he might have failed. For a jazz musician typically performs a work by following some sort of prescription while simultaneously exploring that work in her improvisations, and her audience appropriately attends to both things—appreciating her playing of the piece and, for instance, admiring the unexpected melodic resources she discovers in it.10 So jazz performers are users of interactive artworks, on Lopes's account of these roles, and that sounds implausible. However, Gaut's insight about interactivity is enough to exclude traditional work‐performances, including jazz improvisations, from that domain, since jazz performances are typically for an audience distinct from the performers themselves. Indeed, even if some jazz musicians improvise on a work for themselves, it is not obvious that this is a mode of engagement prescribed by the work. That is, it is not obvious that jazz standards are, even in part, interactive works.

In short, Gaut is right that interactivity is essentially a matter of one person's simultaneously playing two roles—instantiator and audience. But I believe he is mistaken that gamers thereby perform the works they instance. One way to defend this claim would be to argue that gameplay is not a significant focus of appreciation in the gaming world; the video games themselves play that role (Smuts , §3; Tavinor , 58–59 et passim). Although I am sympathetic to this view, in the absence of some hard empirical evidence for it, it might seem at this point in the dialectic to be question‐begging.11 Alternatively, one might defend this claim on the basis of Lopes's account of work‐performance: the gamer typically lacks the knowledge of the work required to produce a display with features she intends it to have, for the appreciation of an audience, so she fails to meet a necessary condition on work‐performance. But this would ignore significant complexity in Gaut's theory of gamers as performers.

II.A. Assisted Performance

Gaut begins by pointing out that work performance has two aspects: compliance and (performance) interpretation. Compliance is a matter of getting the work right—playing all the notes in the score of a piano sonata, for instance. “Performance interpretation involves further determining the [performance] events’ properties, beyond those required by compliance, so as to suggest or ground a critical interpretation” (, 145). He also claims that performance “[i]nterpretation is not required, but it must always be possible, for performance to occur” (145). This is a bit quick, since Gaut has folded two distinct ideas into his concept of interpretation. Every compliant performance requires “further determining the [performance] events’ properties” (the first part of Gaut's gloss of performance interpretation) since works for performance are necessarily “thin,” that is, they underdetermine their performances. One cannot just play what is represented in the score of a piano sonata, for instance; one must play those notes and rhythms at a particular tempo, with particular dynamics and articulation, and so on (S. Davies , 3–4).12 But this is a weak sense of “interpretation.” A novice pianist might well get all the notes right and play the entire piece within acceptable ranges of tempo, dynamics, and so on, yet produce a robotic or incoherent performance of the work. To produce an interpretation in the strong sense would be to perform it “so as to suggest or ground a critical interpretation” (the second part of Gaut's gloss of performance interpretation), that is, to communicate a conception of the work—the “critical interpretation”—by playing it in a particular way.13 Nonetheless, with these clarifications in mind, Gaut's identification of compliance and interpretation as distinct aspects of work‐performance is unobjectionable.

Playing a video game is different from performing a traditional work for performance, according to Gaut, because (some of) the compliance aspect of the performance is automated.14 That is, assuming that the technology is working properly, one cannot avoid doing (some of) what is merely required at a particular point in the game for properly instantiating it, though one might, like the novice pianist, produce an awkward or uninteresting instance of the game. Gaut calls performances where compliance is automated “assisted performances” and considers the objection that such automation implies that there is no room for the gamer to provide her interpretation of the work she is playing.15 But, of course, since it is only compliance that is automated, the possibility of interpretation is uninhibited (Gaut , 146). Employing the two senses of “interpretation” explored above, we might say that automated compliance still necessitates interpretation in the weak sense—though engagement with the head zombie may be automated, the player must engage with it in some particular way that is under her control—and allows for, but does not necessitate, interpretation in the strong sense—the sum of the player's choices may add up not to a coherent critical interpretation of the game but a workmanlike, robotic, or mindlessly incoherent playthrough.

Attempting to strengthen his defense against this objection, Gaut also argues that in a typical assisted performance not all compliance is automated, since “automated compliance must have its limits if the audience is to have a satisfying experience” (, 146), the idea being, presumably, that to the extent that the automation of compliance is limited, there is even more room for the gamer to express an interpretation in her playthrough. In particular, Gaut says that

the most salient rules from the player's perspective in videogames are the ones that she must follow in order to win the game or more generally to achieve desired results—she must learn how to control her character so as to shoot the alien or find the treasure or save a failing marriage. When successful in her aims, she is following a rule to bring about these results; if compliance with these rules were automated, there would be no challenge and no enjoyment. (146)

There is some confusion about the nature of game rules in this passage. Consider Monopoly. In order to “win the game or … achieve desired results” in that game, one must learn when to buy property and when to refrain from doing so, when to develop one's houses into hotels, how to conceal one's strategy from other players, and so on. But when one is successful in those aims, one is not “following a rule to bring about these results” in the sense in which one follows the rule that “[i]f you do not throw doubles by your third turn [in Jail], you must pay the $50 fine” (Hasbro n.d., 3).16 If you fail to employ a winning strategy, you do not thereby break a rule, that is, cheat or otherwise fail to play the game. And one could produce (surely someone has produced) a computer version of Monopoly in which compliance is automated (that is, you cannot cheat without hacking the game). If this is right, then it is not correct that “automated compliance must have its limits” in order for a game to be satisfyingly challenging. Monopoly played online with fully automated compliance will be just as satisfyingly challenging as Monopoly played on a physical board. What must have limits of a certain sort for a game to be satisfying is rather the rules themselves, as anyone who has tried to create or even modify a game well knows.17

There is a second confusion here about the nature of assisted performance. Gaut argues that there is more room than there might be for a player's performative interpretation in typical video games because compliance is typically not completely automated. But since interpretation (in both the weak and strong senses) only kicks in once compliance is achieved, whether or not compliance is automated is irrelevant to a work's scope for interpretation.18 For a rather baroque analogy, imagine a sonata for digital keyboard by an avant‐garde composer. Some passages are for traditional performance—the performer must play the notes in the score, but has the usual leeway with respect to details of tempo, dynamics, articulation, and so on—while other passages are “assisted”—the keyboard ensures that the right notes are played and the performer has control only over the details of tempo, dynamics, articulation, and so on. If Gaut were right about the interpretive benefits of limiting automated compliance, the performer of this work should have more scope for providing his interpretation of the work in the traditional passages than in the assisted ones, since more compliance is automated in the latter. But this is clearly not the case. The performer's “freedom” to fail to comply (for example, to play wrong notes) in the traditional passages does not give him more room to provide an interpretation in either sense. He has just as much control over the details of tempo, dynamics, articulation, and so on (the weak sense of interpretation), and he can hardly use his freedom not to comply to provide more of (?), or a more interesting, performative interpretation of those passages (the strong sense of interpretation).19

Does the concept of assisted performance help Gaut to make the case that gamers are (assisted) performers? As we have just seen, Gaut is correct that the objection that assisted performance precludes the possibility of interpretation fails. The concept also enables the proponent of gaming as performance to answer the objection based on Lopes's epistemic condition on work‐performance. Recall that Lopes requires a performer to “generate [a] display … as a result of knowing what features it must have in order for it to be a display of that work and … with an intention to generate a display which has those features” (, 79). The objection to gamers’ being performers based on these conditions is that gamers do not typically know what features a display of a game must have in order to be an instance of that game. Gaut's concept of assisted performance handily provides a response to the objection: the conditions that Lopes supplies here are at best necessary for traditional performance, but not for assisted performance. The automation of compliance lifts from the performer's shoulders the epistemic burden of knowing what features something must have to be an instance of the work in question while still leaving her room to provide an interpretation of the work.

And yet it still seems to me that gamers are not performers, even allowing for the possibility of assisted performance. It would be nice to have a knockdown argument at this point, but all I have is (i) a reductio showing that something is wrong with Gaut's view, and (ii) some considerations suggestive of where it goes wrong. Here is the reductio: consider again a computer‐based version of Monopoly in which compliance is automated.20 When you create a display or instance of this work (that is, play a game of it), your “actions instantiate the work,” that is, according to Gaut, you “have the key capacity of performers” (Gaut , 143). Does the fact that you may not know all the rules of the game (because they are enforced by the computer) prohibit your actions from being a performance? No, since in assisted performances, compliance is automated. Lest one think that this automation of compliance precludes you from being able to offer a performative interpretation of the work, note that the way in which you play the game—the choices you make in the space of possibilities left open by compliance with the rules—seems to be a potential locus for interpretation in both the strong and weak senses: you might have made different choices at various points in the unfolding of the game (the weak sense of interpretation) and you might employ those choices “so as to suggest or ground a critical interpretation” (the strong sense of interpretation)—a critique of capitalism, for instance. Thus, Gaut's discussion implies that when you play Monopoly on a computer, you are performing that work, just as a musician might perform one of Bach's sonatas for solo violin (with the difference, of course, that the former performance is assisted).

This shows that something is wrong with Gaut's account.21 More is necessary for work‐performance, even assisted performance, than Gaut considers sufficient. A comprehensive account of work‐performance would be invaluable for diagnosing the problem, but all I can offer is two kinds of consideration that seem to me to be reasons for thinking that typical video‐game play is not work‐performance. First, note that in contemporary video games of the sort that Gaut is most interested in, as opposed to computer Monopoly, it takes a while to complete a single playthrough of the entire game. Grant Tavinor, for instance (no slouch when it comes to gaming), notes that it took him over 56 hours to initially complete Grand Theft Auto IV (, 117). This exploration period, which constitutes most of the time that many people, at least, spend with a given game, seems more like the period a musician spends learning a new piece than like performance(s) of the piece.22 Relatedly, note that this period is typically discontinuous. I presume that Tavinor did not stay up for 56 hours straight (while assistants provided him with caffeinated beverages?), in order to complete Grand Theft Auto IV the first time he played it.23 But even if he had, his player‐character presumably died many times as he explored the work, figuring out what the (automatically complied with) rules of the game are and what is possible within those enforced constraints. Both kinds of discontinuity are atypical of traditional performance (but typical, mutatis mutandis, for traditional practice and rehearsal). This may not be decisive, however, since some similar discontinuities exist in traditional performance, and others are difficult to exclude from the realm of performance a priori. For instance, there are unmeasured breaks between movements in the performance of a symphony. Moreover, though symphonies are nowadays typically performed continuously, in past centuries it was not uncommon for individual movements to be played in isolation or for other works to be interpolated between the movements. Further, one can imagine an experimental work that is to be performed from the score “at sight,” that is, without preparation, the performance to be considered complete only when the performer has got through the entire score playing at least, say, each measure continuously with no errors. Anyway, even if these considerations are compelling, they leave open the possibility, to which I will turn shortly, that expert gamers may perform the works they have mastered.

The second kind of consideration provides a stronger reason for thinking that gameplay is not work‐performance, if one more difficult to establish. It is simply that there is no role for a performer prescribed by (the designers of) typical video games. As a result, one can no more perform such a work than one can perform, for example, a painting.24 Establishing this point conclusively would require at least two things: (i) an accurate description of the creative activities of the designers of typical video games, and (ii) a theory of how such activities constitute, or at least affect, the norms of interaction with video games (or perhaps artworks more generally). The second topic would take us too far afield;25 the first is a task better suited to philosophically informed psychologists and sociologists. I can, however, point to two aspects of at least paradigmatic work‐performances that are not possessed by typical video‐game play. First, performance is paradigmatically communicative in some broad sense. It is no accident that performers and audiences are typically distinct, any more than it is an accident that most letters are written to someone other than the writer. Perhaps one can perform a piece of music for oneself, but this, like a self‐addressed letter, is at least atypical. This does not show that one cannot be performing for oneself or others when one plays a video game, but it casts doubt on the possibility of a performance art that is essentially self‐directed, as an interactive performance art would have to be. More bluntly, I doubt that many gamers conceive of themselves as communicating something to themselves as they play. Their activities strike me as falling instead along a spectrum from exploration of the game to exercising mastery of it.26 This surely often involves communication from the designers to the gamer and increased appreciation of the game by the gamer, but neither of these is communication by the gamer to herself. (All of this seems indisputable when considering traditional games, such as Monopoly, and I cannot see any relevant difference between such games and typical contemporary video games.) The second, though somewhat overlapping, aspect of paradigmatic performances that is not shared by typical gameplay is their serving as vehicles for (performative) interpretation (in the strong sense discussed above). Though one might talk of a gamer's mastery of a particular video game or the brilliance of her choices and strategy in navigating the game, it is not common to talk of the interpretation of a game communicated through, or implicit in, a given playthrough. I do not mean to rely for this point on mere verbal usage; the point is that neither gamers nor those who follow other gamers commonly conceive of their gameplay as an act of communicating a particular conception of the game they are playing, that is, as embodying an interpretation. (Again, all of this seems indisputable when considering traditional games, such as Monopoly, and I cannot see any relevant difference between such games and typical contemporary video games.) Once more, none of this makes it impossible for the gamer to provide an interpretation of the game to herself (or to others) by playing it, but it does suggest that the idea that typical video games are designed for the gamer to play such an interpretive role is implausible.

II.B. Expert Gamers and Exemplary Playthroughs

What of expert gamers? Although Lopes resists the idea that typical gameplay is performance, he concedes that expert playings could be work‐performances in virtue of the expert player's fulfilling the crucial conditions of his characterization of the performer role (Lopes , 80). Since I have argued that those conditions are not necessary, ignoring as they do the possibility of assisted performance, and no one argues that they are sufficient, this argument is now irrelevant. Anyway, it is not clear that gaming expertise affects the arguments just given. Is it any more plausible that an expert Monopoly player performs the game than a tyro? Does the expert player seem to be communicating anything to himself in his play, let alone a critical interpretation of the game? Such rhetorical questions are dangerous, however, since a popular line of objection to my thesis points to the immense and still growing popularity of websites and events catering to the spectatorship of gameplay. Proponents of gameplay as performance argue that this shows that video games (if not Monopoly) can be, and are, performed for audiences. The gamers who participate in these practices are clearly aware of their audience, and it would be implausible to deny that many of them are performing for that audience. But this is true of many other competitive activities that are not essentially performative. Poker has become a surprisingly popular spectator “sport,” with many tournaments broadcast on television and the web, generating many poker celebrities. These players are surely performing for their viewership, but this does nothing to transform poker itself into a performance art. The crucial point, again, is that there is no role for a performer to play prescribed by the rules (or institutions) of poker. Similarly, there is no role for a performer to play prescribed by most video games.27

Would it make any difference if it turned out that video games, unlike poker, Monopoly, or Solitaire, are artworks?28 I do not see how. A speech may not be an artwork yet be written for performance in the relevant sense. There is a further subtle ontological difference between poker and contemporary video games (or speeches, for that matter): contemporary video games are created by fairly well‐defined teams and released on particular dates, while poker presumably developed over a long period of time through widely distributed informal establishment and modification of playing conventions. This raises further interesting questions about how norms emerge from human activity, but it does not seem to cast any doubt on the existence or even robustness of such norms. So, again, I fail to see how this affects the question of whether poker or video games are for performance. (Folk songs differ from “art songs” in just this way, but both are clearly for singing, whether or not the former are artworks.)

It might help to make explicit the distinction between performing in the production of (an instance of) an artwork and performing the work. Consider the films of Picasso painting on glass.29 These capture entertaining performances in which an artist produces an artwork.30 But what is captured is not a performance of that artwork. Even in the traditional performing arts, there are parts of the performance in a broad sense that are not parts of the performance of the work. Think of a pianist's theatrical mopping of his brow. He does this for the audience (hence “theatrical”), the same way that an athlete might play up her pain following a tackle—both musician and athlete are performing. But in neither case do we have (even part of) a performance of a work.

There is one final possibility to be considered as a case of gameplay that is work‐performance. Gaut argues that certain exemplary performances (of musical works, for instance) are works of art in their own right—“performance‐works,” as opposed to “work‐performances,” in David Davies's helpful terminology (D. Davies , 19)—and that certain exemplary playings of video games could similarly be works of art in their own right (Gaut , 144–145). If my arguments above have succeeded, then if a playthrough is a performance‐work, it is not so in virtue of being an exemplary work‐performance, since it is not a work‐performance at all. But could not a playthrough qualify as a performance‐work on its own merits, as it were, independent of the ontological status of the game itself? Many have argued, for instance, that “pure” improvisations, such as Keith Jarrett's Köln Concert, are works of art in their own right, despite being performances that are not of any preexisting (or indeed reinstantiable) work.31 Could an expert playthrough be just such a work of art?

Some have argued that ephemeral events, as opposed to persisting objects, are simply the wrong kind of thing to be works of art.32 Setting such general concerns aside for the sake of argument, the first thing to note is that one can incorporate just about anything into a performance‐work. For instance, Marina Abramović’s Imponderabilia involves a naked man and woman standing in either side of a doorway, between whom the audience must squeeze. The piece, originally enacted by Abramović and her partner Ulay in 1977, was recreated for her 2010 retrospective exhibition, The Artist is Present, at the Museum of Modern Art in New York, where it was enacted by other people. Assuming the 2010 event is a performance‐work, the fact that a particular doorway is (or was) an essential part of that performance‐work does not tell us anything particularly interesting about performance art, performance‐works, or, most relevantly for my concerns here, the ontology of doorways.33 And it makes no difference if what is incorporated into the performance is, unlike the MOMA doorway, an artwork in its own right. For instance, a performance by the Elevator Repair Service of their work Gatz arguably incorporates the entirety of F. Scott Fitzgerald's The Great Gatsby,34 but, just as with Imponderabilia and its doorway, this tells us nothing about the ontological nature of Fitzgerald's work. More simply, a recitation of Fitzgerald's novel, even if it were so beautiful and moving that it qualified as a performance‐work in its own right, would not show that the novel is a work for performance nor that the recitation was a performance of that work. (This example is similar to that of Picasso performing in creating a painting.) So even if it is the case that certain expert playings of video games are performance‐works, this does not tell us anything interesting about the ontology of video games or their playings.

Finally, notice that expert playthroughs are unlike free improvisations in that the gamer's performance relies on interacting in prescribed ways with a highly structured preexisting artifact: the video game. Whether this is relevant to the possibility that such playings are performance‐works may depend on one's theory of art, but at least one philosopher has provided arguments that might be extended to suggest that the highly structured nature of video games militates against playings of them counting as performance‐works. David Davies defends a neo‐Goodmanian conception of art, according to which something is art if it is intended to be appreciated for how it articulates a content in an artistic medium, where “artistic medium” is cashed out in terms of “symptoms” of the aesthetic, such as syntactic and semantic density, repleteness, exemplification, and so on (D. Davies , 14–17). Davies points out that where one's performance is of a highly structured work for performance, for example, a Romantic piano sonata, the constraints on performing the work will be such that there will be little room for the performer to articulate his own artistic content. This is not to say that the performer cannot provide an illuminating and valuable interpretation of the work he performs, but Davies distinguishes such derivative content from the original artistic content required for the creation of a new work of art. I have argued that video games are not works for performance; but if they are artworks at all, they are at least highly structured artifacts designed for players to explore interactively. It thus seems plausible that, like the classical pianist, the expert gamer will be constrained in what artistic contents she can articulate using the game.

One might resist this line of argument by maintaining that (some) video games are, on the contrary, extremely “open,” in the sense that they provide a virtual space in which the player can do a wide range of things. (“Sandbox” games such as Minecraft are the obvious examples here.) Again, Davies provides a potential model in his discussion of (some) jazz standards, such as Miles Davis's “Blue in Green” and Thelonious Monk's “Straight, No Chaser,” which he argues are composed “expressly to function as a frame for improvisation” (D. Davies , 155). There is a tension, however, between Davies's assumption that these standards are works for performance and his arguments, just considered, about the difficulty of a work‐performance's achieving the status of a performance‐work. For if the thickness of a Romantic musical work practically precludes its accurate performances from being performance‐works, then the very thinness of a standard composed to be a basis for improvisation would seem to militate against its articulating an artistic content of its own. So it seems we might be forced to choose between bestowing artwork status at best on either the game or (some) expert playings of it.35 Alternatively, we might consider sandbox games to be less like thin musical works and more like musical instruments, the result again being that the range of possibilities afforded by the game militates against its being an artwork in its own right.36

II.C. Two Final Issues

For all I have said, might there yet be video games that are for performance, ordinary playthroughs of which are performances of those games? Perhaps the best actual candidates are games based on traditional performance arts, particularly music, such as the Rock Band, SingStar, and Guitar Hero franchises. In these games, one appears to play and sing popular songs by manipulating controllers with more or less superficial resemblances to the instruments of a standard rock band and by actually singing into a microphone. Though originally designed for home use, there is now a significant tradition of live performance events using these games.37 One difficult issue these cases raise is what work, if any, is being performed. It certainly seems plausible that I sing the 1982 song “Eye of the Tiger” when I select that option in Rock Band, rather than the 2008 interactive computer work of the same name.38 But it also seems plausible that I am at least “using” the interactive computer work. Am I performing in either case? As I have already discussed at length, there is something nonparadigmatic about performing for oneself, though I do not rule out that possibility. And in the case of events with an audience distinct from the performer, I see no reason to deny that one is performing for that audience. But it seems much more plausible to me that one is performing the song than the interactive computer work in either case. Moreover, in cases where there is a separate audience, one's performance is not itself interactive, since the audience is not the “user.”

I have kept the arguments in the previous paragraph rather brief because, even if Rock Band et al. are not actual interactive performance works, nothing I have said rules out the possibility of such works. A philosophically inclined game designer convinced by my arguments might take them as a challenge to produce an indisputably interactive work for performance. All he would have to do would be to create an interactive computer work (preferably clearly a game39) with a prescribed role for a self‐directed performer to play. He could make the prescription explicit in the game itself, or in the accompanying instructions. When one interacted with this game in the prescribed way, one would be performing the game for oneself, with greater or lesser scope for varied performative interpretations of greater or lesser value, depending on the details of the game. If my arguments in this article have been convincing, then it should be clear that this game would be highly atypical compared to the vast majority of video games currently played on computers and consoles around the world. Although my focus has been on arguing that such games are not works for performance, I think that the considerations I have raised in favor of that thesis are also suggestive of reasons why video games are not typically for performance. In short, they are interactive, that is, their instantiators are their audiences, and performance is paradigmatically communicative or other‐directed.

Do multiplayer games complicate the issues I have discussed? Not much, it seems to me. It is no more paradigmatic for a group of people to perform a work for themselves than it is for a single person to perform for himself. After all, such a group does not produce a plurality of performances; they collaborate on a single performance. And though a group of gamers may (though they need not) plan a cooperative strategy for playing a game in advance of or during their gaming, it seems just as unlikely as in the single‐player case that they conceive of this strategy as aimed at communication of an interpretation of the game, to themselves or others.

III. CONCLUSION

In sum, video games are not works for performance because they are not created with a prescribed role for a work‐performer to play, and typical gameplay differs significantly from paradigmatic work‐performance. Thus, even if she may perform in playing the game, the player cannot give a performance of the game any more than one can give a performance of a painting, novel, or other nonperformance artwork. There may be exceptional video‐game playthroughs that are performance‐works in their own right, but this does not show that the games being played in such cases are works for performance or thus that the creators of such performance‐works are performing those games. After all, a performance artist may enact a performance consisting entirely of his playing a round of Solitaire. This would show neither that Solitaire is a work for performance, nor that in this instance the artist is performing Solitaire.

Video games are a very recent invention, and we sometimes assess the artistic potential of new media poorly. I have not offered any assessment of the artistic potential of video games or interactive computational technology more generally. But I have argued that those who believe that to play a video game is to perform a work are incorrectly assimilating a new medium to traditional artistic media with which we are more familiar. Whether or not video games are works of art in the first truly new artistic medium since the invention of film, I believe that we will have to develop new theories in order to understand what goes on when we interact with them.40

Footnotes

1.

Aaron Smuts (, §4) makes a similar argument.

2.

For ease of discussion, I will often talk as if video games are artworks, but everything I say could easily be translated into nonart terms. While on terminology: I use the terms “player” and “gamer” interchangeably to mean simply someone who plays a video game, and “gameplay” for that activity.

3.

This is not to say that my conclusions will be of no interest if it turns out that video games are not artworks. For what it is worth, I agree with the growing consensus among aestheticians that video games at least have the potential to be artworks and that some existing video games probably are artworks, while as yet no video game qualifies as a great (or perhaps even a good) artwork (Smuts ; Tavinor , chap. 9; Lopes , chap. 7. For dissent, see Rough ; for replies to Rough's arguments, see Conrad ).

4.

I hesitate to use the term “instance” since the “branching” nature of video games and other interactive works makes them quite unlike traditional ontologically multiple artforms, the implications of which for ontology and appreciation I hope to explore elsewhere.

5.

It may be that there is more to the display of a video game (or interactive computer works more generally), such as bodily states of the user. But this will not affect my arguments.

6.

Lopes's account also makes (the prescription of) a certain mode of engagement—exploration—essential to interactivity. This is problematic to the extent that one might master an interactive work without its ceasing thereby to be interactive. That is, there might be other proper modes of engagement with interactive works (compare Nguyen forthcoming). One response to the objection would be to follow Aaron Smuts in claiming that the interactivity of a work is relative to the skills of a given user (, 63–66). Alternatively, one might respond that a person who has mastered an interactive work is no longer capable of properly appreciating it. Since I side with Gaut on the definition of interactivity, I need not resolve this problem for Lopes's account.

7.

Perhaps a boilerplate disclaimer about “semantics” is worthwhile here: we could, of course, use “interactive” to label any of a range of phenomena. But it seems to me that the phenomenon that Lopes, Gaut, and others are interested in here includes many video games but not, for instance, the kinds of magic performance just referred to in the main text.

I should also note that I draw on Lopes's and Gaut's discussions of interactivity because they seem to me to be the most promising in the literature. For a useful philosophical survey of the range of theories of interactivity prior to Gaut's and Lopes's work and defense of yet another, see Smuts (). For a games‐studies perspective, see Landay ().

8.

In a recent article, Grant Tavinor, a little more cautiously, asserts “the prima facie resemblance of the playing of some video games to performances in art forms such as theater” (, 25). But though he talks of gameplay as “performance” throughout—and indeed in the title of—this article, he also acknowledges that “[n]one of [his discussion] necessarily means that video games are a performing art in the sense that chamber music or theater are performing arts. … [M]any … art forms can be described as involving performances without thereby counting as performing arts” (, 27). The complexity of this issue is suggested by Tavinor's shifting over time from the view that gamers are not performers because their actions are not subject to aesthetic evaluation (, 58–59) to the view that perhaps they are for the reasons Gaut gives (Tavinor , 571–572) and thence to the more cautious view just quoted.

9.

For some careful exploration of the variety of ways in which ontologically multiple works are instantiated, see D. Davies (, , and ), Abell (), and Gover ().

10.

In fact, I have argued that jazz musicians do not typically perform works (Kania ), but for simplicity I assume here that they do. I believe that this argument could be recast in terms of my view of jazz without impacting the conclusion.

11.

I turn to the spectatorship of gameplay later in this article.

12.

According to this terminology, works are “thinner” or “thicker” according to how much is required of the performer for compliance. “Thinner” works are those with fewer work‐determinative properties than “thicker” works (S. Davies , 3–4).

13.

In fact, the nature of performative interpretation and its relation to critical interpretation are matters of some controversy. See Levinson (), Maus (), Thom (), and Neufeld ().

14.

Thanks to a referee for suggesting that I discuss this aspect of Gaut's account in more depth.

15.

Compare Lopes's consideration of the objection to the value of computer art from the “creativity sink” (, 28–29, 85–87).

16.

There may be other senses in which one follows a rule in employing a strategy. Perhaps one has developed a strategic algorithm and is following it. Perhaps all cognition is rule governed, and thus all actions follow rules. But neither of these is a case of one's following the rules of the game.

17.

I skate over many fascinating issues in the philosophy of games here. For an introduction to some of these, see Bernard Suits's classic The Grasshopper (), which includes precise accounts of failing to play a game by trifling, cheating, and spoilsporting, and Thi Nguyen's excellent recent overview of the philosophy of games ().

18.

Some would dispute this claim. Aaron Ridley (, 105–131) and Julian Dodd (), for instance, both argue that one can give an insightful performance of a musical work (that is, one with a good performative interpretation) without even aiming to secure compliance. But, in part because Gaut implicitly rejects it, this is not the place to address such a possibility.

19.

It may of course be easier for the performer to execute his performative interpretation in the assisted passages, since he does not have to worry about getting the notes right. But that is a different issue.

20.

If the multiplayer nature of Monopoly raises complications, consider instead computer Solitaire or a version of computer Monopoly in which the “other players” are computer‐generated proxies.

21.

Of course, one might bite the bullet and claim to have discovered that all (automated?) gameplay is performance. But that seems less appealing than giving up on the idea that gamers are performers.

22.

Compare Tavinor's recent comments on video‐game play:

The most interesting part of a video game playing, from an artistic perspective, is the “interpretative performance” where the player explores the game to discover its interactive potential in a way that sheds light on the game's content and meaning. … Such interpretative performance frequently involves playing the game repeatedly to explore how different decisions affect the world and deliver different stories. If the video game work is … partly composed of an interactive algorithm, then interpreting the meaning of the work as a whole … will naturally depend on such repeated playings. (, 27)

Note that it would be odd to consider a musician's exploration of the nature and interpretive possibilities of a work in this way a performance; it sounds much more like practice or rehearsal, albeit in the service of (critically) “interpreting” the work. Thanks to David Davies for proposing the idea that gameplay is more like rehearsal than performance.

23.

I must admit to hypothesizing about Tavinor's gaming habits here, but if I am wrong about them, they are surely atypical.

24.

David Davies () argues that all works of art are performances. But even that view does not imply that all works of art are works for performance. See also D. Davies ().

25.

See D. Davies () for some helpful recent work on this topic.

26.

On gameplay as exploration, see Lopes (, 67–84) and Tavinor (, 27). On gameplay as mastery, see the references and discussion of expert gamers in the next subsection.

27.

Two dissimilarities between poker and video games that may be relevant here are that video games are (1) a relatively new phenomenon, and (2) a broad class of things, as opposed to one particular game. (Of course, there are many variants of poker, but they are all gambling card‐games; by contrast, as mentioned at the beginning of this article, “video game,” at least in ordinary English, comprises a wide variety of kinds of things.) It could be that the popularity of video game spectatorship will give rise to a significant independent tradition of video games for performance (or video game “instruments,” to presage a suggestion from the end of this section). But, for the reasons given elsewhere in this article, these will not be interactive works.

28.

Thanks to a referee for prompting me to address this point.

29.

See, for example, https://www.youtube.com/watch?v=CkRS3wDg1xU.

30.

I use “performance” in the strict sense here. These are not just documentary films of Picasso at work in his studio; they are events staged for the camera in which Picasso is clearly aware of, and performing for, his audience.

31.

For example, Alperson (); S. Davies (, 16–19); Hagberg (); and D. Davies (, 135–148).

32.

For example, Thom (, 28); Brown (, 353 and 366n2); and Kania (, 397–399).

33.

It is a nice question whether the 2010 event was in fact a work‐performance of a repeatable work or some kind of artist‐sanctioned replica of the ephemeral 1977 work.

34.

The complete text of the novel is spoken by the cast in the course of the play. My only hesitation is over whether recitations of ordinary works of literature, such as novels, count as instances of them. See Kivy () for some provocative claims in this area.

35.

It is worth emphasizing that the last two paragraphs depend heavily on D. Davies's theory of art, which is as controversial as anyone's.

36.

I am indebted to Sam Taylor for this analogy. Compare Tavinor's claim that “Minecraft might be better understood as a prop for telling stories … than a storytelling medium itself” (, 30) and Jay David Bolter and Diane Gromala's claim that “we perform [digital interactive works] as we would a musical instrument” (quoted in Gaut , 141) (a rather odd claim, since one does not perform a musical instrument).

37.

See, for example, the Rock Band events forum: http://forums.harmonixmusic.com/categories/rock-band-events.

38.

See Derksen and Hick () for an exploration of whether such events count as performances of the songs in question. Perhaps tellingly, the idea that one might be performing an interactive computer work in addition to or instead of the song does not seem to have crossed Derksen and Hick's mind.

39.

Richard Rouse III () argues that Rock Band and its ilk are nongame video games, since they fail to offer their players “interesting choices,” which he takes to be the point of games. I ignore such considerations in keeping with my liberal approach to what counts as a video game in this article. See again Suits () and Nguyen ().

40.

For helpful discussion of earlier versions of this essay, I thank Julie Post; my audience at the Seventh Annual Auburn Philosophy Conference: The Ontology of Art, in 2015; my commentator, Dom Lopes, and the audience at the American Society for Aesthetics Annual Meeting in 2016; and my audience at the Workshop on the Philosophy of Games in Salt Lake City in 2016, especially the organizers of that workshop, Thi Nguyen and Brock Rough. Thanks also to Daniel Conrad for a year of stimulating philosophical discussion of video games more generally. Finally, thanks to two referees for this journal, whose suggestions prompted many significant changes.