Spoors, G. (2000). Codes of Computer Gameplay. ANZCA Proceedings.

Codes of Computer Gameplay

Abstract

This paper briefly reviews some existing theories about narrative and interaction in computer games. It then extends Barthes analysis of narrative codes to accommodate theories about computer interaction, principally David Myers' theory of symbol transformations and Espen Aarseth's theory of ergodic texts, using the computer role-playing game Vampire: The Masquerade as a source of examples. The methodology developed provides a basis for more complex analyses of the relationships between game structure, processes of interaction, and discourses in which computer gameplay is situated.

1. Introduction

Mainstream research on computer games over the last few decades has emphasised sociological and psychological measurements of the influences, or effects, of games on youth, largely in response to parental and governmental concerns (for reviews of this research see Durkin, 1995; Kasvi, 2001). More recently, research in the cultural studies paradigm has turned to game aesthetics (Darley, 2000; Myers, 1990), identification (Friedman, 1995), spatiality (Aarseth, 1998), narrative (Landow, 1992; Frasca, 1999; Myers, 1991), and discourses of capitalism and gender in which gaming is situated (Cassel and Jenkins, 1998; Kinder, 1991; Stallabras, 1993; Skirrow, 1990; Turkle, 1995). However, the relationship between game structure, processes of interaction, and discursive formations has not yet been adequately theorised. This paper begins with a review of existing research in the field, then extends Roland Barthes' (1975) analysis of codes of narrative to incorporate David Myers' and Espen Aarseth's theories about computer game narration and interaction.

2. Narrative and Interaction in Computer Games

The narrative elements in many computer games are limited to introductory text or “canned” (pre-filmed) sequences (for example, at the start and end of each level). Such sequences are often seen as “cute or funny but basically irrelevant to their play” (Turkle, 1984, p. 66), in that they function as backdrop or metaphor for the action (Juul, 1998). Andrew Darley takes a different approach when he argues that by recounting the sequences of action in games we can construct narratives of “near-epic proportions” (2000, p. 152). However, such narratives are highly repetitive, fragmented and impoverished when compared with other narrative forms, as they lack closure, have little psychological or narrative depth, and are interrupted by frequent game saves, death and replay (p. 152-3).

While computer games are supposedly distinguished from other media by their use of interaction (Stallabras, 1993), many games have a pre-defined, linear sequence of events which is not substantially altered by player choices. Consequently, computer game narratives are usually seen as limiting the freedom of interaction (Darley, 2000, p. 154). However, recent attempts to theorise interactive narratives by analogy to “hypertext” indicate that the relationship between interaction and narrative may be more complicated than this (Friedman, 1995; Landow, 1992). Unlike traditional narratives, which are characterised by a beginning, middle and end and a predetermined relationship between writer and reader, a person navigating hypertext constructs his or her own text in a non-linear fashion. As David Myers (1991) argues, the plot-structure of computer games may similarly be understood in terms of event-probability. If at one end of a “computer game's interactivity continuum” (p. 296) event-probabilities remain largely unaffected by player choices, at the extreme end the player becomes the game-designer, with his or her choices determining the structure of the game text. Certainly, recent computer role-playing games (CRPGs), such as Baldur's Gate , offer choices that lead to a non-linear narrative, and text-based Multi-User Dungeons (MUDs) allow players to participate in the construction of not only non-linear and open-ended narratives but virtual identities and communities ( Baym, 1995; Danet, 1998; Kolko, 1994; Reid, 1994).

Ted Friedman, however, concludes that “the constant feedback between player and computer in computer games is far more complex than [the] simple networking model” provided by the paradigm of hypertext (1995, p. 74), and that the prevalence of interaction with non-textual interfaces problematises the analogy between reading and gameplay. Myers (1990) also concludes that literary analysis of computer games (that is, of their language, character and plot) errs because games are not read, but played. He argues for an “aesthetics of play” (p. 28), and identifies three dominant aesthetics of gameplay: “[1] Game as challenge . . . providing [or] being a challenging opponent . . . [2] Game as social activity . . . and [3] Game as meditation” (1991, p. 385). Darley uses the term kinaesthesia to refer to the physical component of this play, and concludes that “the various kinaesthesia that comprises a major part of these (relatively) semantically shallow texts constitutes their primary attraction” (2000, p. 194). For him, narrative in computer games is “ de-centred . . . in a subordinate position within the overall formal hierarchy that constitutes the game aesthetic” (p.151).

The problem with this approach is the underlying assumption, which Jesper Juul (1998) puts forward as his argument, that narrative and game are “two separate phenomena that in many situations are mutually exclusive.” Juul only maintains this distinction by confusing one particular kind of (literary) narrative with narrative in general and not adequately defining the terms “play” and “game.” Narrative frames do, certainly, function as a metaphor for the action in many computer games (Juul, 1998; Herz, 1997). However, narratives may motivate play (Fuller and Jenkins, 1995), and the term play is often used to describe the process of reading and/or interaction, to foreground the fact that meaning is determined not simply by a text, but through the agency of the reader/player (Aarseth, 1997; Frasca, 1999; Hutchinson, 1983; Myers, 1991). As Aarseth (1997) argues, while there may be differences between the categories of narrative and gaming, “the difference is not clear-cut, and there is significant overlap between the two.” It suffices to observe here that since many adventure and CRPGs incorporate narrative and gaming elements in complex ways, any analysis of the distinctive experience of gameplay needs to consider how codes of narrative and gaming function together.

Myers (1991) offers one model of this by relating theories of structuralist narratology to poststructuralist theories of play. He begins with the premise that, unlike most narratives, the structure of computer games is not linear but recursive. In Populous , for example, the sequence: Power à Transformation à Population à Power, might recur indefinitely. He then refers to A. J. Greimas' (1987) semiotic square, which maps the possible logical structural relations between units of meaning. In this model, the opposites of each term of a binary opposition provide two additional oppositions that are logically contrary to the initial two terms (see figure 1).

Figure 1. Greimas' (1987, p. 66) semiotic square adapted to computer gameplay.

As Myers puts it: “In Railroad Tycoon , for instance, there are three opponents to the game player, and each represents a slightly different sort of opposite to the game player—and to each other” (p. 341). The computer game player (s1) is opposed to a computer game opponent (s2), but also may be opposed to the logical opposites for each term: game world event-objects (not-s2) and real world event-objects (not s-1), respectively. An opponent presupposes (and is the negative of) a player; a game event-object presupposes (and is the negative of) a real-world event-object (the game itself). Since an event-object in a game (for example, a train crash), and a real-world event-object (usually the game itself), may either/both help or oppose the player, these terms logically contradict the other terms.

Myers adapts this model to computer game play by arguing that each iteration of game's basic sequence is contextualised by shifting relationships between these semantic possibilities. He refers to this as “recursive context shifting” (p. 343). In Railroad Tycoon , an opponent “might be used as a helper (to block the invasion of a more formidable opponent) or as an extension of the player (by absorbing the opponent's railroad company into his or her own)” (p. 342). When a player defeats all other players, the game does not end:

a player who has previously been playing opposite individual game elements . . . now plays against the game as a whole. The original helper . . . is transformed from game context to game design context, and so forth, in a spiral of increasing complexity (p. 342).

So while, from a structuralist perspective, closure is consequent upon the final unit of structure in a linear sequence, computer games foil closure through a non-linear process of recursion and the spiralling of the semantic field into a different symbolic arrangement (p. 342). Since this process only ends when play ends, “we begin to lose touch with the atemporal permanence of structuralism and slide towards an increasingly deconstructionist interpretation of the play process” (p. 342).

Unfortunately, Myers analyses games that do not have a linear narrative superimposed upon the iterative sequence, so he does not account for the more complex ways that narrative units are distributed and integrated in most adventure games, CRPGs and MUDs. If we accept that meaning is always unstable and negotiated, the task becomes how to trace the dialectic between processes of play and the structuring processes against which they operates. To this end, I argue that the codes of narration theorised by Roland Barthes (1975, 1982) can be extended to incorporate processes of interaction in computer gameplay.

3. Nine Codes of Gameplay

Narratologists initially studied the relationships between the minimal units of narratives in terms of, among other things, sequences (in which events are distributed paradigmatically, in chronological order), nuclei (indispensable units in a sequence), catalyzers (units which amplify or fill in a sequence), and macrostructures (in which events and event sequences are integrated syntagmatically according to logical/causal relations) (Barthes, 1982). However, it is not enough to look at the relationship between units in a narrative, as narratives are only (re-)produced through the act of reading. While a narrative may unfold in a linear sequence, the act of (de-)coding a text is a non-linear process : readers move back and forth through the narrative along and (re-)produce it along paradigmatic axes of distribution and syntagmatic axes of integration. It is necessary, then, to consider the codes by which reader's make are able to perform this work.

I consider nine codes here. The first six codes, deriving from Barthes (1975) work, are applicable to all narratives, though I am primarily concerned with their work during gameplay. These codes are: the proairetic code , the semic code , the hermeneutic code , the symbolic code , the reference code , the diegetic code . The last three codes incorporate theories about interactive narratives and gaming, and I refer to them as the ergodic code , the computational code , and the kinaesthetic code . Each of these codes “is one of the forces that can take over the text (of which the text is the network), one of the voices out of which the text is woven” (Barthes, 1975, p. 20-21). Many of these codes work simultaneously and effect the work of the other codes. Indeed, when the last three codes are passive, for example during cut-scenes, computer game narration may be virtually indistinguishable from print and film narration.

The methodology I suggest here, like Barthes', involves identifying minimal units ( lexies ) of narration or gameplay and identifying how they are structured by these codes during gameplay. A theoretical model is, of course, only as useful if it allows us to understand its object(s). However, I do not have space for an extended analysis, and use Nihilistic Software's (1997) computer role-playing game (CRPG) Vampire: The Masquerade—Redemption ( V:TM ) to help define the codes and their work.

3.1 CODES OF Narrative

3.1.1 The Proairetic Code

The proairetic code “provides the basis of events and sequences, proliferating linearly and irreversibly” (Cohen and Shires, 1988, p.119). While playing a computer game, we decode the sequence of events by organising them into groups under some generic title of action (walk, encounter, fight). This involves organising events according to a temporal logic, but nuclei are not separated from catalyzers, nor are sequences organised macrosequences or a macrostructure. However, computer games tend towards shifting levels of detail and generality in the organisation of sequences.

For example, a new game in V:TM begins with an introductory cut-scene, the events of which a reader will organise at a level of generality common to film narratives: Christof (C) fights heathens; C is felled by an arrow; C awakes in a Convent and sees the nun Anezka; the two characters talk; the Archbishop Geza, interrupts enters and speaks; C sleeps. When C awakes and the player gains control of C, the keyboard and mouse are used to move C. This control may be broken up into such events as: C moves forward; C moves forward; C turns to the left; C turns to the left; C looks to the left; C moves forward; C moves forward and to the left. Such minimal event-units (mapped to player keystrokes and mouse clicks) may proliferate ad nauseam , and, potentially, ad infinitum , leading to a seemingly endless enchaining of catalyzers, but (upon their completion) all of these events may be organised into a single sequence: C walks across the room. In computer games, such proliferation of detail violates the “threshold of functional relevance, that which divides the narratable below the non-narratable, sequences below which are taken-for-granted” (Heath, quoted by Culler, 1975, p. 143). While there is a certain level of generality in our descriptions (C walks across the room), computer games draw attention to units below this level of generality (C moves forward; C moved forward; C moves forward).

As Michel de Certeau has argued, e vent sequences generate space (1974). For example, a sequence in which C walks across the room implies the space of the room and the movement of C through this space. However, while all narratives may generate space in this sense, Aarseth (1998) stresses that:

The defining element in computer games is spatiality. Computer games are essentially concerned with spatial representation and negotiation, and therefore a classification of computer games can be based on how they represent—or, perhaps, implement—space.

Aarseth uses Henri Lefebvre's terminology to argue that gameplay is a “spatial practice” that generates “representations of space”: space conceptualised as a formal or logical system of relations (Lefebvre, 1991, pp. 38). For example, a sequence of spaces—the possible co-ordinates within a certain area—may be subsumed to the generic title of a space: C's room in the convent. Events and spaces may be, but are not necessarily, mapped against one another, as a single event may describe complex movement through more than one space (C walks through the streets of Prague), and multiple events may occur in a single space (C sleeps in his bed; C is woken by a scream; C gets out of bed; C retrieves his sword; C leaves the room). The proairetic code, then, provides the basis of sequences of events and sequences of space.

Figure 2. Screen shot of V:TM , with Christof's character window open.

3.1.2 The Semic Code

The semic code “provides the basis of character traits” (Cohen and Shires, 1988, p. 119). These traits are allocated to individual characters to define our expectations of their role in the sequence of events, but may be grouped thematically to provide chains of connotations. For example, we may assign the trait of “strength” to one character, but the trait may also function thematically, recurring throughout the text to provide “flickers of meaning” (Barthes, 1975, p. 19). As the narrative of a computer game develops, characters may accrue semes found in traditional print and film narratives, such as “trustworthy,” “talkative” and “informative.” However, the semic code in computer games also incorporates character statistics and all other game variables visible to the player. In V:TM , each character has a fixed series of semes, such as “health,” “bloodpool,” “frenzy,” “humanity,” “strength,” “dexterity,” “stamina,” and so on, and each is assigned a different numerical value (see figure 2). These traits may function thematically : “health” may code the historical period or setting (the medieval city of Prague may be un healthy: decadent, corrupt), and “blood” may code a person or group in physical terms (sensual, engorged, strong) or psychological terms (bloody-minded, ruthless). As characters go up levels, they may gain additional semes by completing quests (they become increasingly “heroic”) and acquire new skills or abilities (“disciplines”), such as “blood healing.”

However, the semes in computer games fluctuate much more frequently than in most narratives. In V:TM , each character's numerical trait for “health” or “bloodpool” is represented as a ratio against a (current) maximum. In figure 2, Christof's (C's) health is 70/100, his bloodpool is 80/80. During a battle, C's current health may drop from 70 to 50 (50/100). If he then had enough experience points to go up a level, C's maximum health might be increased to 110, and if he rested his current health would reach this new maximum (110/110). To account for the changing values of game variables, we can distinguish between dynamic semes , to refer to a trait's current value (in figure 1, C's current health is 70); static semes , to refer to a trait's current maximum value (in figure 2, C's maximum health is 100); and potential semes , to refer to traits which the character may be able to acquire or to a trait's absolute (static) maximum within the game's diegesis (a game like V:TM might, for example, place a numerical limit on a character's health, so that the player does not become too powerful and upset the game's balance).

3.1.3. The Hermeneutic Code

The hermeneutic code “provides the basis of a macrostructure linearly and irreversibly directed towards closure” (Cohen and Shires, 1988, p. 119). This is the code of suspense, in which nuclei and catalyzers are distinguished on the basis of key questions, enigmas and plot developments, and sequences are (re-) organised into macrosequences and a macrostructure. The hermeneutic code often refers to other narratives—or genres, such as romance, fantasy, science-fiction, horror—as a basis for expectation.

If, as Herz (1997) argues, the plot of many computer games is the story of player actions, we can identify recurring game macrostructures related to movement and combat. These macrostructures will determine the organisation of event sequences and the basis of the game's suspense: “Will I fall?” “How much damage did it do?” “Did the vampire hit me?” “Did I hit it?” “Am I dead?” In adventure and CRPG games, however, the hermeneutic code may also function in more complicated ways, as it does in many print or film narratives. In V:TM , the following macrostructure develops: Christof, a Christian knight, is “embraced” by a vampire from the Brujah Clan, and attempts to maintain his humanity and faith as he struggles with his love for the nun, Anezka, and fights against other vampire Clans. This unfolding narrative macrostructure, which draws from print and film genres of romance, fantasy and horror, is continually interrupted by macrostructures of movement and combat common to most computer games. This may involve a periodic displacement of the narrative macrostructure by macrostructures of movement and combat, so while some extended sequences might not develop the macrostructure at all, some short sequences may advance it significantly. However, the centrality of game macrostructures and the narrative macrostructure varies throughout the game. It does not suffice to say, then, that narratives in computer games, then, are simply de -centred, as Darley puts it: game and narrative macrostructures are subject to dynamic (non-linear) processes of centring, de-centring and re-centring.

3.1.4. The Symbolic Code

The symbolic code “provides the basis of representation through reversible binary oppositions” (Cohen and Shires, 1988, p. 120). The symbolic code organises the signs defined in the proairetic and semic codes according to increasingly abstract oppositions. “The symbolic code inscribes the text as a site in which the privileging of one binary term over another is both staged and exposed, legitimised and placed in jeopardy” (p. 125). So while the semic code links connotations to characters, events and objects, the symbolic code organises these groupings in relationship to one another on the basis of similarity and difference.

For example, in V:TM , the protagonist Christof is both muscular and heroic, but he is also so sentimental and faithful that he takes on traditionally feminine attributes. Vampires also have both masculine and feminine traits: while the penetration of the neck of a victim is a phallic act, the vampire's messy act of sucking blood evokes breastfeeding and menstruation. Yet vampires in V:TM are distinguished from one another on the basis of the traits of the Clans to which they belong, and, more importantly, whether they live anonymously alongside humans or reveal themselves in an attempt to destroy them. Christof, in identifying with the former, suppresses the feminine and evil traiting of vampires and his own personality, and the game ultimately creates a general opposition between, on the one hand, Christof's attempt to preserve his life, humanity, faith, masculine and reason, and, on the other hand, the undead, feminine, inhuman (vampiric), profane, mad and reckless forces that would destroy him.

As Myers indicates, Greimas' semantic square can be used to show how binary oppositions can be formulated not simply in pairs but in terms of relationships between four terms (s1, s2, not-s1, not s2). Myers' process of “recursive context shifting”—the spiralling of the game's semantic field into a macrostructure—occurs along these axes. However, Myers argues that this spiralling progresses towards increasing complexity, when it may lead to a recurring patterns of varying complexity. For example, a complex series of oppositions between several human and computer-controlled players as well as several simultaneous game events may be followed by a more basic conflict between two surviving human players. Furthermore, while Myers uses Greimas' square to map orientations towards play in the game macrostructure (relationships between players, opponents, game events and the game itself), symbolic transformations also occur in the narrative macrostructure. Early on in V:TM , an opposition is set up between the human protagonist Christof and the vampire Ahzra, but, after defeating Ahzra, Christof is bitten by a vampire from the Brujah Clan and becomes one himself. The Brujah, who respect human life, are opposed to vampires who seek to destroy or enslave humanity, and Christof comes to fight alongside them. However, Christof also struggles continuously against the primal urges that threaten to turn him into “the Beast.” When Christof turns into “the Beast” he attacks the nearest character, friend or foe, and the player is unable to control him until the frenzy passes. We can map these symbolic oppositions as follows:

Figure 3. Greimas' semiotic square applied to symbolic transformations in the narrative macrostructure of V:TM .

The spiralling of the game into a “macrostructure” is better seen, then, as a complicated process of distribution and integration of narrative units that may involve shifts between multiple (game and narrative) macrostructures. This requires the co-operation of multiple codes. Moving Christof around Prague changes game variables (C's co-ordinates move from X 1 ,Y 1 ,Z 1 to X 2 , Y 2 ,Z 2 ) but does not necessarily advance the game's macrostructure or symbolic organisation. When the player moves Christof into the lair of the vampire Ahzra, however, this instigates Ahzra's speech, which introduces enigmas such as: “Who are Cainites?” and “Why are vampires in Prague?” Ahzra's subsequent defeat is decided by the semic code (Ahzra's “health” and “bloodpool” are reduced to “0”), leading to a symbolic transformation in the narrative macrostructure (Christof is superior, Ahzra is inferior; the goodness of Christianity triumphs over the evil of the heretical). This is accompanied by a shift from an opposition in the game macrostructure between character (Christof) and opponent (Ahzra) to an opposition between player (having to guide Christof back to Prague) and game (the game world that must be navigated).

3.1.5 The Reference Code

The reference code “provides the basis of seemingly extra-textual referentiality” (Cohen and Shires, 1988, p. 120) and involves a text referring to “a science or a body of knowledge” (Barthes, 1974, p.20) outside the text, be it “physical, physiological, medical, psychological, literary, historical” (p. 20). These references naturalise the “truth” of the narrative, inasmuch as they draw from ideologies familiar to the reader. The reference code comes into play in the coding of the four cities and historical periods in which V:TM is set: Prague, Vienna, New York and London, in that, through the digital architecture, music and text, these settings retain some authenticity. Furthermore, the film or print genre(s) to which a computer game refers ( horror, science fiction, fantasy, post-apocalyptic, mystery) will also determine what is real—believable/unbelievable—in the game's diegesis. For example, since V:TM refers to the genre of fantasy and horror, the fantastical events that occur in these otherwise historical settings are accepted (vampires with magical powers walk the streets of London in 1999).

3.1.6. The Diegetic Code

The diegetic code encompasses the codes and conventions of the narrative genre (horror, adventure, and so on) and the narrative medium (print, film, computer game) (Cohen and Shires, 1988, p. 130-131). The work of the diegetic code can be understood in terms of what Gerard Genette (1980) has called focalisation . Focalisation refers to the dynamic relationships between whoever supplies the narration ( narrating agents ), the characters whose point of view focuses the narration ( focalisers ), and what is being narrated ( focalised ). In figure 4, for example, we might say that the camera is the narrating agent, the characters Wilhelm and Christof are the focalisers, and the painting on the wall is what is being focalised. The way focalisation is handled affects the way a narrative is understood. When a character (focaliser) functions as the narrating agent, the narration tends to be seen as a subjective account, and readers tend to identify with a focaliser (say, the central hero) more than the focalised (say, the villain). Focalisation also manages disparities of knowledge between narrators, characters and readers, in that the narrating agent (say, the camera) may reveal more than a focaliser (a character) may see (a basic example of is evident when a film audience cries out to a character: “Turn around, it's right behind you!”).

Figure 4. Focalisation in V:TM.

Figure 5. Direct address in V:TM

Print and film narratives have distinct codes and conventions that manage focalisation. Print narratives, for example, have first-person, second-person, third-person modes of narration, past, present and future tenses, nested narratives, and stream-of-consciousness (see Lanser, 1981), while film narratives have the shot/reverse shot, establishing shots, the 180 ° rule, the 30 ° rule, and the orchestration of actor's within the frame (Lapsley and Westlake, 1988, p.139-141). These all work to reinscribe the shifting and multiple perspectives around the ideal of an individual's privileged point of view, identified analogously (in film) with the camera as eye/I (p. 140). While standard computer game interfaces often draw upon these codes (figure 4), computer game narration is distinctive. To take one example, since the player in V:TM , like in many other 3D games, can enter first-person mode by pressing a key, it can be argued that t he shot/reverse shot is built into the game engine around the unifying perspective of the player. While all narratives position readers as subjects, (Lapsley and Westlake, p. 59), the player continually (re-) positions him/herself as a subject through interacting with the game, helping to contain the difficulty an interface may have in maintaining visual and conceptual intelligibility. Indeed, since players in V:TM are able to manually operate the camera position, there may be a sense in which they feel responsible for failures in the game's narration. However, the player's ability to intervene as a narrator or focaliser in the diegetic code is governed by the ergodic code.

3.2 Codes of interaction

3.2.1 The Ergodic Code

Aarseth (1997) uses the term ergodic —deriving from the Greek words ergon (work) and hodos (path)—to refer to any text whose structure is physically altered by the activity of the “reader” (hypertext being exemplary). While readers of film and print (non-ergodic) texts perform the work of reading “in their heads,” the user of an ergodic text

also performs in an extranoematic sense. During the cybertextual process, the user will have effectuated a semiotic sequence, and this selective movement is a work of physical construction that the various concepts of “reading” do not account for. (1997)

The ergodic code , as I define it here, provides the basis for admissible choices for interaction at the level of the interface and the extent to which these interactions affect the work of the other codes. The ergodic code is transparent when the user is familiar with the interface, but is visible when being it is customised, learnt, or is frustrating the player (because of poor design or lack of mastery).

The ergodic code governs when the player may interact. During a cut-scene this may be limited to pressing <ESC> to end the playback, but the timing of interactions is usually determined by iterations of the computational code (see below), turn-taking (each player/character takes a turn, one after the other), and the time it takes to complete certain events (it may take a few seconds in game time for a character to reload a gun, and the character can perform no other action until that action is completed). The ergodic code also determines the permissible objects of interaction (objects that may be moved, picked up; characters that can be talked to or attacked; icons and menus). To facilitate identification of “active” objects, games often use such techniques as highlighting of objects or icons (figure 6) and/or changing icons and cursors when an option is available (in figure 7, the cursor has changed from a dagger to a gauntlet).

Figure 6. Dagger cursor near highlighted object.

Figure 7. Ccoins highlighted and the cursor transformed into a gauntlet.

The work of the ergodic code is evident in the way it affects the other codes. The ergodic code functions almost continuously with the proairetic code, in that sequences of events and spaces that constitute the narrative units are often consequent upon the player's interaction. For example, the sequence of Christof crossing the room to pick up gold coins is only achieved through a series of player interactions.. The ergodic code also works almost continuously with the diegetic code in determining (possible) relationships between narrating agents, focalisers and what is focalised. Indeed, the options available to the player during a particular gaming sequence depend upon focalisation. A player of V:TM cannot see an object unless Christof is nearby and the camera angle is correct, and Christof cannot pick up the object until the cursor is in the right spot.

The ergodic code works with the semic code when a player's actions trait him/her in a particular way (for example, by a defensive, aggressive, furtive or reckless style of attack). Of course, while in many games the player is already identified as a generic type (hero, soldier, thief and so on), the tendency is for games to allow players to select from different characters. In CRPGs and MUDs, the player is not only able to choose the “alignment” of a character (for example, good/evil/lawful/chaotic/ neutral) but a whole range of physical or behavioural traits. During gameplay, player choices also form part of the text's reference code. For example, when, in V:TM , the player makes Christof perform a “good” deed (helping people), this choice functions both as an index of an ideology of Christian ethics, and of the player's own conduct. When a player's conduct is inconsistent with the character's alignment (when goodly Christof kills an innocent person), there may be consequences (Christof loses “humanity”). In short, the player's ideological predispositions become part of the cultural codes referenced by, and incorporated into, the game.

The hermeneutic and symbolic codes often function simultaneously with the ergodic code when a player's choice affects the unfolding of the text's macrostructure and symbol transformations. In linear adventure and RPG games (such as Diablo ), this principally involves the activation of a pre-planned sequence of narrative developments and conflicts. In games that offer non-linear elements, the ergodic code functions with the hermeneutic and symbolic code by irreversibly closing off possible sequences and/or symbol transformations. Furthermore, even if a player is interacting almost continuously with the game, there are only a few key moments when their choices will produce a nuclei that will develop the narrative macrostructure. Nonetheless, the action of the player may function almost continuously to reinforce symbolic oppositions. For example, the act of repeatedly striking a key that makes Christof attack an vampire will act out an opposition between the Good Protagonist Christof (who symbolises humanity, faith, reason and so on) and the Evil Antagonist (who symbolises inhumanity, faithlessness, madness and so on).

3.2.2. The Computational Code

The computational code refers to the software (program code) and hardware (platform) which generates the game text (see Anderson, Holmqvist and Jenson (1993) for a theorisation of computers and computer programs as sign systems). The computational code involves the linear processing of (digital) code according to Boolean logic, though some games may facilitate fuzzy (analogue) feedback systems. While many game engines are patented and re-used, each game has its own unique computational code which declares variable parameters and type (narrative-units, classes of event sequences, available character traits), including well-defined oppositions (win/lose, alive/dead) and event-conditions (kill the enemy, find the object, capture the flag).

The difference between the ergodic and computational code is analogous to the difference between what Turkle calls a culture of calculation —characterised by a recognition of the “depth” of the computers inner workings (1995, p. 32)—and a culture of simulation —characterised by the Windows interface, and a tendency to take things “at (inter)face value. . . . if it works for you, it has all the reality it needs” (p. 24). In a culture of simulation, players are not interested in underlying mechanisms and learn by experimenting (“tinkering”) with the interface. The ergodic code collapses into the computational code when interaction with(in) the game occurs at the same level of abstraction as software (for example, when MUD players program objects or a player hacks the code), but since gameplay is performed at the level of the interface, the computational code may be provisionally bracketed out when analysing gameplay.

3.2.3 The Kinaesthetic Code.

While the ergodic code governs the player's ability to intervene in the generation of the game text, its operation is generally symbolic. The kinaesthetic code, by contrast, refers to the distinctive experience of gameplay. While Darley refers to the kinaesthesia of gameplay, Turkle has elsewhere referred to the “altered state” of gameplay: “Call it ‘muscle memory,' call it ‘flow,' call it ‘trusting your instincts' . . . [s]killed video game players experience [an] immediacy of knowing their game with more than their head, and the experience is exhilarating” (1984, p. 81). Friedman, on the other hand, describes the experience of gameplay by appropriating Donna Haraway's use of the term “cyborg” (cybernetic organism). The term cyborg, used in this sense, does not refer to the physical augmentation of the player by technology, but to the cybernetic circuit the player forms with the computer, and consequent feeling that t he computer is “like an organic extension of one's consciousness” (1995, p. 83). The term is more apt for PC games, which are focused around a keyboard and mouse and tend towards more cerebral modes of interaction than arcade, console or VR systems.

While the ideal of kinaesthesia in a game may be physical experience in the “real” world, and some VR technology may come close to reproducing it (Holmes, 1997), most computer games offer a poor simulation. Furthermore, the kinaesthesia of gameplay emerges from a negotiation between the physical experience of interacting with the hardware (the keyboard, mouse, monitor, joystick, gamepod, 3D projected screen, VR gloves and/or helmet, and so on) and the experience of interacting with the virtual space of the game world (the identification of spatial sequences and the recognition of the player's location in the narrative space, governed by the proairetic and narrative codes, respectively).

For Aarseth (1998), the virtual space of a game is defined principally by “representations of space”; he dismisses what Lefebvre refers to as “representational spaces”—“ space as lived through its associated images and symbols” (Lefebvre, 1991, pp. 39)—as primarily aesthetic. However, the effect of representational spaces (places) on the experience of the game's virtual space is relative to the centrality and depth of narrative during a gaming sequence. The semic code will define places in relationship to characters' traits; the hermeneutic code will define spaces as significant in terms of their relationship to plot developments; the symbolic code will define spaces in terms of sequences of oppositions; the reference code will refer to “real” spaces outside the narrative, and the diegetic code will identify spaces in terms of their verisimilitude. The ergodic code will determine to what extent the player participates in the generation of space, and, along with the computational code and kinaesthetics codes, constitutes the “virtual” space of the gaming experience.

4. Conclusion

Analysing the work of the above codes provides a means of tracing the non-linear processes of distribution and integration in computer games. The usefulness of such an approach is that the processes of interaction in computer games cannot be neatly compartmentalised: any of these codes may potentially be at work in any game and become central to its gameplay during a particular sequence. It is the complex negotiation of the multiple codes of narrative and gaming (and of different forms of spatiality) that determines the distinctive character of any act of gameplay.

However, this methodology is not an end in itself. An archaeology of computer game discourses (Darley, 2000; Jarvinen, 1998) is necessary to explain why different player types, styles of gameplay and computer game genres are associated with some codes more than others. The methodology above merely points to problems with existing analyses of computer games and provides a basis for a more complex analysis of the relationships between game structure, processes of interaction, and the discourses in which gameplay is situated.

References

Aarseth, E. (1997). Cybertext: Perspectives on Ergodic Literature [on-line]. Available WWW: http://www.hf-fak.uib.no/cybertext/Ergodic.html [2001, 4 April]

Aarseth, E. (1998). Allegories of Space: The Question of Spatiality in Computer games [on-line]. Available WWW: http://www.hf.uib.no/hi/espen/papers/space.html [2001, 4 April]

Anderson, P. B., Holmqvist, B. & Jensen, J. F. (Eds). (1993). The computer as medium. Cambridge: Cambridge University Press.

Barthes, Roland. (1975). S/Z . (R. Miller, Trans.). London : Jonathan Cape.

Barthes, Roland. (1982). A Barthes reader . (S. Sontag, Ed.). New York : Hill and Wang.

Baym, N. K. (1995). The Emergence of Community in Computer-Mediated Communication. Cybersociety. Computer-Mediated Communication and Community . Thousand Oaks, California: Sage Publications.

Cassel, J. & Jenkins, H. (1998). From Barbie to Mortal Kombat: Gender and Computer games . Cambridge, Massachusetts: MIT Press.

Cohen, S. & Shires, L. (1988). Telling Stories: A Theoretical Analysis of Narrative Fiction . London: Routledge.

Culler, J. (1975). Structuralist Poetics: Structuralism, Linguistics and the Study of Literature . Ithaca, New York: Cornell University Press.

Darley, A. (2000). Visual Digital Culture : Surface play and spectacle in new media genres . London: Routledge.

De Certeau, M. (1984). The Practice of Everyday Life . (S. Rendell, Trans.). Berkeley: University of California Press.

Durkin, K. (1995). Computer Games Their Effects on Young People: A Review . Sydney: Office of Film and Literature Classification.

Frasca, G. (1999). Ludology Meets Narratology: Similitude and differences between (video)games and narrative [online]. Available WWW: http://www.jacaranda.org/frasca/index2.htm [2001, 4 April].

Friedman, T. (1995). Making Sense of Software: Computer Games and Interactive Textuality. In S. G. Jones (Ed.), Cybersociety. Computer-Mediated Communication and Community , (73-89). Thousand Oaks, California: Sage Publications.

Fuller, M. & Jenkins, H. (1995). Nintendo and New World Narratives. In S. G. Jones (Ed.), Cybersociety. Computer-Mediated Communication and Community , (pp. 57-72). Thousand Oaks, California: Sage Publications.

Genette, G. (1980). Narrative Discourse . Oxford : Basil Blackwell.

Griemas, A. J. (1987). On meaning selecting writings in semiotic theory . (P. J. Perron & F.H. Collins, Trans.). Mineapolis: University of Minnesota Press.

Holmes, D. (Ed.). (1997). Virtual Politics. Identity and Community in Cyberspace . London : Sage Publications.

Herz, J. C. (1997). Joystick nation : how video games ate our quarters, won our hearts, and rewired our minds . Boston: Little Brown and Company.

Hutchinson, P. (1983). Games Authors Play . London: Methuen.

Järvinen, A. (2000). Game Archeologies [version 2.0] [on-line]. Available WWW: http://www.uta.fi/~tlakja/GA/index.html [2001, 4 April].

Juul, J. (1998). A clash between game and narrative: Interactive fiction (Why computer games do not tell good stories and why this is not a problem.) [online]. Available WWW at: http://cmc.uib.no/dac98/paper/juul.html [2001, 4 April].

Kasvi, J. J. J. Games, Gamers and Gaming Culture. Thoughts and studies on the computer and video game phenomenon [on-line]. Available WWW:

http://www.knowledge.hut.fi/projects/games/gamelinks.html [2001, 4 April].

Kinder, M. (1991). Playing with power in movies, television and video games: From Muppet Babies to Teenage Mutant Ninja Turtles . Berkeley: University of California Press.

Kolko, B. E. (1994). Building a World with Words: The Narrative Reality of Virtual Communities [on-line]. Available WWW: http://www2.iup.edu/en/workdays/KOLKO.HTML [2001, 4 April].

Landow, G. (1992). Hypertext . Baltimore: Johns Hopkins University Press.

Lanser, S. S. (1981). The Narrative Act: Point of View in Prose Fiction . Princeton, New Jersey: Princeton University Press.

Lapsley, R. & Westlake, M. (1988). Film Theory : An Introduction . Manchester, United Kingdom: Manchester University Press.

Lefebvre, H. (1991). The Production of Space . (D. Nicholson-Smith, Trans.). Oxford: Basil Blackwell.

Myers, D. (1990). A Q-Study of Game Player Aesthetics. Simulation and Gaming, 21 (4), 375-396.

Myers, D. (1991). Computer Game Semiotics. Play and Culture, 5 , 334-345.

Reid, E. (1994). Cultural Formations in Text-Based Virtual Realities [on-line]. Available WWW: ftp://ftp.lambda.moo.mud.org/pub/MOO/papers/ CulturalFormations.txt [2001, 4 April].

Skirrow, G. (1990). Hellivision: An Analysis of Video Games. In M. Alvarado and J. O. Thompsons Eds.), The Media Reader (pp. 321-338). London: BFI Publishing.

Stallabras, J. (1993). Just Gaming [on-line]. Available WWW: http:www.cc.rochester.edu/College/FS/Publications/Stallabras.html [2001, 4 April].

Turkle, S. (1995). Life on the Screen. Identity in the Age of the Internet . New York: Touchstone.