Agency: Responses to the DIGRA session
These applications of the term aren't uniform. Another problem.
We keep coming back to it in circles. Yet another problem.
However, some good work is being done on it.
At least, we recognise that all the “satisfying power to take meaningful action and see the results of our decisions and choices” etc isn't really an explanation. Neither is the formal and material cause idea useful mainly because it still leaves too much to human volition.
Noah Wardrip-Fruin brings in the discussion of the machine's agency and sort of links with the actor-network theory (see his paper). I kind of like the ways in which he identifies the 'waxing and waning of agency' and how he slashes the simplistic proportionality equation of realism and agency.
However, I do not think that this approach adequately discusses the aleatory element in games (especially for me, in RTS) and how that affects and upsets the choice equation. Also not sure it accounts for the process of moment-to-moment perception of choice by the player and also the perception by the player of choices exerted by the machine (connection also with involvement and also context: a deeply involved player thinks he is fighting the Roman army and not a program on his pc). Further, I think that the term 'agency' itself is problematic especially with its humanist connotations (and indeed the morass of other connotations that relate it 'free will' and a whole lot of other things). In any case, it is problematic whenever it implies 'pure' choice. In videogames, the human choice is also a non-choice (on the level of the machine). Justin Parsler's very lucid explanation of how he uses the concept of weak agency (and strong agency) is useful here. In a very inspired smoking break, in the presence of Miriam Eladhari, a very informed academic whose name I forget and yours truly, Parsler described 'weak' agency as the choice to choose between a certain number of shirts in a shop and not to make the initial choice of what shirts are going to be there. In a sense, I think that this is the case on even other levels of freedom and choice. For example, even the number of shirts that can be made for you depends on the availability of the material and labour and so on. Choice is never the absolute free will claimed by Renaissance Humanists like Pico della Mirandolla ('Oration on the Dignity of Man'). Neither is it free will given by Divine Grace as a one time dispensation (i.e. God as Supreme Being makes a one-time grant of a mechanism of free choice that humanity can use until Doomsday) . Nevertheless, we experience the sense of choice and do not live in a wholly deterministic world, whether that of a Calvinist God or that of Agent Smith. Please, a subtler understanding of agency is required.
An 'extra value' airtight definition of agency might be available soon on superstore shelves in varicoloured cans. The cans, however, may be empty most likely, caveat emptor. If game researchers wish to use the term 'agency' in a qualified sense that accounts for the process of experiencing choice while recognising machine constraints, then the sense in which they define agency needs to be spelled out with no ambiguity. Wardrip-Fruin does this quite well and this sets his analysis apart from the sweeping generalisations made by some earlier commentators. However, it is important that game researchers don't only speak to themselves. When we talk to outsiders, the sense of what we say should not mislead them into construing the process as one of human-centred choice. To avoid any liberal humanistic connotations, I use the value-neutral term 'action' as the starting point for my analysis of the processes (choice-making/acting/experiencing) that go on in the game-machine-human complex.
I find it easier to look at agency from a Deleuzian perspective in my PhD dissertation. After yesterday's session, I feel I must at least state the issues I perceive with the problem of 'agency' in videogames and especially so, since recent scholarship is also tending to move in a similar directions. The aim is to enter the discussion- with yet another point of view.
Below is an extract from Chapter Seven, 'Playing in the Zone of Becoming I: Agency and Becoming in the Videogame, from my PhD thesis, The Zone of Becoming: Game, Text and Technicity in Videogame Narratives (Nottingham Trent University, 2009), pp. 220-260. Please cite with my permission.
Murray’s conceptions of agency have provoked much critical response. The above analysis of procedural authorship clearly shows that the action in videogames occurs in a process of interaction between player and machine rather than being located as embedded agency. Further, contrary to Murray’s anthropocentric model, the player as the protagonist or subject is not homogenous and absolute; neither is the participation in a game just a wearing of a mask or a journey into the Holodeck, as will be shown subsequently, here, and in the following chapter. In fact, how much of the playing-subject is human and how much machine is a moot question. Considering these issues, a theory of an anthropocentric and embedded agency is insufficient in explaining the process of action in videogames. Recent commentators, therefore, argue against this model and also take into account the issue that Murray calls attention to but does not pursue: the ‘call of the machine’. Critics like Atkins and Krzywinska express their scepticism about earlier conceptions of agency and Atkins briefly refers to these as the ‘illusion of individual agency’,i a phrase that will be significant in the subsequent sections of this chapter. A more sustained criticism, however, has been made by critics like Poremba and Susana Tosca who approach the problem from different perspectives. It will be instructive to identify these approaches first because they create the base for creating the model of ludic action in the subsequent sections.
Poremba quite clearly argues against the earlier conception of ‘embedded agency’ which is how she identifies the model proposed by Murray and followed by theorists like Klastrup. Commenting on GTA III, Poremba concludes that its agency is ‘difficult to attribute [and can be seen as] lying somewhere in a nebulous region between player, designer and system’.ii Though an issue such as agency, which has always been hotly debated in other contexts, obviously attracts a lot of controversy, recent game studies criticism generally is in consensus with the description above. The so-called nebulous region has, of course, attracted much critical attention and this chapter will also attempt to locate and explore the zone of ludic action and agency.
In fact, though she does not mention it, Poremba’s account clearly illustrates a Derridean supplementarity between the various situations in which agency might be possible within a game. The phrase ‘situations in’ has been purposely chosen over ‘types of’ as a reminder that this description does not aim to divide agency into separate types with different ‘ordered centres’. The various elements associated with agency — the player, the designer and the machine — are not distinct entities. In fact, Poremba’s analysis reveals that they cannot be characterised as originary and derivative as Murray’s model does. Poremba states that ‘further work needs to be done to explore new models of agency that accommodate a more complex relationship between game designer, player and the game itself’.iii While the issue could not have been better expressed, the term ‘agency’ still poses problems especially because of its connection with human-centred choice and the problems of reconciling this with the bipartite process of action in videogames.
Murray’s conceptions of agency have provoked much critical response. The above analysis of procedural authorship clearly shows that the action in videogames occurs in a process of interaction between player and machine rather than being located as embedded agency. Further, contrary to Murray’s anthropocentric model, the player as the protagonist or subject is not homogenous and absolute; neither is the participation in a game just a wearing of a mask or a journey into the Holodeck, as will be shown subsequently, here, and in the following chapter. In fact, how much of the playing-subject is human and how much machine is a moot question. Considering these issues, a theory of an anthropocentric and embedded agency is insufficient in explaining the process of action in videogames. Recent commentators, therefore, argue against this model and also take into account the issue that Murray calls attention to but does not pursue: the ‘call of the machine’. Critics like Atkins and Krzywinska express their scepticism about earlier conceptions of agency and Atkins briefly refers to these as the ‘illusion of individual agency’,iv a phrase that will be significant in the subsequent sections of this chapter. A more sustained criticism, however, has been made by critics like Poremba and Susana Tosca who approach the problem from different perspectives. It will be instructive to identify these approaches first because they create the base for creating the model of ludic action in the subsequent sections.
Poremba quite clearly argues against the earlier conception of ‘embedded agency’ which is how she identifies the model proposed by Murray and followed by theorists like Klastrup. Commenting on GTA III, Poremba concludes that its agency is ‘difficult to attribute [and can be seen as] lying somewhere in a nebulous region between player, designer and system’.v Though an issue such as agency, which has always been hotly debated in other contexts, obviously attracts a lot of controversy, recent game studies criticism generally is in consensus with the description above. The so-called nebulous region has, of course, attracted much critical attention and this chapter will also attempt to locate and explore the zone of ludic action and agency.
Poremba’s account is representative and thorough. She argues for a model of agency that will account for the game designer’s agency, player agency and the emergent and artificially intelligent system’s agency. Besides making conscious choices to explore, configure, experience and react with the guided environment of the game system, the player often subverts this environment by using external tools (additions or modifications to the game’s code) or by exploiting latent possibilities in the game’s code (as in the ‘Hot coffee’ mod in GTA: San Andreas) or in its logic (the ‘hooker cheat’ in GTA III, as mentioned by Poremba). In all the cases mentioned above, player agency is possible only in response to the ‘call of the machine’. The modification and subversion of gameplay certainly falls under the category of constructivism described by Murray but in that case it is necessary to realise that this is a machinic constructivism. An awareness of the machinic affordances is not only required for modifying and subverting gameplay; it is essential for the process of play itself. As Wright, Boria and Breidenback, in their analysis of creative player actions in online FPS videogames, make it clear, ‘Playing is not simply mindless movement through a virtual landscape, but rather movement with a reflexive awareness of the game’s features and their possible modifications’.vi Poremba supports their conclusion in her essay on agency in GTA III and maintains that this is indicative of the fact that agency in games needs to be seen in terms of newer models which move the analysis beyond the limitations inherent in the notion of embedded agency. She also states that player agency and designer agency are not discrete binaries but rather they exist as interdependent categories. According to her,
Game designers have expressed pleasure in player’s creative actions — even ones that clearly go against design intention and extend the boundaries of the game. Conversely from a player perspective, gameplay is often about determining what the game designer wants (i.e. how to play the game) rather than a constant drive for increasing agency.vii
This assertion illustrates a clear shift in the understanding of procedural authorship from Murray’s separation of the design perspective and the gameplay to a more supplementary relationship between the two. In fact, though she does not mention it, Poremba’s account clearly illustrates a Derridean supplementarity between the various situations in which agency might be possible within a game. The phrase ‘situations in’ has been purposely chosen over ‘types of’ as a reminder that this description does not aim to divide agency into separate types with different ‘ordered centres’. The various elements associated with agency — the player, the designer and the machine — are not distinct entities. In fact, Poremba’s analysis reveals that they cannot be characterised as originary and derivative as Murray’s model does. Poremba states that ‘further work needs to be done to explore new models of agency that accommodate a more complex relationship between game designer, player and the game itself’.viii While the issue could not have been better expressed, the term ‘agency’ still poses problems especially because of its connection with human-centred choice and the problems of reconciling this with the bipartite process of action in videogames.
Commentators such as Atkins suggest that the experience of agency is illusory; the chief reason for this is a reaction to Murray’s notion of agency as free choice. Susana Tosca examines this issue in detail through a critical analysis of the Blade Runner game (1997) created by Westwood Studios. Blade Runner is an interactive adventure game — one of the last of its kind; though it wasn’t commercially as successful as its FPS rivals such as Quake 2 (1997) and Half-Life (1998), it still has a considerable fan-following and figures in many game studies analyses. It requires the player to play as Ray McCoy, a blade runner employed to ‘retire’ replicants; McCoy is similar to Deckard, the protagonist in Ridley Scott’s film and Philip K. Dick’s novel. The issue of whether to have sympathy for the replicants or to kill them, a major philosophical question in both the book and the film, is incorporated into the game as player choice. The game has thirteen different ‘official’ endings which depend on what chain of actions the player follows in the game. Player choice is, therefore, responsible for determining the player’s character within the game as well as the fate of the various characters. This is how it looks from the player’s point of view but that, however, is not the only perspective. Louis Castle, the designer of Blade Runner, in an interview with Pearce, describes how this works from the point of view of the game:
If you play the game as if you are a replicant, then the game treats you as a replicant. If you play the game as if you were a Blade Runner human, it treats you like you’re a human. So people perceive that at some point they’ve made a choice that puts them on one track or the other, which isn’t the case at all. It’s based on how you play the game, whether you hunt the replicants, whether you kill them, whether you let them go. Those things give us clues as to what you think you are—and at any given point, you can switch over. You can go halfway through the game and go "Oh, my gosh, I’m really not a human after all, I’m a replicant." And just turn mid-stream and start saving the replicants. And that’s okay. The game lets you do that.ix
In the above comment, the way the game constructs the playing subject is important. Castle’s language, especially his usage of phrases like ‘the game treats you’ or ‘the game lets you do that’ clearly indicates that the game is also an actor or a player. For the (human) player, the choices she makes may seem all important- they may even seem to reflect the player’s character. For the game’s logic or algorithm, the case is different. Here the response is input-based, as Castle states.
The subject is determined by the actualisation of technical choices. Tosca makes a similar point in the following comment:
Each action matters towards the end and that we contribute to the evolving story as we go. Trying to guess which actions those are, and how they lead to each conclusion, is a sort of narrative reverse engineering where, in my opinion, the pleasure of the game lies. And once we know, of course, we can always exert our free will and choose another path.x
Tosca’s statement is important because it highlights a dichotomy. First, there is the idea of each action contributing to the evolving story. This is part of the process of configuring and interacting with the game’s algorithm. Hence agency seems to involve both the player and the game algorithms together with their technical affordances. Tosca’s idea of the process of back calculation or as she calls it ‘narrative reverse engineering’ is also in consonance with this kind of agency in that such activity still involves the game’s logic as an equal partner in the process. The problem arises, however, when she speaks of exerting free will to choose another path. This sounds as if it is arriving at the same conclusions as the earlier conceptions of embedded and anthropocentric agency. Tosca’s qualifying comment in a later statement, however, shows a contrary position: ‘Blade Runner creates a digital suspension of disbelief that players are willingly drawn into through the excitement of the different moral choices, where trusting our implanted memories will bring us the illusion of free will’.xi
This is a statement that needs careful attention: the memories that allow the player to reverse engineer or, in simpler words, to reconstruct a narrative actualisation, are not just human memories. They are also a part of the machinic memory in that they are steps in the algorithm that the game follows. In the example of Sands of Time in Chapter Six, the saved games were attributed as the Prince’s (and therefore the player’s) memories. The ‘free will’, in this context, is an illusion simply because the choices made by the player are not entirely free but rather bound to the affordances of the machine algorithm.
Once the player returns to the point of deviation in a game that is being replayed (for example, from a saved game), she encounters a series of choices and has the opportunity to exercises choice yet again. Beneath the apparent vital nature of the player’s emotional choice, which the game convincingly portrays, lie the game choices and these are primal in determining the path of actualisation. The player perceives moral choices and memory whereas the game algorithm contains its algorithmic choices and pathways. The two coincide when, as Tosca says, there is a ‘suspension of disbelief’. The suspension of disbelief, intrinsically related (in the nature of the supplement) to agency, will merit a separate analysis in Chapter Eight.
The present discussion will return to the question of memory. Not surprisingly, Tosca uses the phrase ‘implanted memories’, a concept that is all too familiar from Blade Runner texts, to describe the experience of memory in videogames. Those familiar with the Blade Runner movie will remember the famous scene where Deckard (Harrison Ford) administers the Voigt-Kampff test to Rachael (Sean Young). At the end of the test, it is revealed that Rachael, unknown to herself, is really a replicant. She does not know that some of her memories are not real: they are ‘implants’ from Tyrell’s sixteen-year old niece. While the player’s memories are literally not ‘implants’ as in Dick’s novel or Ridley Scott’s film adaptation, they are reconstructions of a series of in-game choices: they are as much memories as part of game algorithm. Hence, after accessing these to replay a game sequence, the player willingly becomes part of game system and executes another algorithm.
For the player, to choose not to kill replicants may be a moral choice, but it is also a choice informed by the machinic attributes of the game and its specific algorithm. For example, the player in Doom does not have the choice not to kill the monsters that appear in the game. It is of course possible to subvert the original game using cheats and mods but as noted earlier, to do even this involves restrictions in the game program.
Tosca’s conception emerges as more complex than mere non-agency. The ‘illusion of agency’ most certainly includes and allows for choice. Here, choice is, however, a decentred phenomenon: it is not the prerogative of either the (human)player or the machine algorithm. These entities themselves occur as supplements to the other, as already observed in earlier chapters. The element of choice therefore occurs within the (human)player-machine algorithm complex. Given this supplementarity within which choice operates in videogames, it is possible to relate this to the earlier examples of supplementarity between writing and reading or game and play where the elements in the relationship are all in-play. Even in conceptualising agency and choice in videogames, it is possible to see them as being in-play. This notion has significant implications in the way the phrase ‘illusion of agency’ can be read. By ‘illusion of agency’ something different is to be inferred. The use of the word ‘illusion’ here is perhaps fortuitous but it serves the purpose marvellously. The etymology of ‘illusion’ (as derived from ‘illude’, which can mean ‘make sport of’, albeit used pejoratively) contains the Latin root ludere or ‘to play’.xii It is possible to read the term differently from what was perhaps the intended meaning: one can read ‘illusion of agency’ as the "making ludic of agency" and this reflects the process of interaction and response between the (human)player and the game algorithm. In the case of videogames, it is important to remember that the game is also an artificially intelligent machinic algorithm. The possibility of choosing the action in videogames is therefore always related to the ‘call of the machine’.
From Agency to Becoming: A Deleuzian Understanding of Choice in Videogames
The altered conception of agency, as described above, marks a major shift from the earlier human-centred concept of free will to a relationship between the player and the machine that can be more clearly understood in terms of a bipartite process of action. Commentators such as Galloway already have already started thinking about the bipartite process as being a supplementary one. For him,
One may start by distinguishing two basic types of action in videogames; machinic actions and operator actions […] Of course, the division is entirely artificial — both the machine and the operator work together. […] The two types of action are ontologically the same.xiii
Galloway quite rightly identifies the importance of studying the action in videogames as a more accurate way of analysing gameplay. He stresses, almost axiomatically, that ‘if photographs are images, and films are moving pictures, then video games are actions. Let this be word one for video game theory’.xiv While maintaining the importance of an action-based approach for game studies, Galloway notes that there is no clear division between machine and operator actions. This account also illustrates the supplementary relationship described above. In the first chapter of his book, Gaming: Essays in Algorithmic Culture, Galloway launches directly into a discussion of action, in digital games as being performed ‘step by step [and] move by move’xv by operator and machine. As the base foundation of his analysis, he reads games in terms of the ‘action-image’ as described by Deleuze. However, he does not engage with the concept of videogame action within a Deleuzian framework in any detail. The importance of the concept makes it merit further analysis and it will be seen in this and the following chapter that the process of involvement of the player and the ludic action that characterises gameplay finds its best explanation when analysed within a Deleuzian framework.
The analysis of ludic action within a Deleuzian framework, however, may be opposed by various commentators. As mentioned earlier, Bogost’s objection to such an analysis was that the ‘local operations’ within such a ‘nomadic’ structure would deny any factor of deliberation in digital games. For him, it is difficult to locate agency in the workings of the Deleuzian manifold since he sees the multiplicity as being characterised essentially by the element of the aleatory. Such a reading of Deleuze is open to contestation.
Bogost, however, is not alone in his objection. Hayles, quoting Mark Hansen, notes that ‘Deleuze and Guattari are much more thoroughgoing in their deconstruction of the liberal humanist subject and of "subjectification" in general. As Mark Hansen comments, "D+G do not shift the locus of agency [... but] dissolve the role of agency altogether"’.xvi She, however, adds that ‘they too recuperate agency at crucial points […] They warn the reader against giving up agency altogether’.xvii Hayles agrees with Hansen that Deleuze and Guattari wish to deny agency but she maintains that they cannot avoid it because ‘through their performative language, they exercise agency even as they deny it […] Deleuze and Guattari cannot avoid inscripting into language, the agency implicit in their performance of desire’.xviii While she is right in stating that Deleuzoguattarian theory does take into account the exercise of agency, her assertion regarding its intention to deny agency is controversial. Hayles’s argument is drawn from her reading of A Thousand Plateaus where Deleuze and Guattari do not directly address issues of agency. Such a reading misses the more direct analyses of agency and subjectivity in Deleuze’s earlier works, such as his treatises on Hume and Spinoza, which also play a key role in shaping the main body of his work including the texts where he collaborated with Guattari. Aurelia Armstrong, commenting on Deleuze’s modification of the Spinozist conception of agency states that in Deleuzian (and indeed, Deleuzoguattarian) thought, quite differently from earlier notions, ‘agency is conceived of as a movement, which evades the definition of the individual in terms of forms and functions and the delimitations of its capacities, whether such a definition is biological, psychiatric or political’.xix Armstrong further maintains that the ‘growth of agency is shown to consist in a becoming-active, in the increase and enhancement of “individual” powers through their combination with the powers of other, compatible individuals and things’.xx This is obviously quite different from liberal humanist notions in which agency is situated as the free choice of the individual; it is also equally different from the totally aleatory scheme of events.
In the analysis of temporality in Chapter Six, the Deleuzian idea of the manifold was compared to phase portraits of molecular movements where the population of trajectories as a whole influence the course of any action. Agency should be seen as an analogous and related experience. In an emergent structure, agency can only be thought of in terms of the options for acting within a framework of the constraints imposed by the actions of connected elements. Further, the concept of ‘becoming’, which runs as a key theme throughout the whole thesis, is equally important in speaking of agency. True, agency is action but it is actually the ‘becoming-active’; in this process, the individual’s subjectivity is experienced in a complex manner due to the actions performed by her within the system. ‘Becoming’ has already been introduced in Chapter Two as the ‘zone of indiscernibility’xxi occupied by the subject: the player in the computer game does not act as if free of her machinic persona and neither does she get totally absorbed in such a persona. Instead, as explained in the subsequent chapter, her experience can be described as a ‘becoming’. In game studies, the concept that corresponds most to this is well-known as ‘immersion’. The subsequent analysis will, however, indicate the problems in seeing this as being a separate phenomenon. Instead, both immersion and agency need to be viewed as merged concepts that constitute the core of the process of ‘becoming’. As already discussed in the context of videogames, an altered conception of agency is being put forward here: this conception is based on action and on movement or ‘becoming’ and it moves beyond the more traditional ways in which game studies and other analyses of machinic media conceive of agency.
However, it is obvious that despite their apparent differences with Deleuze, both Hayles and Hansen are in agreement regarding the two aspects of agency described in Deleuze. Total free will for the (human) player is not the case in videogames because of the pervasive presence of the (machine) algorithm and because during gameplay, the machine can also be considered a player and the human player a part of a certain algorithmic sequence. The first issue would be the emergent patterns present in videogames that preclude any totally determined act on the part of the human agent. Secondly, the human agent, in becoming part of the game experiences a complex subjectivity that any conclusion of pure agency difficult to envisage. Both of these issues are described in Deleuze’s formulation of the action-image in the ideas of action as actualisation and as resulting in a ‘new mode of being’ for the agent.
In fact, it might be argued that Deleuzian ideas of agency are not so different from Hayles’s own, especially when seen in a broader Deleuzian context. Hayles maintains that ‘if the posthuman implies distributed cognition, then it must imply distributed agency as well, for multiplying the sites at which cognizing can take place also multiplies the entities who can count as agents’.xxii Her position is similar to that of Poremba and Tosca, described above. It is also the point of entry to Galloway’s application of the Deleuzian action-image to videogames and to its extension to discussions of agency. Distributed agency is seen as resulting from distributed sites of cognition. This is similar to the Deleuzian explanation provided by Armstrong: agency can only be conceived of in connection with the actions of connected elements; hence, to use Hayles’s term, it is ‘distributed agency’. More needs to be said about distributed agency in the subsequent discourse on the action-image. From this analysis, it is possible to conclude that the Deleuzian framework used in this thesis does not support a denial of agency as some critics suppose; instead, it effectively brings together the different aspects of the discussions on agency and helps view the process within a more representative framework. Nevertheless, within this framework, the earlier approaches need to be sufficiently modified and some significant changes must be made. The first of these would be to replace the term ‘agency’ itself.
The analyses of the computer game narrative show that the process of gameplay is not deterministic from the point of view of either the human or the machine, but the use of the term ‘agency’ gives it that connotation, especially when considered in the light of its liberal humanist history. The subsequent analysis will, therefore, use a more representative term for the process and one that is well supported by the Deleuzian framework that provides adequate tools for studying the process; the concept in question is ‘action’.
Well, that's pretty much what my position is and I am posting this in response to yesterday's speakers as well as to share ideas with Justin. I am happy to enter into a more extensive discussion.
'Literature is a combinatorial game that pursues the possibilities implicit in its own material [...] but it is a game that at a certain point is invested with an unexpected meaning' - Italo Calvino