Michael Mateas responds in turn

Michael Mateas responds in turn


Narrativists vs. ludologists, material vs. formal constraints: Michael Mateas replies by identifying actors’ roles in each division.

To Brenda Laurel

I am thrilled with Brenda Laurel’s positive response to both the theoretical model and technical agenda presented in the chapter. This work obviously builds on her own pioneering work in this area. In this short note I’d like to comment on two issues: the noted inconsistencies in my use of Aristotelian causal nomenclature, and the idea that interaction must have a real influence on the plot.

Laurel notes that my statements, “formal cause is the authorial view of the play,” and “material cause is the audience view of the play,” are a misuse of the Aristotelian causal nomenclature. The actual work of authoring is correctly understood as an efficient cause, while Aristotle proposes no causal role for the audience. But what I mean to highlight by these admittedly somewhat sloppy statements is not the author or audience viewed as a cause, but rather what sort of information is directly available to author vs. audience. The author, through the act of authoring (efficient cause), arranges the elements both materially and formally. But while the material arrangement of the elements is more or less available to the audience, the formal arrangement is not. The author knows things about the play, such as why a character must be this character for this whole action (formal cause), that the audience does not. The audience must work from what is directly available to the senses, and hopefully, by following the chain of material causation, eventually recapitulate the chain of formal causation. So in referring to the “authorial view” and “audience view,” I am attempting to highlight this asymmetry in knowledge between author and audience. The chain of formal cause is available to the author in a way that it is not available to the audience. And the chain of material cause is in some sense designed for the audience, as it is the ladder they must climb in order to understand the whole action.

Similarly, a player in an interactive drama becomes an author, and thus, as an efficient cause, contributes both materially to the plot and formally to elements at the level of character on down. But these contributions are constrained by the material and formal causes (viewed as affordances) provided by the author of the interactive drama. And hopefully, if these constraints are balanced, the constrained freedom of the player will be productive of agency. In these discussions I elided efficient cause and went straight for a discussion of the material and formal causes that the act of authoring puts in place.

Andrew and I agree fully with Laurel that interaction should not be spurious but should really matter, and that to matter it should have an influence on the plot. With Façade we are trying to build something where the player “carves out” a whole action out of the space of dramatic potential represented within the system. In terms of the model, a system that provided the opportunity for much hand wringing and smiting without it having an effect on the plot would suffer from the same material imbalance of some puzzle-based adventure games, in which there is plenty to do, but no formal affordances providing guidance as to what you should do when or why. The formal affordances must communicate to the player how the smiting and hand-wringing matter, and plot changes are one way that the formal affordances “speak” to the player.

To Gonzalo Frasca

Interactive drama, in its Aristotelian conception, currently inhabits a beleaguered theoretical position, caught in the crossfire between two competing academic formations, which I will style the narrativists and the ludologists. The narrativists generally come out of literary theory, take hypertext as the paradigmatic interactive form, and use narrative and literary theory as the foundation upon which to build a theory of interactive media. Ludologists generally come out of game studies [e.g. Avedon and Sutton-Smith 1971], take the computer game as the paradigmatic interactive form, and seek to build an autonomous theory of interactivity (read: free of the English department), which, while borrowing from classical games studies, is sensitive to the novel particularities of computer games (this is sometimes described as a battle against the colonizing force of narrative theory, as Eskelinen does in First Person). Both camps take issue with an Aristotelian conception of interactive drama, finding it theoretically unsophisticated, an impossible combination of game and narrative (though of course the camps disagree on whether this should be decided in favor of game or narrative), and technically impossible. Frasca’s comments, coming out of the ludology camp, are organized around three specific objections: Aristotelian interactive drama creates an impossible-to-resolve battle between the player and the system, confuses first and third-person perspectives, and is technically impossible. (For an example of a narrativist’s objections to interactive drama, see the exchange between Bernstein and Stern in First Person.)

Frasca argues that a conception of interactive drama that attempts to create a strong sense of closure with a well-formed dramatic arc introduces a battle between the player and system for control. If the system decides the ending, we have guaranteed closure without interactive freedom; if the user decides the ending we have guaranteed freedom but possibly no closure. Further, if the player is playing a prescribed role, such as Gandhi, we either have to limit interactive freedom to maintain the player’s role (and story arc) or provide interactive freedom at the expense of the role (and story arc). Both these arguments have the following form: story means fate, interactivity means freedom (doing whatever you want), therefore interactivity and story can’t be combined. However, the whole point of my chapter is to replace the vague and open term “interactivity” with the more specific term “agency,” and to then argue the conditions under which a player will experience agency: a player will experience agency when material and formal constraints are balanced. This is not the same as “a player will experience agency when they can take arbitrary action whenever they want.” So in the case of choosing the ending of an interactive story, the player does not need the ability to make arbitrary endings happen to feel agency. A small number of authorially-determined ending configurations can still produce a strong feeling of player agency if reachable through sequences of player actions within a materially and formally balanced system. Similarly, a Gandhi story can still produce a sense of agency without providing Gandhi with a chain gun or rocket launcher. If an interactive Gandhi story left weapons and power-ups lying about, but used some heavy-handed interaction constraint (like the cursor turning red and beeping) to prevent the player from picking them up, then the experience would certainly be offering material affordances (“here’s a gun for you to pick up - oops, not really”) not balanced by the formal affordances (the dramatic probabilities of the Gandhi story), resulting in a decrease in the feeling of user agency. If, however, the Gandhi world never provided access to such weapons, and given the plot it never made sense to think of using such weapons, the player would still experience agency, even in the absence of access to plasma cannons. Interactive story designers do not have to be saddled with the impossible task of allowing the player to do whatever they want while somehow turning it into a well-formed story; creating a sense of both story and agency (interactivity) requires “merely” the hard task of balancing material and formal constraints.

Note that the neo-Aristotelian theory does not prove that if you build a system which materially balances more complex formal affordances, the player will experience both agency and “storyness.” But neither do Frasca’s arguments prove that this combination of agency and “storyness” is impossible. This is an empirical question. But the neo-Aristotelian theory has the advantage of providing a constructive plausibility argument that can inform the technical research agenda required to search for an empirical answer.

Frasca argues that Aristotelian interactive drama confuses the first-person gaming situation with the third-person narrative situation. A narrative is an already-accomplished structure that is told to a spectator. A game is an evolving situation that is being accomplished by an interactor. Since an already-accomplished static structure is not the same thing as a evolving, dynamic situation, then, the argument goes, narrative and game are fundamentally dichotomous. What this argument denies, however, is the possibility for hybrid situations. For example, the storytelling situation, in which a storyteller constructs a specific story through interaction with the audience, is such a hybrid. In this situation, the audience is both spectator and interactor, and the evolving story only becomes an already-accomplished structure at the end, yet still has story properties (e.g. interpreted in accord with narrative conventions) in its intermediate pre-completed forms. Aristotelian interactive drama is similar to this storytelling situation; through interaction the player carves a story out of the block of narrative potential provided by the system.

Finally, Frasca argues against Aristotelian interactive drama on the grounds of technical impossibility. It is very difficult for a human author to write a single drama. It would be even more difficult to write multiple dramas, in real time, in response to player interaction. Since the current state of AI is nowhere near the point of producing systems that can write good static plays on their own, then certainly interactive drama is not possible. This argument, however, assumes that an interactive drama system must have the capability to construct stories out of whole cloth, denying human authorship of the AI system itself. But any AI system consists of knowledge (whether represented symbolically, procedurally or as learned probability distributions) and processes placed there by human authors, and has a circumscribed range of situations in which the system can function. The “only” thing an interactive drama system must be able to do is represent a specific space of story potential and move appropriately within this space of story potential in response to player interaction. As argued above, the system doesn’t need to handle arbitrary player actions, but only those that are materially and formally afforded by the specific story space. While still hard, this sounds like a much easier problem than building a system that can do everything a human playwright can do and more. Incidentally, Frasca notes that Augusto Boal’s work provides a better theoretical and practical framework for constructing interactive pieces. But the Boalian technical agenda of building powerful social simulation environments in which non-programmers can use easy-to-learn languages to simulate complex social phenomena is certainly a challenging technical project, and were I so inclined, amenable to a technical impossibility argument (I confess to a dispositional tendency to believe that nothing is impossible).

In conclusion, I’d like to say that I find Frasca’s conception of Boalian “videogames of the oppressed” (as discussed in First Person) extremely interesting, and hope that he pursues this idea. I certainly don’t believe that the conception of interactive drama described in my chapter is the only proper conception of interactive story-like experiences. Nor do I believe that all interactive experiences must be assimilated to the concept of narrative. The ludologists commonly use examples such as chess, Tetris, or Space Invaders in their analyses, and I agree that such games are most profitably studied using non-narrative analytic tools (but conversely, denying any story-like properties to games such as The Last Express, Grim Fandango, or Resident Evil also does not seem profitable). However, I reject the notion that games and stories are fundamentally irreconcilable categories, that experiences which provide the player with a sense of both agency and story structure are impossible. The neo-Aristotelian theory, and the concrete system that Andrew and I are building, are a theoretical and empirical investigation within this hybrid space of interactive story.


Avedon, Elliott M. and Brian Sutton-Smith (1971). The Study of Games. New York: Wiley.

back to Cyberdrama introduction