Secret agency is at issue in Frasca's response, which denies the application of Aristotle to the open-ended interactivity of gaming.
Aristotelian interactive drama has been the holy grail of many theorists and designers for more than a decade. Certainly, Mateas' strategy of combining Brenda Laurel's ideas with Janet Murray's is an interesting one. However, the concept itself of Aristotelian interactive drama still does not convince me. Its main problem is that the nature of traditional drama is ontologically different from so-called "interactivity." Aristotle thought of drama as something that is performed for an audience. Agency, as Mateas affirms, implies that "the player should be more than an interactive observer." When the user is playing interactive drama, she does not care about her experience attaining an "organic whole" because immersion prevents her from taking an outside view of her actions. As Joseph Bates' bus station experiment suggested (1993), players do not care about plot incoherency as much as external observers do.
Basically, I find in Mateas' proposal the same problems that Espen Aarseth [Aarseth 1997] pointed out about Brenda Laurel's interactive drama model. For example, the notion of Aristotelian closure does not fit into the computer environment. Aarseth wonders who should decide the ending: the user or the system? If the user is enjoying the experience, why should the system put an end to it? And who should select the genre: the computer or the player? In addition to this, Aarseth observes that human dramatists do not have to build several coherent alternatives of a play that comply with Aristotle's requirements. The fact that this task would be extremely hard for a creative human makes very it unlikely that, given the current state of AI, a computer could achieve it in a satisfactory way.
I agree with Mateas when he describes the imbalance problems of adventure games but I am not able to find a "sweet spot" between agency and plot simply because those two elements do not form a continuum. Narrative plot is a structure for spectators. Time works differently in narratives and games, as Markku Eskelinen [in First Person, and in Eskelinen 2001] suggests.
If, as Mateas affirms, players "should not be over-constrained by a role" and they should be encouraged "to be themselves" then they will expect a degree of freedom of action that is incompatible with Mateas' goals. Imagine, as game designer Tim Schafer once suggested to me, that you wanted to create a game where you play the role of Gandhi. How would you give agency to players while preventing them from turning peace-loving Gandhi into a Quake -like killing machine? The traditional solution would be to put authorial constraints and, for example, prevent the player from using potential weapons. Of course, you can try to fool the player by tricking him into doing what the author wants her to do. But, as Mateas admits, this option fails after the software is used a couple of times. I can only envision two possible solutions to this dilemma. You could either kill replayability by creating disposable software that could only be experienced once [Frasca 2001] or you could build a non-immersive - and therefore non-Aristotelian - environment where players would not "be themselves" but rather encouraged to become aware of their own performances while trying to perform coherently to their character's personality, as some professional RPG players do.
I personally see the potential of the computer not as something that can extend representation and narrative but rather as a laboratory where players could experiment while building and deconstructing the rules of simulated systems. The twentieth century produced interesting alternatives to Aristotelian drama that are built around agency and therefore do not need major modifications to be applied to software design. Non-Aristotelian dramatists such as Augusto Boal provide a much robust theoretical and practical environment for creating interactive pieces, as I have suggested in my essay in First Person.
Nevertheless, I salute Mateas and Stern's effort in building Façade because, unlike most theorists - and I include myself on this list - these authors are actually building a piece of software based on their research. This is quite unusual in a field where both industry and academia are busy cloning the same old ideas over and over.
References
Aarseth, Espen (1997) Cybertext: Perspectives on Ergodic Literature. Baltimore: The Johns Hopkins University Press
Eskelinen, Markku (2001) "The Gaming Situation" in Game Studies 1. http://www.gamestudies.org/0101/eskelinen/
Frasca, Gonzalo (2001) "Ephemeral Games: Is it barbaric to design videogames after Auschwitz?" in Cybertext Yearbook 2001. Research Centre for Contemporary Culture, University of Jyväskylä. Also available at: http://www.jacaranda.org/frasca/ephemeralFRASCA.pdf
Kelso, M.T., Weyhrauch, P. & Bates, J. Dramatic Presence. PRESENCE: The Journal of Teleoperators and Virtual Environments, Vol 2, No 1, MIT Press. Also available at: http://www-2.cs.cmu.edu/afs/cs.cmu.edu/project/oz/web/papers/CMU-CS-92-195.ps