Monday, August 16, 2010

Player Autonomy and Games, part 2

This article is part two of a two part series on autonomy and choice in games. You can find Part 1 here. I owe a great deal to Kwame Anthony Appiah’s The Ethics of Identity for inspiring and shaping the majority of these posts.

Last week I discussed choice in games from a philosophical liberal perspective. In agreement with John Stuart Mill’s assessment of the value of individuality and life plans, I posited that choice in games is valuable because choice is important in fashioning happy lives for ourselves. But just as in real life, not all choices matter equally. Those derived from plans are best, even when outcomes seem sub-par or align with the decisions of others. Yet even the most ideal and self-reflective choice is influenced by numerous factors. Once again, this troubles our understanding of choice in real life and in games. What does it mean when a player’s autonomy is called into question?

Many of our valuable identities are formed while we are confined to a preexisting set of options. As Appiah points out, for a self-constructed identity to make sense and be valuable to the creator, “it must be an identity constructed in response to facts outside oneself, things that are beyond one’s own choices.” While a woman may find value in joining a women’s social group based upon her personal identity, she could also suggest her external reality as a woman constrains her autonomy in that she cannot feasibly pursue a dream career as a linebacker for the Falcons (at least not in today’s society). In game design terms, how do I differentiate and evaluate an authored constraint (NFL regulations) and a condition for a meaningful autonomous choice (joining a social group)?
Before I move on, let’s take an example from Appiah: Imagine Michael, a store owner, who is robbed at gunpoint and forced to open the store vault for his assailant.

Michael formed the intention to open the vault after considering all the options; although he wouldn’t have made his decision without coercion, he could have done otherwise. Did he form his intention autonomously? Reasonable people may differ.

Now imagine this incident was a prank gone bad. The robber was actually his close friend, and the gun was a toy water pistol. If Michael had been better informed, he would have chosen his actions differently. In a video game, almost all information is authored and controlled. How much information do players need to make meaningful autonomous decisions? Conversation choices in Dragon Age: Origins, for example, may lead to romantic relationships with NPCs. Can a person who wants to sleep with Leliana but makes the wrong dialogue choice, thereby locking themselves out of that option, be considered to have autonomy? Do they have less autonomy than someone with intricate knowledge of the immutable dialogue tree? What of someone who makes their dialogue choices unaware that romantic relationships can occur at all? In this case, information (or the lack thereof) can be both a constraint on, and a condition for, meaningful choice.
Now let’s imagine Michael had access to both his own pistol and a nearby silent alarm. With more options, Michael may have chosen differently. Does Michael have more autonomy in this case? Similarly, could we say the streamlined Mass Effect 2 gives less autonomy to its players than the first Mass Effect? Can we be said to have more autonomy, and therefore more access to meaningful choice, in a game with more weapons than another? What about increased dialogue options? When Mass Effect 2 offers a moral dilemma during a heavily authored loyalty quest, with only two options, can we say the player has only partial autonomy? “Does the rich bachelor idler have more autonomy than the family man with a full time job,” Appiah asks. “Such are the imponderables posed by talk of options.”
To answer any of the above questions with a resolute yes is to say we can have partial autonomy, an idea Appiah rejects. “The notion of partial autonomy suggests the conceptual possibility of ‘full’ autonomy: if we have difficulty making sense of the latter, we should find the former equally elusive.” He continues:

Talk of partial autonomy is an ill-fated attempt to split the difference between two standpoints: one in which I have autonomy and one in which I do not...oblivious of the distinct purposes served by each of the two standpoints.

To bring Michael back into the equation, “if we’re interested in the conditions that make people act a certain way,” Appiah says, “we tell one narrative: here’s how Michael acts under these conditions. If we’re interested in retribution and blame, we tell another: look at what that villain did to poor Michael!” In video game terms, players are always potentially autonomous, regardless of linearity, depending on what sensations game designers seek to evoke. Therefore, it is nonsensical to say one implementation of choice and autonomy in a game is better or more “immersive” than another. “When is something a condition of choosing and when is it a constraint upon choice? Inevitably, how we answer is a matter of interest-guided judgement, of what standpoint we adopt.”
How would a game undoubtedly rich in choice fit into this paradigm? Perhaps Jason Rhorer’s Sleep Is Death could be considered the closest a game has come to an unobstructed player experience. Some consider the game to be the pinnacle of cooperative storytelling. Yet this is troubling. Since Rhorer created the game’s assets, is he considered an additional storyteller within his sandbox world? What about the constraints and knowledge barriers on the player designing the environment for the other? How much cooperation must take place in order to call it cooperative? Using such vocabulary is to take one of Appiah’s “standpoints” and reveal your interests, not to prove any objective claim about storytelling. Sleep Is Death is but one way to evoke a suite of sensations, itself full of conditions and constraints.

What about a highly authored 2D side-scrolling experience like Canabalt? Players have only two choices: to jump or not to jump. Does this mean the game is less engrossing or less meaningful than a choice rich experience? Of course not. The game executes its two options in service to an expertly crafted sensation. Players can even identify with protagonists in linear narratives like this one (although Canabalt may be a poor example), so as long they have a well developed rationale. In that sense, meaningful autonomy is much like Kant’s take on free will: “We have to act as if freedom is possible even though we can’t provide any theoretical justification for it,” Appiah notes, “it can’t be explained, only defended.”
Can we say choices in games narrative matter more than choice in gameplay? Is autonomy more important in one or the other? Again, the answer depends on what questions you seek to answer and what sensations you seek to evoke. To say a player lacks narrative autonomy from the game designer is to make a statement about your explicit or implicit goals and values. A game which provides no narrative autonomy fulfills a different purpose than a decision rich experience.

Accordingly, choices can be incredibly powerful, but they need not be. Morality systems (with a few exceptions), for example, undermine meaningful choice by offering player autonomy of actions but not over belief systems. What would it mean to offer a choice of various communion wafers to an atheist? More options do not equate to autonomy, and therefore meaning and immersion either. No game designer should feel burdened by popular trends one way or the other. Above all else, respect players as autonomous beings, each who value exercising their given faculties to accomplish plans and enjoy the sensations of gaming, regardless of designer constraints.


  1. “The notion of partial autonomy suggests the conceptual possibility of ‘full’ autonomy: if we have difficulty making sense of the latter, we should find the former equally elusive.
    Talk of partial autonomy is an ill-fated attempt to split the difference between two standpoints: one in which I have autonomy and one in which I do not...oblivious of the distinct purposes served by each of the two standpoints."

    So help me out here. Is there no room of the application of a Hegelian dialectic here? How would that be oblivious of the distinct purposes served by the two standpoints? What does Appiah and/or you see as the purposes of the standpoints?

  2. @ Joshua

    To answer your question would be to step out on a very precarious limb. My understanding of Hegel's dialectic is almost non-existent. Based on my admittedly shallow understanding, I don't think Appiah would say his ideas are incompatible with Hegel's.

    The standpoints serve to mechanisms by which actors understand their actions and the actions of others. They are derived by what sorts of questions you seek to answer, and both are valuable.

  3. Maybe I'm just being thick. But I'm not understanding the message of the article. It sounds like you are saying two things.

    1. We can't pull off a real-world amount choice in games so we settle for little systems that give the impression of choice.

    2. There in no middle ground when it comes to how much choice a game offers because one either has full autonomy or no autonomy.

    3. But ultimately none of that matters because whether there is an illusion of choice or not, any kind of game can be fun.

    Or perhaps I got 2. wrong and the point is that full, and no-autonomy are non-existent categories?

  4. @ Joshua

    Sorry for being unclear. I was mostly exploring these ideas about choice by coming at it from a philosophical perspective. That is, I wanted to know why we take the value of choice for granted in games, and what that means for game design.

    My point about autonomy was meant to call out the pursuit of the "right amount of choice" in games as a useless. Having more or less options in real life is meaningful ONLY in terms of what questions you are seeking to address. In game terms, having more or less options is meaningful ONLY in terms of what sensations designers seek to evoke.

    So in that way, your reading of the article is correct. My point was to emphasize that choice in games has no inherent value, and thus should be considered only in service to other design intentions, which may be improved or weakened by the implementation of player decision making.

    There are other bits in there I was poking at, but I think that's the gist of it.