Earlier this week, we pointed you towards an interesting paper by Georgia Tech Professor Fox Harrell, which dealt with the surprisingly complex politics of avatars and identity in online games. Sadly, it seems many did not get much out of it.

No, judging by the comments in the post it seems many decided to read simply the headline of the piece (which, as an angle to entice readers into something a little heavier than we're accustomed to, could have been better-presented on our part), and not the suggestion to read either a fuller piece or Harrell's whole paper elsewhere. In the interests of presenting Harrell's thoughts on the matter in full, then, he's been so kind as to present this post.

Top: A screenshot from Harrell's interactive game/poem "Loss, Undersea" (left), and a range of possible avatar transformations (right) (you can watch a video of the project in action here)

Advertisement

Gamers are beautiful, so think of this as a love letter to you. I love how we can circle the wagons when the medium we care for so much is assailed. So, let me tell you directly: my goal is to support your creativity in gaming and other digital media forms. In recent days, I had the pleasure of being interviewed by Elisabeth Soep for boingboing.net on the topic of research into identity representation that I have been conducting. This article, "Chimerical Avatars and Other Identity Experiments from Prof. Fox Harrell," also had the distinction of having been reblogged on Kotaku under the sensationalistic headline "Making Avatars That Aren't White Dudes Is Hard." I am thrilled to see the dialogue started by my fellow denizens of gamerdom, however the title and article misstated my aims. In this line of my research (I also invent new forms of AI-based interactive narrative, gaming, poetry, and other expressive works), I am interested in two things:

1) New technologies for creating empowering identity representations, not only in games but in social networking, online accounts, and more.

2) Using these new technologies to make avatars and related gaming systems more artistically expressive.

Advertisement

What I have called "Avatar Art," can make critical and expressive statements regarding identity construction themes including changing moods, social scene, marginality, exclusion, aesthetic style, and power (yes, including gender and race but certainly not exclusively). My own works construct fantastic creatures that change based on emotional tone of user actions or based upon other people's perceptions rather than the players'. My real efforts, then, are quite far removed from the goal of creating an avatar that "well, looks like [I do]!"

Read the original article too. And, for your convenience and in the spirit of dialogue and genuine desire to engage and grow, I offer a list of 10 follow-up thoughts that I posted to the comments on the original.

1) On race. The points argued in the article do not primarily revolve around race. Really, since this is about research, the goal is to imagine technologies that engage a wider range of imaginative expression, social awareness/critique, fun, empowerment, and more.

Advertisement

2) On personal preference. The game examples discussed represent personal preference. One is allowed to prefer Undead that look more mysterious (such as "lich-like" or other similar Undead types — the idea is a male analog to the female Undead which can look much more like the Corpse Bride) than like a Sid Vicious zombie on steroids. One is also allowed to believe that such options would break the game maker's (Blizzard's) coherent cartoony aesthetic driven by the game's lore. The larger point is that issues like aesthetics, body-type, posture, and more, are meaningful dimensions. In the real world or tabletop role-playing it would be easy to simply imagine these attributes — they do not need to be built into rules. Yet, in software they are implemented through algorithmic and data-structural constraints. Why not imagine how to do better without allowing players to break the game or slow things down?

3) On the bigger picture. The game examples I raise are, to some extent, rhetorical devices. They address fashion, body language, gender, culture, and more. The idea is that in the real world there is an incredible amount of nuance for representing identity. Identities are much more than race and gender. Identities change over time, they change based on context. Research is forward looking — why not imagine what it means to have technologies that address these issues and how we can use them effectively. That includes making coherent gameworlds and not bogging people down during or before gameplay. The rhetorical devices may be more, or less, successful. But the point remains that this is a *hard* problem.

4) On back-end data structures and algorithms. The research mentioned does not focus primarily on external appearance. It focuses on issues like emotional tone, transformation, change, community perspectives, stigma, and more. As noted, these are internal issues. But we can go further. New computational approaches are possible that do not reify social identity categories as discrete sets of attributes or statistics. Categories can be modeled more fluidly, and new game mechanics may result. My GRIOT system allows for AI-based composition of multimedia assets, including characters in games. Let's imagine and create technologies that can do more — and then deploy them in the most effective ways whether for entertainment, social critique, or social networking.

Advertisement

5) On fiction as social commentary. The approach argued for may also help to make fantastic games begin to approach the nuanced analyses of fiction writers like Samuel R. Delany, Joanna Russ, or even the introspective metaphysical work of Haruki Murakami. There is a tradition of fantastic fiction as social critique. Tabletop gamers may know of the game "Shock: Social Science Fiction" as a good indie example of this.

6) On characters different from one's self. The article does not point to discomfort with playing characters such as elves with pale skin, or suggest that one should inherently feel uncomfortable playing a role that is far from a real life conception of identity. Rather, it begins with the ability to happily play characters ranging from elves to mecha pilots. This is a wonderful affordance of many games. But even more, it is great to be able to play non-anthropomorphic characters and many other options. I have done research on this issue to describe different ways that people related to their characters/avatars: some are "mirror players" who want characters that want characters that are like themselves, others are "character users" who see their identities as tools, and others still are "character players" who use their characters to explore imaginative settings and alternative selves in playful ways (this is the nutshell version). However, no matter what, the types of characters in games are often related to real world social values and categories. It can be disempowering to encounter stereotypical representations over and over.


A screenshot of the character creation questionnaire in Ultima IV

7) On alternative models. Someone mentioned text-based systems and systems that use other characteristics such as moral choices to determine characters (c.f., Ultima IV). That is exactly the sort of thing being argued for here. Meaningful character creation — not just tired archetypes and game-mechanics oriented roles. Someone else mentioned modding and suggested that not modding may be a mark of laziness. Yet, the goal here is actually building new systems that can do better! Certainly less lazy than adapting existing systems. And this effort is proposed with a humble, inviting attitude. When new systems fail, the input of others (such as those commenting here) can make them better still! Works like "Loss, Undersea" and "DefineMe: Chimera" are just early examples of artistic outcomes or pilot work built in some cases using an underlying AI framework I have designed called the GRIOT system. This endeavor is called the Advanced Identity Representation (AIR) Project ("advanced" not because of hubris, but because it is possible to go much further than current systems allow).

Advertisement

8) On platforms. The research mentioned looks at not only games, but also at social networking sites, online accounts, and avatars. There are some strong overlaps between them, despite the obvious differences. Looking at what each allows and does not allow can yield valuable insights.

9) On this guy, that guy, and the other guy. Offering appropriate constraints for gameworlds and allowing for seamlessly dynamic characters is important. Ideally, one outcome of this research would be ways to disallow "That Guy" (described as a particular type of disruptive role-player) to ruin the game. That said, labels (like "That Guy") can obfuscate the issues at hand. So can a focus on details rather than the general potential of exploring new possibilities. The goal is not to offer every nuanced and finicky option, but rather to illustrate what some potential gaps might be. People are complicated, any elegant technical solution that enriches role-playing in games seems desirable. But this needs to be done in a sensible way that adds meaning and salience to the game. Examples like the ranger and mesmer classes in GuildWars: Nightfall are really just to describe how there are many categories that are transient, in-between, marginal, blended, and dynamic. Probably more than there are archetypical categories. Let's think about how to enable these categories in software.

10) On the goal. The ultimate goal is not a totalizing system that can handle any customization. Rather, it is to realize that our identities in games, virtual worlds, social networking sites, and related media exist in an ecology of behavior, artifacts, attitudes, software and hardware infrastructure, activities (like gaming), institutional values and biases, personal values and biases, systems of classification, and cognitive processing (the imagination). In the face of all of this complexity, one option is to develop technologies to support meaningful and context-specific identity technologies — for example rather than just superficial race, gender, masquerade masks, and the tinting of elves, let's think about how to use all of these to say something about the world and the human condition.

Advertisement

Thank you all for considering these ideas, even those who disagree. Your concerns may have been clarified, and they may have been exacerbated, but this is what productive dialogue is all about.