As an ethnomethodologist and conversation analyst, someone who scrutinizes the micro organization of talk-in-interaction through audio-visual recordings, what intrigues me most about virtual worlds is that they take face-to-face conversation as their communicative metaphor (many of them do anyway). Often when I play, I notice ways in which avatar and chat systems differ, in interactionally consequential ways, from the system of real-life face-to-face. And I ask myself, "Why did the designers do this?" "Was it too hard to follow real life?" "Were they going for a different effect?" "Or did they just not know enough about face-to-face as a system?"
Now whenever I propose that designers might benefit from following real-life, I inevitably get rebuked by some for using the R-word... realism. "You can't translate real-life into the virtual." "Virtual worlds are supposed to be fantastic, aren't they?" "Playability is more important than realism." "We can make the virtual world better than the real." It seems that realism has fallen out of style. So I'd like to step back for a moment and briefly examine what we mean by "realism" in art and media.
First, although "accuracy" may be important in
training simulations, e.g., flight simulators, what matters in art and media is
the appearance of accuracy. Richard
Bartle (2004:320) suggests about depicting the
flight of a virtual arrow, "this doesn't have to be done by applying
Second, art and media can be believably realistic in many very different ways:
1. Narrative - Does the story seem plausible in real life? Story elements, such as settings, environments, events, characters, entities or behaviors, can be more or less realistic or fantastic. Contrast non-fiction with fiction with SciFi/fantasy with surrealism.
2. Visual representation - Do the people, places or things, no matter how fantastic, look and move like they could exist in the physical world? Gollum (2003) is fantastic in a narrative sense, but nonetheless he looks and moves (through motion capture) as if he were really walking side-by-side with Frodo and Sam (not 2D, stop-motion animated or made of clay). In contrast, Sully (2001) has a cartoony form, but his long fur moves as if blown by real wind (through sophisticated physics simulation). (See "cartoony" game art.)
3. Object interaction - Can you interact with objects like you can in real life? In Half Life 2, you can destroy countless wooden crates in realistic-looking ways by shooting them, but you can’t sit on them and lay out your weapons to examine them. In Second Life, objects are (slightly) more likely to enable you to interact with them in appropriate ways: sipping from a coffee cup, laying and fidgeting on a beach towel, sitting on most horizontal surfaces.
interaction - Can you interact with other people like you can in real
life? On the one hand, the "public privacy" in MMOs is a fantastic
feature that enables you to "talk behind someone’s back" while your
avatars are standing face-to-face (through private chat). On the other hand, in
EverQuest II, players can make
"eye contact" in somewhat realistic ways by clicking on an avatar to (have
your avatar) look at it, as well as by
simply moving very close to it (through a clever "auto-glancing"
trick). While the latter is mostly aesthetic, the former (when used) can be a
useful resource for displaying one's focus of attention and indicating the
recipient of one's utterance, as conversation analysis shows people do
systematically in real life.
So when we say that some art or media seem "realistic," we usually mean that only particular elements of them appear realistic while others may appear fantastic. It's always a mix. Given this, when might it be desirable to make certain elements of social interaction in virtual worlds realistic?
1. When the aesthetic style of the world is realistic (i.e., there should be consistency in the look and feel). For example, the overall look of EverQuest II is realistic. It has some realistic features in the social interaction system, such as "eye contact" and "auto-glancing," which nicely complement the visually realistic avatars who have believable anatomy, cast shadows, bounce light off of shiny armor, sway and blink just right. However, there is often an inconsistency when these same realistic-looking avatars stand silent and still like zombies for too long (as in most MMOs) while their players type long chat messages or perform other concealed activities. Ethnomethodology shows that people's ability to locally manage everyday social settings depends on the observable-reportable nature of each other's actions. When these actions are hidden by design in virtual worlds, it impacts players’ ability to achieve mutual intelligibility. So revealing them not only makes avatar interaction appear more realistic, it also makes interaction easier for players to manage.
2. When real life has a better solution to an interactional problem. The text chat in most MMOs does not easily enable tight coordination of actions between players because, at any given moment, you can't see if the other is in the process of saying something. As conversation analysis demonstrates, the ability to monitor utterances-in-progress is critical to enabling tight coordination in interaction. So when tight coordination really matters to players (e.g., in guild raids), what do they do? They abandon clunky text chat all together and switch to a real-time medium, voice (i.e., TeamSpeak). However, real-time turn construction can also be approximated with text by posting messages on a word-by-word as in There. In other words, public turn construction is better for tight social coordination than private turn construction.
3. When they translate well. The ability of real human bodies to navigate physical space and thereby to "approach" or "face" other human bodies, translates very nicely into virtual worlds. In contrast, peripheral vision and proprioception, which impact players' ability to know what their own and others' bodies are doing and therefore their ability to use gesture and facial expression, do not.
4. When you want skill transfer to real life. If the virtual world is being used to teach real-life communication and coordination skills through simulation (e.g., Forterra?), then it will require accurate simulation of certain key features of real-life conversation.
Face-to-face conversation is a technical system - like human anatomy, physical forces or optics - that is challenging to model in believable or accurate ways. Therefore, when realism in avatar interaction is desirable, the findings and methods of conversation analysis and other fields can have practical value for designers of virtual worlds (for more see "Doing Virtually Nothing: Awareness and Accountability in Massively Multiplayer Online Worlds").
So when do you think the R-word is appropriate in virtual world design? When is it not?