NICK YEE VISITS METANOMICS - MARCH 17, 2008
ROBERT BLOOMFIELD: Welcome, everyone, to this special edition of Metanomics, on this
4-sim Amphitheatre on CMP Isles one, two, three and four. Metanomics presents an
interview or panel discussion on business and policy issues in virtual worlds every Monday
at 11:00 A.M. Pacific time, Second Life time.
But this a special edition as we’re also a part of the Life 2.0 Conference. So thanks to United
Business Media’s Think Services Metaverse Group for letting Metanomics be part of
Life 2.0. And, to those of you who are new to Second Life, coming in for the Life 2.0
Conference, welcome. As always, thanks to our sponsors of Metanomics: SAP, Cisco
Systems, Generali Group, Saxo Bank, Kelley Services, Sun Microsystems and, of course, to
my own institution Cornell’s Johnson Graduate School of Management, for supporting me in
this effort, and the Second Life Cable Network for filming and distributing these discussions
with such fascinating people, like today’s guest, Nick Yee.
Let me remind everyone that Metanomics has a Facebook group and Twitter account, both
called Metanomics, as well as a web site, metanomics.net. We encourage you to join the
Second Life group called Metanomics so you can join in, in the backchat during this event,
although I should warn people that the backchat is having a little bit of problems today, the
group chat is. So we’ll primarily be, I think, relying on local chat.
So as you have questions, comments, reactions to what we’re saying, we love to know
you’re alive out there so type away in the local chat. If you’re here on CMP, feel free to
instant message me if you’re hearing this from any one of our event partners. This event’s
going on, on sims across Second Life, from Muse Isle to Colonia Nova, Rockliffe University
and the Outreach Amphitheatre of the New Media Consortiums Educational Community
So let’s see. Let’s get started with our show. This winter season of Metanomics has been
pretty heavy on game developers and executives and light on academics. We’ve had
Robin Harper and John Zdanowski, VPs of Linden Lab; Robert Gehorsam, president of
Forterra Systems; Michael Wilson, the CEO of There.com; Chris Klaus, CEO of Kaneva.
The closest we’ve really gotten to academia has been David Wortley, director of the Serious
Games Institute, and last week Richard Bartle, who’s best known as a game designer, but is
currently a lecturer on game design in the UK.
Well, today I am delighted to turn back more squarely toward academia and welcome
Nick Yee, who got his Ph.D. from Stanford’s Department of Communication, and has been
conducting peer-reviewed research on games, gamers and the more generically named
virtual collaborative environments. Nick started grad school at Stanford in the fall of 2003
with the Department of Communication, and he has been working with Jeremy Bailenson
conducting his largely experimental research with a touch of demography in there as well.
He graduated in June of ’07 and is currently a research scientist at the Palo Alto Research
Center, called PARC. So, Nick, welcome to Metanomics.
NICK YEE: Thanks so much for inviting me, Rob.
ROBERT BLOOMFIELD: It’s really a pleasure to have you on. You’ve been incredibly
productive for such a young guy, so I’m really looking forward to hearing what you’ve got to
say. Now, my hope for this discussion is to cover two types of material. Obviously, I want to
talk, as much as we can fit into an hour, about the specifics of your very prolific research
endeavors, and so we’ll see what we can get done. But I also want to talk a little more
generally about the outlook for virtual world research and researchers in academia. So if I
could, I’d like to start just by asking you, what took you into academia and into a
communications department in particular?
NICK YEE: I guess I knew all this back in undergrad, because that was when I started
doing online surveying research, and I started that because these two seniors a year ahead
of me were looking at the personality differences among three different game genres, at
first-person shooter games, real time strategy games and MMOs. And at that point, none of
us had ever played an MMO so the Psych Department went out and got us a copy, and I
was a lab techy, so I installed it. And we all tried the game, but I was the only one who liked
it. And so that’s how my online game research started. I finished undergrad, I worked for a
couple years in Chicago, and I was still doing the online survey on the side, so I said, “I’m
still doing this; I might as well be doing it more full time.” So I started looking at grad
At that point one of the things I noticed was it was hard for me to study what I was interested
in studying in psychology at that point because there was so still much a focus on the
violence and aggression agenda. And I was much more interested in the culture of this
space, and it was harder to justify that within traditional psychology. So I was starting to look
at HCI Group’s media studies departments, as well as communication departments, and so I
applied to a couple places, and then Stanford was the one that really made the most sense
at the end.
ROBERT BLOOMFIELD: And so right now is that where you would say--like if you’re an
aspiring academic looking at Ph.D. programs in Virtual Worlds and gaming communication
departments would be one of the first places to look?
NICK YEE: I think it really depends on--apart from thinking about what themes and issues
you’re interested in, it’s also important to think about what methodologies you’re comfortable
with. So coming out of a pure psych background from undergrad, the only method I had was
experimental paradigm, and I was getting a little tired of a pure experimental approach. And
so getting into the [AUDIO GLITCH] program helped me look at survey methods and more
qualitative techniques as well. So for people who are interested in this space, it’s also
important to consider what methods you’re interested in. For example, for some people
doing interviews and participant observation from an anthropology or sociology perspective
may be better aligned with their own interests. One thing that wasn’t obvious to me at first
when I was applying to grad school was that, apart from the thematic and issue alignment,
it’s also really important to consider the methods alignment.
ROBERT BLOOMFIELD: And you’ve been, I guess, primarily experimental with a healthy
dose--well, I guess actually you’ve done experimental, you’ve done demographic where
you’re really surveying representative populations in games, and then you’ve also been
doing some of, I guess, what is often called qualitative research on the alliances in World of
Warcraft, where you’re interviewing and observing natural behaviors. So you pretty much
seem to span many of the social science methodologies.
NICK YEE: Yes. I think, for me, what was most interesting was looking at the same topic or
issue from a variety of different perspectives, and what was most satisfying for me was--I
was doing mostly survey work in the gaming space, and a lot of people would say, “Oh, but
it’s all very biased. How do you know that you’re not just [AUDIO GLITCH] population”? But
then I started doing work, for example, looking at more server-side data, like for example at
PARC’s PlayOn Group, or doing more experimental designs. Really, that was always really
satisfying, looking at the same issue from multiple angles and really coming up with a
ROBERT BLOOMFIELD: Now, you’re at PARC, Palo Alto Research Center now. Can you
give us a sense of, I guess, what took you there from Stanford and what it’s like to be at a
research organization that is not a purely academic research organization?
NICK YEE: Yes. I started at PARC as a summer intern the summer of my second year in
the Ph.D. program and really loved that work. And they said, “Would you like to stay after
the summer”? And so I actually stayed with them part-time for the last two years of my Ph.D.
program. So when I was finishing with the program, I was looking around for what makes
sense for after the Ph.D. The thing that struck me was that I really wasn’t sure, but I wanted
to keep my options open. So what I did was I said, “I’ll take the PARC job halftime. I’ll still
continue some research on the side with my advisor, Jeremy, at Stanford, and I have
another gig elsewhere.” But really just take the time to think about where I want to move
At PARC, the thing that was so fascinating was, when they started doing the PlayOn Group,
when they started doing the census research of World of Warcraft, gathering pseudo
server-side data of players from five different servers and really looking at the social
dynamics of what factors predicted guilt survivability. I thought that was such fascinating
research, with being able to, in a sense, make an end run around the game deliverer, not
having to ask them for data directly, but being able to siphon that data out ourselves was
really exciting. And so really one of the reasons that compelled me to stay there, I like
working with the people that I work with. One of the things that’s good at PARC is that there
are a variety of different projects, and so they really encourage you to spread out into new
areas and to bring your expertise from past spaces into new areas. So it’s just really
ROBERT BLOOMFIELD: That’s great. So actually, we have a question from one of our
audience members early on here, which is--as I mentioned before the show, Nick, we had
Richard Bartle on here last week, and we’ll talk a little bit more about how your research
relates to his player types. But the question here is, this comes from Fleep Tuque:
Richard Bartle commented that he no longer--or rarely can enjoy playing games as a player
because his experience as game designer causes him to see the game through that lens.
How has your research changed your personal experience of game-playing? And if you do
still play for fun, which games are you playing?
NICK YEE: What was funny was, in my dissertation year, I actually managed to level in
World of Warcraft, I managed to level Draenei Shaman from level one to level 70 in my
dissertation, year so I look back and I think that’s kind of funny.
ROBERT BLOOMFIELD: Does your advisor know this?
NICK YEE: He does, but he doesn’t play, so I don’t think he knows exactly what that
ROBERT BLOOMFIELD: It’s all in the name of research.
NICK YEE: Right. It really hasn’t affected my game-play in a negative way. I think the thing
that’s really affected my game-play negatively is that there really hasn’t been much
innovation in the MMO space. I think that, more than anything, has really dampened what I
find in MMOs now or whether I’m willing to try out a new MMO. It wasn’t particularly the
research that did it, I don’t think.
ROBERT BLOOMFIELD: Well, I have to say now the first part of your answer didn’t sound
at all like Richard Bartle, but the second part sure did. And anyone who wants to check on
metanomics.net, you can see the transcript or look at the SLCN video of Richard Bartle,
who, I would say, shares your concern about the quality of new MMOs.
But let’s turn now to your research. So when I pulled all my notes together, I realized that in
preparing for this interview I had to read 14 research papers.
NICK YEE: I can’t believe you read them all.
ROBERT BLOOMFIELD: Well, this is me. I guess I’m an achiever type or something,
according to Richard Bartle. But anyway, so the way I see it we can organize your work into
three different categories. There are the studies of how virtual environments affect the way
we perceive ourselves and others; there’s the demographic work on who participates in
games and why; and then there’s the work on social interactions in virtual worlds and,
particularly, your research on guilds in World of Warcraft. And so what I want to do is start
with that first category and, in particular, start talking about the Proteus Effect. Can you just
tell us where this name comes from and what it means?
NICK YEE: Sure. We labeled the effect the Proteus Effect from the Greek god Proteus, who
was able to change his physical form at a whim. So there’s this passage where it talks about
he’s changed from a tree into a lion and so forth, and we used the Proteus Effect label to
talk about an effect that we had found in a series of experiments. We were looking at
whether, and in what ways, the avatars that people use in virtual environments change the
ways that they behave. And a lot of people who had been looking at computer-related
communication had been looking at features of the environment, such as anonymity or the
lack of social cues and so forth. But, for me, looking at online games in virtual worlds, I was
real interested in the self-representation issue as a fundamental issue that one of the big
things about being online is that you can literally change your height, your weight, your
gender at a click of a button. You know, what does that mean when we have these flexible
Protean bodies? And so that’s really what the effect labels. And so, in a nutshell, what we
found was that the avatars that people choose have a huge impact on how they behave.
In one experimental study where we put participants in either attractive or unattractive
avatars, the participants in the attractive avatars were willing to walk closer to a confederate
stranger, so a research assistant posing as another user in the environment, and they were
also--the participants in the attractive avatars were also more willing to disclose personal
information about themselves via introductory prompts. So that’s just in a nutshell how we
came to that term of a way of thinking about what it means to have these very flexible fluid
ROBERT BLOOMFIELD: Now, a lot of the research you did on the Proteus Effect is
inspired by Bem’s theory of self-perception. So can you just give people a sense of what
that theory is and how it relates to avatar embodiment?
NICK YEE: Absolutely. So Bem’s theory of self-perception essentially says that, in a lot of
cases, we’re not sure how to think about ourselves or how we really feel about things. And
what we do is we observe our own actions and our own behaviors from a third-party
perspective, and infer from the observed actions how we would behave. So there’s this one
early well-known study where when you put people in black uniforms, they become more
aggressive. They pick more aggressive games to play. There’s another study where, when
they put someone in a nurse’s uniform, they’re less likely to use--and this was in a
teacher/learner paradigm where they use electric shocks. This was the pre-Milgram days.
They found that when you put someone in the nurse’s uniform, they’re less likely to use the
electric shocks than if you put them in a uniform that resembled a Ku Klux Klan uniform. And
so drawing from that research we were saying, “What happens in the Virtual World? Can we
leverage this research in the Virtual World where, not only can we manipulate costumes or
what people wear, but literally what people look like”? And so that’s where the attractiveness
and the height research came out of was leveraging that.
ROBERT BLOOMFIELD: So I’m looking at the backchat, and I’m seeing a lot of questions
where people are asking about the method that you use, so questions like Ana Herzog is
asking: How do you define “attractive”?
NICK YEE: Yeah, a great question.
ROBERT BLOOMFIELD: So I understand you have done some research in Second Life,
and we’ll get there shortly, but a lot of your research is actually done in the lab where you
assign avatars to people. So I’m thinking in particular of the two papers, one is the called the
Proteus Effect, the effect of transformed self-representation on behavior, that’s with your
advisor, Jeremy Bailenson. And then there’s another that takes us to offline behavior as
well, that is, not only with Jeremy Bailenson, but a few colleagues at PARC. But both of
those use a very similar methodology where you’ve got headsets. You create your own
virtual collaborative environment. So can you just give us a visual sense of what’s actually
going on in that experiment?
NICK YEE: Yes. So in all these experiments we try to control for everything except what
we’re manipulating. So let me just address the attractiveness issue real quick. For the
attractiveness one we actually assigned people to avatars, and so we were able to take a
set of pre-generated faces, have coders write them, and then generate the attractiveness
scores. So that’s how we picked the attractiveness issue.
But let me turn back to what these experiments typically look like. So in the lab we actually
run an immersive virtual reality setup. People put on the helmets, and the helmets have a
stereoscopic lens so they see two displays, one for each eye. And so they see it in
stereoscopic 3D. There is a light that four cameras in the corners of the room pick up. And
so when the person moves in the physical space, what they see in the lens is updated so
that they feel like they’re moving in the virtual space. There’s also an accelerometer on the
back of the helmet so when they tilt their heads, again, their viewpoint changes accordingly.
So when I said that the participants walked closer to the confederate stranger earlier on, I
meant they literally physically walked closer to the stranger, as well as, of course, virtually.
So that’s how that setup looks like.
So when they want to move in the world, they don’t use a keyboard in our lab, they just walk
to the direction they want. One of the interesting things, for example, is that when we talk
about eye gaze in the lab, we can actually track that fairly precisely by taking the
measurements from the accelerometers.
ROBERT BLOOMFIELD: So let me just be a little more clear in my own mind. I mean, let’s
say I’m a participant in this experiment. When you say “I walk closer to someone else,” I’m
in a room alone, but I see in my head, because of this headset, I actually see myself as
being in a virtual world, and so it’s as if I’ve walked closer, avatar to avatar. Is that right?
NICK YEE: Yeah, correct. So when you’re in the immersive VR setup, you’re actually in a
first-person perspective. You see a room in front of you and when you take one step
physically, your virtual scene updates so it’s pushed closer to you. So it’s as if the physical
step you took had a corresponding action in the virtual world.
ROBERT BLOOMFIELD: Okay. Now, you assign people to more attractive and less
attractive avatars as one of your key manipulations. And so now one of our audience
members, Ana Herzog, was asking how you define “attractive,” and I’ve got a slightly
different question, which is: How do people know they are themselves, or their avatar, is
NICK YEE: That’s a great question. [AUDIO GLITCH] this part out before. And this was
actually one of the tricky things that we had to come up with, was how do we convince
someone and let them know how they look like? So we devised a virtual mirror--great
question; sorry I didn’t mention this earlier--so they move in a virtual world. They first turn
around, and we tell them to turn around, and there actually is a virtual mirror. And we say,
“This is how you look in the virtual world.” And to convince them that that’s actually the case,
we ask them to take a step toward the mirror and to roll their heads and so actually see their
virtual mirror reflection doing those actions. And it’s done in a fairly convincing way, and we
tell them, “This is now other people in the world see you,” and then they turn back around,
and so they’re in first-person perspective again, but it’s the virtual mirror that lets them see
how they look like.
ROBERT BLOOMFIELD: Okay. And how about the answer to Ana’s question? You used a
group of independent raters to--
NICK YEE: Correct. So all the faces--
ROBERT BLOOMFIELD: --assess whether an avatar was attractive.
NICK YEE: Correct. We had a database of faces, and we had a separate pool of raters rate
all the faces. We calculated the average attractiveness for each face, and then we picked
faces at the unattractive end, at the attractive end, and faces at the average. And so the
other thing that’s also worth mentioning is it’s really important when we ran this study, that
it’s not the confederate’s response to attractive or unattractive faces that creates the finding.
So what we use the average face for is from the confederate’s point of view, from the
research assistant’s point of view. They always see the participant as having an average
face, so they’re blind to condition.
So the only person who knows whether they’re attractive or unattractive is the participant.
It’s literally in their own head. And so that’s how we teased apart the self-perception effect
from what’s commonly known as behavioral confirmation. So when I interact with somebody
who’s attractive, I treat them differently, and that leads them, in turn, to behave differently.
So when we ran the studies, it was really important for us to make sure we isolated and
teased apart those two effects. And that, when we were studying self-perception, that we
really were studying self-perception.
ROBERT BLOOMFIELD: And I’ll say this is, to me, one of the most fascinating things about
the types of social psychological research you can do in virtual worlds, is that not everyone
sees the same thing. I can see myself as tall, but you can see me as being short. You might
think you’re a bunny, and I see you as a wolf.
NICK YEE: And actually, a lot of the studies that we’ve run at the lab focus on the fact that
everyone thinks they share the same virtual space, but in a lot of cases they don’t. And so
we’ve run a lot of studies where we’ve had these kinds of manipulations. We actually did
another study with the height, where we did exactly that, where the participants thought they
were tall, but the research assistant saw them as just average height. So again, it’s all in the
head of the participant.
Of course, as you said, there’s this real interesting case where what if the participant thinks
they’re tall, but everyone else thinks they’re short? So there are all these really interesting
ways of using the virtual space to run these really cool studies that can tease apart really
complicated theories in psychology.
ROBERT BLOOMFIELD: Yeah. Now one of the, I guess, key results that you found in your
first Proteus Effect Paper was that the people who are assigned taller avatars, meaning, I
guess, that they feel taller, even though others don’t see them as taller, they play a very
simple negotiation game, and you find that they’re more aggressive and also more effective
in the sense that they end up with more money. And then what you do in the second paper,
which I thought was particularly interesting, is that you then take that--it seems like largely a
replication of that experiment--but then, after they’ve done that, you then have them meet
someone in the flesh and do a very similar task. And again they end up--there’s this residual
effect that, after being taken out of their tall avatar, negotiating as the person they actually
are--they end up continuing to be more aggressive and get better outcomes. So why does
that happen, in your eyes?
NICK YEE: As soon as we ran the studies in the virtual environment, and we saw several
studies that produced consistent findings, the very next natural question was: Oh, my God,
does this actually carry back out to subsequent interactions? Particularly because the
average online gamer in a game like World of Warcraft, they play 20 hours a week, and it’s
spread over that week. So if they get constant reinforcement from that environment, how
does it carry back out to the face-to-face interaction? So that was the motivation behind it.
And we found that it did indeed transfer. So we did it with the height one. We actually also
did it with the attractiveness one. So we ran the study again in the virtual environment, but
when they came out we didn’t just replicate it, we said, “Does it carry out other kinds of
interactions”? So we actually created a mock dating website, and we sold it. We told the
participants that it was a separate study altogether. And I think Tammy just had this
question before. So what we did was we had them run the attractiveness study in the virtual
environment. They came out, and we said, “Okay, that was the end of that study. Here’s a
We were studying how people date online. And what we did was we put them in front of a
mock dating website, and we made them go through filling out a profile, and then we
showed them nine photographs--real photographs, not 3D avatars--of people of the gender
that they’re interested in, and the nine photographs were selected because they filled the
spectrum from unattractive to attractive. Again, these were pre-rated. And our question was
do people who had an attractive avatar ten minutes ago in VR, do they react differently than
the people who had unattractive avatars? And so within that short time frame of ten minutes,
we found that people who had attractive avatars actually picked more attractive potential
partners than the people in the unattractive avatars.
Again, we know now that it looks like these effects carry out for a short amount of time when
people come out of their virtual environment. What’s still not clear is how long those effects
last. It’s very likely that these effects are just totally washed away when they leave the lab,
and it’s 30 minutes, an hour later. But what’s interesting is that even to a short amount of
time that they do seem to have that effect. And, of course, the bigger question is what
happens to people who spend 20 hours a week, half a work week, with a particular avatar;
how do those behaviors cross over?
ROBERT BLOOMFIELD: Yeah, and I think anyone who’s played an MMO or a video game
for hours at a stretch knows I’m not the only one who has been lying in bed with my eyes
closed, seeing myself do whatever it was I was doing in-world for the last three hours. So I
could imagine that there’s going to be a lot more carryover when people are much more
immersed for a much longer time than you did in your study.
NICK YEE: Right.
ROBERT BLOOMFIELD: Now, one of the papers that I thought was particularly interesting
was--this was again with Jeremy Bailenson, “Walk a Mile in Digital Shoes: The Impact of
Embodied Perspective-Taking on the Reduction of Negative Stereotyping in Immersive
Virtual Environments.” You’ve got to love academic papers. If it doesn’t have a colon in it,
it’s not serious work. So in here you have people take on the avatar of an old person or a
young person and provide some evidence, I mean somewhat, I guess, mixed evidence--
NICK YEE: Mixed, yeah.
ROBERT BLOOMFIELD: --that actually being in the form of an old person reduces
age-related bias. So can you talk a little bit about what you see as being the mechanism for
NICK YEE: Yeah. In social psychology, one of the mechanisms that they found that
effectively reduces stereotyping and negative prejudice of other social groups is this notion
of perspective-taking. And here’s what they used to do that they found was effective. So the
simple thing of giving someone a photograph of someone of a different social group and
say, “Imagine a day in the life of this person, and just write a short paragraph about it.” That,
in and of itself, was effective enough to get the participants to reduce the negative prejudice.
And the mechanism that they think is behind this is because, when you ask someone to
take someone else’s point of view, it reduces their sense of the distinction between self and
other. So by asking someone to take someone else’s point of view, it gets them to feel that
the self and other are more similar than not. And so what we did was, we said, “Instead of
doing verbal or narrative perspective-taking, what if we did it directly via an avatar in a
virtual environment”? So that was the motivation behind it.
Our results were a bit mixed. We looked at discrimination and prejudice in a variety of
different scales and measures, and I think we looked at three, and it only came up
significantly strongly for one of those three. So there was some tentative evidence that this
may or may not be effective, but it was an interesting direction to look into.
ROBERT BLOOMFIELD: Now, for all of these results one explanation is this “them” like
self-perception theories, as you said, that we look at sort of ourselves from the outside to
actually interpret who we are on the inside. There’s also the possibility that it’s a priming
explanation, that you’re just sort of triggering people to think in certain ways so they could
see another avatar that’s old, and it might have the same effect. You actually did some work
to try to tease those out.
NICK YEE: Yeah. There have been some really interesting studies on priming, and here’s a
really seminal study that really gets the point across. So in one study they presented--let me
think which one’s better--they presented key words related to old age and being elderly to
participants. They’re either presented elderly-related words or very neutral words, and what
they found was people who were exposed to the elderly key words walked slower to the
elevator exit when the experiment ended--
ROBERT BLOOMFIELD: I did see that study.
NICK YEE: --than those who got neutral words, right. And so in the literature of priming,
there’s this finding that when people are primed with key words from a specific domain, they
unconsciously behave more like those variables. So in our study, when we showed
someone an attractive avatar, one counterclaim is that they were just being primed by
attractiveness, and attractiveness primes friendliness, and that’s all you’re seeing. It has
nothing to do with this self-perception stuff.
So what we did to tease those two apart--again, here we were leveraging some aspects of
virtual reality--was, in one condition, we ran the same manipulation. They saw themselves in
the mirror, and they did head moving and so forth. In the other condition, in the priming
condition, we said, “Okay, if it’s just priming and it’s not you being in this body that’s
important, then, if we showed you a video playback of someone else who looked in the
mirror, and we told participants that you’re looking at a video screen of someone else who
was in this room”--so they’re getting the exact same visual stimulus as someone else did.
The only difference is that it’s not them. They’re not embodied in that avatar. And so that’s
what we did to tease those two effects apart. And what we found was that the effects were
much stronger when there was embodiment, when people believed that it was their avatar,
when they believed that this was them, not simply that they were primed by some third-party
ROBERT BLOOMFIELD: You know what I love about that study is that researchers are
always emphasizing the importance of matching the method that you use to the questions
that you’re asking. And it’s hard to imagine a better method than that to get at this distinction
between self-perception theory and priming.
NICK YEE: Absolutely.
ROBERT BLOOMFIELD: Now, before we move on to the next basic research topic, we’ve
got a bunch of questions. So we’ve got two classes of questions. One, they’re coming from
a whole bunch of people so I’m just going to paraphrase. I get the sense people are asking
about endogeneity. You were very careful in your studies to assign people to avatars that
are tall or short or unattractive or attractive so that you could know that it was actually the
manipulation, your controlled manipulation of height or attractiveness that generated their
behavior. But in virtual worlds, in fact, people choose their own avatars. So can you say
much about sort of the feedback loops that might arise? Do you think there are feedback
loops where someone chooses an attractive avatar? They see themselves as better but, of
course, other people see them as more attractive and treat them nicer, which is reinforced
and so on. Have you done research on that? Or is that maybe on your “to do” list?
NICK YEE: That’s something on a “to do” list. So you know, of course, looking at user
behavior in avatars, you’ve really also got to look at the choices. So in our studies, we focus
on the effect side of the issue. But absolutely, the choice side is just as important. But it also
happens to be a lot messier, because choice studies are inherently non-experimental so
they’re a very different class of studies altogether, but they’re equally important. Some of the
findings we’ve seen is that it has a lot to do with personality variables, it seems. So whether
someone creates avatars that are similar or dissimilar to themselves physically seem to
have some personality variables interwoven in there. So it’s definitely very important. It’s a
very important part of the issue. In our studies, we focus very much on that effect side, but
the choice side is definitely equally important.
ROBERT BLOOMFIELD: Dizzy Banjo asked this, and I think he was not being serious, but
he asked, what is the implication of all this for furries? I don’t know if you want to go there,
but your call.
NICK YEE: [LAUGHING] I’m not sure. Yeah, I’m not sure either. I’m not sure either.
ROBERT BLOOMFIELD: Okay. Well, let’s move on then, and let’s talk about a study that
you did do in Second Life. This was “The Unbearable Likeness of Being Digital: The
Persistence of Nonverbal Social Norms in Online Virtual Environments.” Now, as I
understand this study, you took hundreds of snapshots of avatars in Second Life and had
people code the gender of people who were being close enough that you would call them
being in dyads. And so you had the male-male dyads, the pairs, and the female-female
dyads, and you found that the male-male dyads stand farther apart and share less eye gaze
than the female-female dyads, which is also something that you would expect in real life.
Now, I have heard a number of people criticize this study by saying that in Second Life it
doesn’t matter where your avatar is standing or which way it’s looking. You can use your
camera controls to have a point of view other than in between your avatar’s eyes, and you
can be in Instant Messaging chat and things like that. How would you respond to that type of
NICK YEE: Yeah, it’s two things. First of all, we really were looking at the avatar gaze and
not the user gaze. And the reason we did that was because, just anecdotally from my own
experiences, in games, the first game that came up that let you run through people, I think,
was [Dark Street Hamlet?]. What was interesting was people still ran around other people
because the thought of running through someone was really psychologically uncomfortable.
So we very much were focusing on the avatar level, not where you’re looking, but how you
positioned your avatar. To the bigger question, of course, what about people using IM and
don’t need to talk face to face?
For us, that really was just noise. It was noise that clearly we couldn’t account for, but it was
simply dampening the effect. But even given that dampening noise, we still found those
significant differences, that there were still those same differences in interpersonal space
and eye gaze that varied along in gendered ways that matched what we knew about face to
So you know, one effect that we found was it relates back to the equilibrium theory. So for
example, when you’re in an elevator, when you’re very, very close to other people, it’s very
uncomfortable if you have to look them in the eye because both interpersonal space and
mutual gaze are expressions of intimacy. So when you’re in an elevator and you’re close to
someone, you have to avert your eye gaze, because you’re not very intimate with this
person standing next to you. So it’s the same way--then a lot of party situations, for
example. People will pick the people they’re talking to and look at them and be fairly close to
them, but they’ll angle themselves so they’re not looking at anyone else, even if they’re
back-to-back. It’s that equilibrium theory.
So we actually found support for that in Second Life as well. When you’re looking at social
group cluster in a kind of like small party situation, we see that effect where the closer that
people get to each other, the more likely they avert their eye gaze, again, for the equilibrium
ROBERT BLOOMFIELD: So we’ve got a tag team question here from Intellagirl Tully,
Fleep Tuque and Lyr Lobo, which is, basically, what about problems that some people have
poor motor control or camera control or movement control? How would that interfere with
your results on the study?
NICK YEE: Again, that’s also in the noise--and just anecdotally--and, of course, this can be
something that we can do a study in--is that, even among players who’ve just started using
Second Life or they’re a little slow, they will eventually fix themselves. Early on we ran these
tests where we studied what would we see if this actually can happened, so we actually
logged into Second Life, and we just walked avatars really close to other people to see what
they’d do. And more often than not they backed away, and they backed away within a
ten-second time frame. Of course, this was something that we could go in and get a sense
of how many people have really bad poor motor skills, but I think most people are fine with
the basic [principle?] of backing away and getting closer to things. For those who don’t that
just gets added to the noise. And, again, even with the noise, we did find those significant
differences, all of which were fairly consistent with face-to-face trends.
ROBERT BLOOMFIELD: Now, you conducted another study that looked at people
interacting with one another, but in a rather different way. You had a haptic joystick. So as I
understand what this is--I don’t even know; Greek or Latin for “touch”?--so you actually had
people using--this is in the paper, “A Virtual Interpersonal Touch: Expressing and
Recognizing Emotions Through Haptic Devices,” with Jeremy Bailenson and Scott Brave.
And so people had a joystick that they could move in two dimensions, and they could do it
fast or slow, and someone else would be holding a joystick somewhere else and it would
move in their hand? Is that what’s going on in this study?
NICK YEE: Yeah. So the issue we were looking at--it was one of these really far out, far out
ideas--was can people express emotion via two degrees of freedom on a joystick? And how
well can people express emotion and how well can other people detect that emotion? And
so we set that one and said, “What do you think will happen? Do you think people can do
that”? And so that was what was behind that study. We could [AUDIO GLITCH] say where
we asked people to express emotions and then another set to try to guess what that
emotion was. It was a really interesting study of instead of just doing [type chatter?] by facial
expressions, what are other ways that people can do to express emotion?
ROBERT BLOOMFIELD: And so what you found is people were better than chance at
recognizing emotions through these sort of very simplified handshakes. They were not as
good as they were with real handshakes?
NICK YEE: That’s correct. We found effects where there were some emotions that people
were more likely to take by chance, I think. Like angry was fairly easy, but it still wasn’t as
good as other, more direct, methods.
ROBERT BLOOMFIELD: I mean this seems to be, at this point, one of the problems. I
mean I know that there’s a big literature on the medium and how it affects people’s ability to
communicate. I’ve seen studies where they have people, via email, attempt to be either
sincere or sarcastic, and they feel quite sure that they’ve conveyed themselves clearly, but,
in fact, the people reading their emails are, you know--basically it’s random; they’re no
better than random at being able to figure out sarcasm versus sincerity. So I do worry about
whether it’s in games or work-related use of virtual environments, whether people are going
to have an easy time, really, communicating with one another.
NICK YEE: Absolutely. Yeah.
ROBERT BLOOMFIELD: Okay. Let’s move on to a second big area of your research, which
is on demographics. You know, who is playing games and why. And so as I mentioned last
week, we had Richard Bartle on the show, and he talked about his four types of players: the
achievers, the explorers, the killers and the socializers. Now, in your case, you have five
types of motivations for game play. You’ve got achievement relationships, immersion--
NICK YEE: There’s actually a more recent version of that but, yeah, but I think the
interesting difference between how Richard approached it and how I approached it--and
we’ve actually had fights about this--is that, when I came into the space with a psychology
background, for me what was most important was how do we actually measure these
things? How can we actually tell whether a person is A or B or C or D? And the discomfort I
had with Richard’s model was that it had never been empirically tested. I mean specifically
this: that, for example, Richard proposes in the explorer type that there’s a kind of person
who really likes understand, whose [AUDIO GLITCH] really needs to understand the game
mechanics, who also likes to explore the world and figures out exactly where the geography
ends. So that, for him, is an explorer type.
When I looked at those types, my first question was do we really know that those
motivations grew together? So for example, I know that there are people who like [the
Minmax?], and I know other people who like to explore geography, but are they actually the
same kind of person? Do those two motivations correlate? So for me, my method was let
me generate a list of different motivations that people say they have in MMOs. I’ll have
people fill out the survey telling me how important each of those is to them, and then I’ll just
cluster in the background, and then I’ll know how these things cluster out. So some of the
differences that are different between my model and Richard’s models, for example, that the
people who like to [Minmax?] are not the same people who like to explore the limits of the
Another thing is, in Richard’s model, role-playing is assumed to be an overarching thing, so
there’s no role-playing motivation. To Richard, I think for him it’s either [AUDIO GLITCH] or
social motivation. Whereas, when I looked in and did the [AUDIO GLITCH] the role-playing
motivation was its own thing. They’re people, and it’s not related [AUDIO GLITCH]
motivation, not overarching motivation [AUDIO GLITCH] that I was looking at so [AUDIO
GLITCH] up with the taxonomy from his experiences. But for me, it was more important to
come up with a model where we could really go in and statistically measure [AUDIO
GLITCH] how do people change as they get older? How are men different from women?
Which of those motivations best predicts who’s more likely to spend more or less time or
more or less likely to quit an MMO?
ROBERT BLOOMFIELD: Now, you mentioned gender, and I know you got some pretty
significant gender differences. And then, on the other hand, in some of your other work,
you’ve argued that actually there’s nothing inherent about existing MMOs that makes them
less attractive to women than men. It’s actually the channels that introduce them to MMOs.
So I guess, first, could you talk just a little about what the gender differences are in
motivations for play? And then talk a little bit about these channels [AUDIO GLITCH].
NICK YEE: That’s a great question. For example, when you look at the motivation
differences or any kind of play-style difference in an MMO, and you slice it just by gender
alone, you’ll always come up with gendered finding. So for example, you always find that
men are more likely to do the things that relate to achievement, and women are more likely
to do things that relate to socializing and relating. For a while, early on, that was the slice
that I was looking at. But when you back out from the picture and you look more at social
access, social context issues, you see a very complicated kind of picture. For example,
when you talk to women and you ask women playing MMOs, “What do you think is least
appealing to you? What would you like to see more of? Why do you think there aren’t more
women playing MMOs”? Very few of them talk about the game mechanics. What all of them
talk about are the cultural constraints. So they always talk about the female avatars, how
being the female avatars makes them really uncomfortable. It reminds them that they were
meant to be spectacles within a masculine world. So that’s one.
Another thing is when you look at gender alone, you find those gender differences, but it’s
only when you take other variables into account that you see the full picture. So for
example, when you take age into account, what we notice is that age actually accounts for
twice the variance in the achieving motivation than gender does. And what that means is,
the people who score highest on achievement are young players. Young male players are
the most competitive and achievement-oriented players. And, as people get older, that
decreases significantly. And that decrease by age, the effective age on achievement is
actually twice as high as of its effect on gender. And, of course, we have all this talk about
games for men and games for women, but we hardly ever hear, “What would games for
older people look like, and how would they look different from games for younger people”?
And so for me it was really important to bring other factors back on the table so we’re not
just looking at the gender issue. But there’s another way to slice it, and that’s the access
channels, how people get in the games. One of the really interesting things in MMOs is that
women are more likely to be introduced to the game via their romantic partners, and one of
the ways that female players in MMOs talk about this is that, within the game, there’s this
cultural notion that women don’t really like to play these games, but, if playing with their
boyfriend or husband, it makes sense because it’s their husband who brought them in. So
you know if [AUDIO GLITCH] playing these games [AUDIO GLITCH] dating or marrying a
guy, and then they had to make a rubric, one real interesting [AUDIO GLITCH]. Rob, are
you still there?
ROBERT BLOOMFIELD: Yes, I am.
NICK YEE: Okay. I just [AUDIO GLITCH] Second Life [AUDIO GLITCH] crashed. I might
crash out really soon. One of the things [AUDIO GLITCH] is that women are likely to play
physically in the game space [AUDIO GLITCH]. A really interesting question of people
ROBERT BLOOMFIELD: I’m afraid I may have lost you, in fact. Yes, Nick Yee has
disappeared into the ether. Well, they say that the sign of a good speaker is someone who
leaves you wanting more. And so I think Nick did a very nice job of that. Yes, Jeckel Thor,
Nick Yee has left the building. Now, of course, the interesting thing is, he is probably still
talking. It’s like when the cell phone cuts off and someone is in the middle of their
Okay. Well, anyway, I guess that was not too badly timed because we only had a few
minutes left as it was. I have about 30 minutes left of content and questions I was hoping to
get to, but maybe we’ll get Nick back on the show another time. So anyway, I would like to
extend my thanks, I guess, I don’t know if posthumously is the right word, to Nick Yee for
coming on to Metanomics. I’d like to apologize to the many people who did not get a chance
to ask Nick some questions, and we had actually a really interesting one from
Opal Zavalota(?), on NMC’s sim, about interview research being conducted in Second Life.
Oh, wait. I see I have filled up enough air time that Snowdrift Heron, aka Nick Yee, is back.
Nick, you there?
NICK YEE: Yeah, I’m back. Can you hear me?
ROBERT BLOOMFIELD: Wonderful. Yes. This is what happens when you’re a rock star
like you are is that we get so many people, even on the 4 sims that everyone has lagged.
We’re really just about out of time so I guess, if I could, let me jump to a couple questions
and let you just answer them. So one is from Opal Zavalota, who’s actually at one of our
event partners New Media Consortium’s sim, and he or she is asking, Can you talk a bit
about your interview research methods? I know you did quite a bit of interviewing in World of
Warcraft. Have you done interviews in Second Life?
NICK YEE: Yes, with the qualitative stuff, it’s mostly within the online gaming context and
open-ended questions on surveys. We actually did a study with Demetri where we used the
census data from World of Warcraft to get a representative small cross-section of players in
Guild, and then we had actually a structured interview script, and we actually went into
World of Warcraft and then interviewed them via script. But most of my training has been on
the quantitative side so, on the qualitative side, I’m much weaker, especially in really talking
about complex interviewing methods.
ROBERT BLOOMFIELD: And let’s see. Well, let me ask, is there anything that we missed
talking about that you would specifically would like to let people know about?
NICK YEE: Let’s see. No, I think we covered good ground. The only other chunk of
research was the PARC PlayOn stuff where we did the census scripting and some of the
findings from that, but I think we covered good ground.
ROBERT BLOOMFIELD: I think we’ll have to have you back to cover what we did miss. I
do have two closing questions. The first one comes from Intellagirl Tully, which is, do you
envision new research methods or new approaches to current methods being developed for
virtual worlds research? What would you see at the top of the list?
NICK YEE: Yeah. I just want to answer Renaldo’s question that just came up right here.
You can find all my papers on the website, nickyee.com. They’re all there as PDFs. With the
new research methods, I think what’s most fascinating to me are the kind of studies like
Proteus Effect where we really leverage the possibilities of teasing apart different people’s
viewpoints, being able to really play with reality. So for example Jeremy has this notion of
transformed social interaction. A really simple example is that here I am in this conference
room, and my avatar can really be looking at one point at a time similar to physical space,
but there’s no reason why--because all of you in the audience are watching this from your
own screens--why my avatar can’t be looking at all of you at exactly the same time.
Because that’s possible. Because we really don’t share that same physical space. But if my
avatar were able to do that, he’d be that much more effective because we know eye gaze
helps a lot of things like persuasion, likeability and [enablements?] and so forth.
So for me what’s interesting is thinking about what are ways in which we can break those
physical rules of interaction in interesting ways and leveraging them for social benefits?
Whether that’s changing the physical rules; fool people into thinking that they have the same
viewpoint, but they really don’t. And I think those are the most interesting ways of doing
research in these environments.
ROBERT BLOOMFIELD: So actually the term--that’s a non-zero-sum eye gaze. Is that
what that’s called?
NICK YEE: Non-zero-sum gaze is the term that Jeremy uses for that.
ROBERT BLOOMFIELD: I don’t know if anyone has ever had a video call with
Christian Renault, but he is the master of looking you right in the eye while he is talking, to
make sure he has your attention, even though he’s in the middle of a snowstorm in Iowa
and I’m in my fourth-floor office in Ithaca. And I can just imagine him then staring eight
people in the eye, and they all believe he is talking directly to them. That does seem like an
exceptionally useful research technique and also, I think, something that’s going to be very
important as we think about the future of virtual work.
Okay. We have time for one more question, which is, would you recommend the aspiring
academics follow in your footsteps, getting a traditional Ph.D. to study these rather
NICK YEE: That’s something that I’ve struggled with, too, especially as we’re gaining some
momentum now in the game studies field. Looking back, I think it’s still very important to be
grounded in a particular perspective and methodology, and it’s really hard to get that in an
emerging field. And a lot of the things that we know and the methods we have in traditional
fields are actually highly applicable to Virtual Worlds, and I think are still equally fruitful. So
I’d say it’s very possible--it’s much more an issue of finding advisors and faculty and
colleagues who are open to applying these methods in these new worlds. But whether
you’re in psychology, communication, sociology, law or economics, there are ways in which
there are real interesting crossovers in doing novel and provocative research in virtual
ROBERT BLOOMFIELD: And I’ll say I see, just looking out in the audience, representatives
of a number of different universities, including places I know that are active in virtual world
research. Of course, my own Cornell, your Stanford, and Indiana Emory University, and I’m
sure there are many, many others I am not mentioning.
So Nick, thank you so much for coming on the show today. It’s really been fascinating, and I
hope we’ll be able to get you back on at another time, and you can talk about the next
14 studies that you’ve done maybe in about three months.
Let me also just say to the people who are listening who may not know Metanomics that
well, we have this show every week, and I’m always looking for academics who are willing
to talk about their research because, at this point, so many people are so early on, they
don’t yet have stuff they’re willing to go public with. So if you are an academic conducting
research in or about virtual worlds and you’re willing to talk about it publicly, please contact
me, Beyers Sellers in-world; Robert Bloomfield, email@example.com, in the real world.
And, again, thank you very much, Nick, for being on the show. I hope we’ll see you again.
NICK YEE: Rob, thanks so much for inviting me on. And, everybody in the audience,
thanks for much for coming. This was really, really fun.
ROBERT BLOOMFIELD: Bye-bye, all.
AUDIENCE: Thank you. That was very good.
Transcribed by: http://www.hiredhand.com
Second Life Avatar: Transcriptionist Writer