Get documents about "Why bother sitting in Second Life
Document Sample
Why bother sitting in Second Life?
– A study on the influence of the appearance of reality on behavior –
Tim Roosen
Leiden Institute of Advanced Computer Science, Leiden University, Niels Bohrweg 1, Leiden, The Netherlands
tim@roosen.nl
Abstract: Virtual environments have a large correspondence to our physical world; in this paper
we investigate if this correspondence delimits the full potential of virtual worlds to constraints that
do exist in the physical world but not in the virtual versions. Sitting seems to be a common posture
in Second Life although it has not the functional use as we know from First Life. Observations
show that SL users sit more often in an environment with a stronger resemblance to reality.
1 Introduction
“This is the beginning of understanding that cyberspace has no boundaries, no techno limits, no
geography, no distance.” (Nicholas Negroponte, 23 October 1998).
During the nineties of the last century futurists, such as Nicholas Negroponte, drew an image of the
optimistic possibilities of cyberspace and saw Internet for example as the tool for removing national
borders. Today we know that this perspective on Internet was too optimistic. Internet isn’t the utopia
we had in mind; the pros and cons as we know from First Life (FL) apply to the virtual worlds as well.
With the coming of a virtual world called Second Life (SL) during the summer of 2003 and the large
amount of media attention at the end of 2006, people were again optimistic. SL, as the name promises,
is the chance to leave your FL and live an anonymous and free virtual life. But we will again be
confronted with the same problems we know from our FL, unless we dramatically change the way we
build this virtual world.
2 Research
This project started after several SL users asked me to get dressed while I walked naked on the
virtual Amsterdam Dam Square. Since SL does not have a climate and your avatar can’t get cold or
warm, dressing your avatar is for this reason useless. In view of the fact that this is the main reason we
wear clothes in FL I saw this as an opportunity to “leave my clothes at home” in SL. Why did I disturb
other users with this behavior? Why did they respond to this in such a way that my behavior was
offensive from their perspectives? Eventually I was kicked out of virtual Amsterdam by the owner. I
considered these responses as unconscious nonfunctional FL behavior.
Figures show [1] that since the end of 2006 the number of SL users increased dramatically and it
seems that this increase will go on for some time. Besides this increase in users, islands (environments
build by users to reside in) increase with almost a similar speed. These environments designed with
great precision show strong resemblance with our FL environments. There seems to be an excessive
motivation for building ‘realistic’ environments. This goal is caused mainly by the intention of the
founders:
“…we are trying to create a close reproduction of the actual physical world we live in – one that will
be easily comprehensible and useful to us because it will so closely resemble ours.” (Linden, 2006)
1/9
If you would ask any random person to build a virtual environment without knowing the intention of
the founder, it is very likely that this person would build an environment with a strong resemblance
with reality. Each person carries a large ‘visual heritage’ from our FL. This heritage will in almost
every virtual environment be the foundation of a new virtual environment.
Social sciences and architectural studies have shown that human behavior in FL is constantly being
influenced by many environmental factors: for example the frequency of walking, cycling and sitting
correlates strongly to environmental factors [2, 3]. Could it be that this ambition for SL environmental
designers, in building realistic environments has influence on the behavior of SL users? It seems
plausible to say so.
According to these studies you would also expect that it is quite possible that FL behavior is shown
more often in SL environments where there is a stronger similarity to FL.
A possible reason to think why SL users saw my ‘naked’ behavior as offensive is simply because of
the environment. This virtual Dam Square had so much similarity to reality that SL users, perhaps
completely unconsciously, displayed behavior that could be described as functional FL behavior for
purely nonfunctional SL behavior.
A research that relates strongly to this topic is the research by Nick Yee [6]. His ‘interpersonal
distance’ research measured the interpersonal distance of user’s avatars in SL. The research showed
that the social interactions in SL are governed by the same social norms as social interrelations in the
physical world. It showed that the distance that male-male, female-female and mixed groups are equal
to the distance we take in FL while having a conversation.
It would have been interesting for this research if the relative realism of environments had been a
factor in this research. SL is a virtual world which changes by the minute and these recordings were
made more than six months ago, so it is not possible to use Nick Yee’s data for my research.
2 Hypothesis
In what way could an environment be responsible for behavior? Could it be that there is a relation in
the realism of SL environments and the kind of behavior that is shown by SL users? I formulate the
hypothesis as follow:
“The behavior of SL users is strongly related to the realism of SL environments”
This research will examine the influence of realistic environmental design on the behavior of SL
users.
Before I can measure whether the realism of an environment has influence on the behavior of SL
users, I shall first have to make clear what I think SL behavior is and what behavior I wish to measure,
secondly I have to define ‘environment’ and find a means for assessing the level of realism of an
environment.
3. SL behavior
I consider SL behavior as the SL avatar’s repository of all possible actions within an SL
environment. This repository of actions is understood to mean: walking, flying, sitting, lying down,
talking, buying, etc.
A possible way to measure the actions mentioned above is to look at the frequency of all these
actions in environments that differ to some extent in their realism. It is likely to see differences in the
frequency of actions in the environments, but these differences are not of interest for this research. I
only wish to measure some kind of action that is considered to be both a functional FL action and
nonfunctional SL action.
2/9
With this in mind two types of an avatars possible action attract my attention, namely sitting and
lying down. All possible actions of a visitor have a purposeful function, except for sitting and lying
down. I will make this clear with a couple of examples: in order to transport your avatar you walk or
fly; to communicate with other visitors you talk and to dress up your avatar you buy virtual clothes. But
why would you let your avatar sit or lie down if your avatar can’t get tired?
Lying down and sitting in FL can be considered as actions with a biological function. These actions
are ranked as the two most efficient ways for a human being to save energy [3]. In FL you’ll sit for a
discussion with your supervisor or lie on the beach for sunbathing or enjoying the view, simply because
standing costs you an unnecessary amount of energy.
A SL avatar has the functions for sitting or lying down, but does not have the property for losing
energy. So you would expect that SL users wouldn’t let their avatar sit or lie down while talking with
another user or enjoying the view of the virtual beach, but nothing is further from the truth. When
visiting bars, churches, lectures and beaches in SL you’ll see avatars sitting all over the place.
Could it be that an avatar shows this sitting and lying down posture more often in a more realistic
environment? Do SL users unconsciously copy this non-functional behavior more often when the
environments show more resemblance with FL? Do SL users somehow forget that SL isn’t the ‘real’
world?
Note: The behaviour of sitting or lying down depends on the surface your avatar is on. On a towel
your avatar lies and on a chair your avatar will sit. So I will just use the word “sitting” from now on for
both positions of the SL avatar.
In SL we need to distinguish two groups of sitting avatars with both different motivations for sitting:
1. Commercial sitters
Avatars get paid by the landowner of an environment to do some kind of action; these actions are in
most cases sitting in a location that is created by the landowner. The fact is that landowners get paid
every day by Linden Research (the founders) according to the amount of traffic in their world. Because
your avatar will be automatically kicked out a world when there is no action for some time, the land
owner has created ways to let your avatar do actions and receive money simultaneously. Sitting is
considered to be an action. Now it is possible to let your avatar make money while you are sleeping or
at work in FL.
2. Social sitters
I consider this second group of sitters as the social sitters, because they don’t get paid to sit and they
display this behavior while chatting with other avatars. The reason why they sit while socializing is not
quiet clear to me. They don’t get paid, so why do they sit? This group of unpaid sitters is present in
large amounts in SL. I even caught myself showing this behavior as well. I sat on benches in a church
and lay on towels at the beach while talking to other avatars or enjoying the view. Why did I show such
behavior?
I think looking at the sitting behavior in particular is interesting for this research because I expect it
will fulfil two conditions, namely:
1. An increase or decrease in the frequency of the behavior occurs in relation to a change of
realism of the environment
2. It has no functional use
Besides the sitting example mentioned above I came up with seven other measurable examples
which are part of this research’s main hypothesis; for this research I will only measure one of the
examples:
1) Visitors fly more often in less realistic worlds. Since visitors are able to fly you may well
expect that visitors use staircases more often in shopping malls when the reality of these
malls increases.
3/9
2) The use of pedestrian crossings and footpaths increase if the reality of the environments
increase, in spite of the fact that visitors can’t get hurt by traffic since there is no traffic
(yet).
3) Visitors dress their avatars with swimwear more often at more realistic beaches.
4) Although the reward for commercial sitting is higher in some less realistic places, visitors
prefer to make money in more realistic worlds. There is a higher acceptance of less pay in
more realistic environments.
5) Visitors at the beach prefer rectangle to round towels to sit on.
6) Visitors chat less often in more realistic churches since you should be quiet in FL churches.
Chatting is done by typing, so being quiet by not typing is nonfunctional.
7) Superstitious behavior is shown more often in more realistic worlds. An example of
superstitious behavior that would be shown more often: avoiding walking underneath an
open ladder.
From my experience of SL measuring sitting behavior is much easier compare to the other
examples, since it is a more common and clear behaviour which can be measured with simple
observations.
4 SL Environments
To be able to research the relationship between the realism of an environment and behavior we need
a measurement for the realism of environments. An environment can be described from my point of
view as everything that’s visible to a SL user by leaving the SL software interface aside.
How would I judge an environment on its realism? I know that judging an environment on design
and the consequence of design on behavior is complex. Because human behavior is affected by
thousands environmental factors and is therefore hard to tell which factors are directly correlated with a
sense of being in the real world, it would be difficult to measure affects on behaviour in SL
environments. Take for example the fact that visitors can design their own avatar and therefore choose
whether they look realistic or not. Visitors can choose an avatar that is a realist human body, perhaps
their own, but there are many who choose look like a cartoon hero. However most of the avatars stick
to a realistic looking human form. Since all these avatars are part of the environment this could be an
important factor for a sense of realism, but this is very difficult to measure. Other possible important
factors are the presence of environmental sound or the quality of textures.
In view of the fact that this is not a research about “judging environments on the sense of realism” I
have to restrict myself to one or two factors for judging an environment by its level of realism.
SL gives everyone the opportunity to build a virtual environment, building an environment for this
research would have been the easiest way for measuring behavior if money wasn’t an issue. Building
an environment isn’t the hardest part, but generating traffic to your world it is! To get users into your
environment is only possible when you can offer them something. Since getting free money is the main
reason for users to visit an environment and the budget for this research is zero, building my own
research environment was out of the question. Besides this, it will always be better to measure
behaviour in existing environments so you get a broader variety of observations.
After many hours of SL experience I came up with a way to judge an environment on its level of
realism, because most environments are designed with the intention of appearing realistic, but there
were some exceptions. In some environments the designer clearly intended an environment with less
resemblance to reality. These environments could be interpreted as environments with ‘cartoon’ and
‘fantasy’ themes and therefore judged as less realistic.
4/9
Fig 1. An impression of two types of realism left: realistic, right: non-realistic
For this research I restricted my judgement on reality of an environment to the visual aspects
(textures, proportions etc.) and the intention of the designer. And I only use two levels of judgement:
realistic and non-realistic. (fig 1)
Finding realistic environments was not difficult, since these are the best visited environments. The
non-realistic environments (which are harder to find), that can be used for this research are the
environments of the ‘Furries’. Furries are to be known in SL as fantasy avatars, mixtures of human and
animal. The goal in these environments is to look as non-realistic as possible while inhabiting these
fantasy non-realistic environments. The users here are some of the first inhabitants of SL and from
personal experience they display their scepticism of the commercialisation of SL and are still present in
relatively large amounts.
5. Method & Materials
The approach I chose was a measurement of sitting in three realistic and three non-realistic
environments. Three most popular Furrie environments on the day of observation will be chosen as
non-realistic environments and three random non-furrie environments will be selected for the realistic
environments.
The data for this research was retrieved by observation. In each environment a group of avatars will
be observed for half an hour by the researcher’s avatar. Every five minutes several screenshots from
different angles were taken
The location of observation within this environment depends on the position where the group
gathers. This can be anywhere in the environment but in most cases it was near where avatars entered
the environment. The size of the group and percentage of sitters was recorded every 5 minutes.
Observation was chosen for retrieving data because the environment code is closed to SL users. So
monitoring the actions of all present avatars by reading SL code in these environments was therefore
out of the question.
5/9
6 Results
Realistic Environments
Sitters (%)
Bourton Village 57
Phatland 32
Japan Resort 58
Average (%) 49
1. Bourton Village 2. Phatland 3. Japan Resort
Fig. 2 Selection of three realistic environments
Non-realistic Environments
Sitters (%)
Luskwood 20
Aiot's Toon Wonderland 19
Northstar 22
Average (%) 20
1. Luskwood 2. Aiot’s Toon Wonderland 3. Northstar
Fig. 3 Selection of three non-realistic environments
More detailed research results can be found in the appendix.
6/9
7 Conclusions and Discussion
The results show that the differences in percentage are as hypothesized. Avatars in non-realistic
environments do seem to sit less often than in realistic environments. More research would be needed
before I can say whether the main hypothesis, “The behavior of SL users is strongly related to the
realism of SL environments”, is true. The sitting hypothesis is a small part of the main hypothesis, but
it is possible to say that at least these results do not refute the main hypothesis.
There were some interesting things I noticed while engaged in the observations. For instance, the
locations where avatars gathered in the environments differed to some extend. In all six environments,
avatars gathered in a location that was intended by the owner as a place for gathering. In all three
realistic environments, these locations were a bar with bar stools and a couple of sofas. With the
exception of Aiot’s Toon Wonderland, there was no bar in the non-realistic environments near where
the avatars gathered. In the locations where avatars gathered in the non-realistic environments some
places were designed for sitting but compared to the realistic environments these places were less
‘attractive’ for sitting such as an ‘uncomfortable’ wooden bench or a casket.
This type of research requires more data than I have been able to collect. It seems likely that this
isn’t the right time for measuring and doing this kind of research yet, because the amount of SL users
needs to increase much more in order to make plausible judgements. As I experienced, the community
of visitors who prefer non-realistic to the realistic SL environments is too small.
A possible solution for these problems is to create a “SL laboratory” environment, a place that is
financed and realized by the academic world for SL research. With such an environment better research
can be done in the behavior of users of virtual environments.
While doing this research, an environment that I judged as non-realistic never appeared in the top-20
of most popular places in SL.
So it seems that the most realistic worlds are the best visited worlds and behaviour is in some way
controlled by the owners of these worlds. If you act in ways that would be condemned in real public
life, such as taking off your clothes, like I did before I started this research, then you are likely to be
expelled by the owner. Sometimes it seems as if they are afraid that visitors would disturb the feeling
of realism for other visitors. Fewer visitors mean less money for the owners. So you should behave like
you would in RL, otherwise go and find a non-realistic world!
Aside from the intentions of the founder, it is logical that most virtual worlds look like a realistic
world. SL is most used as a tool for socializing[5]. Socializing in SL could only be easy for a broad
group of users when it has a strong resemblance with reality. If you would like to attract the attention
(of a woman) in a virtual world you should act in such a way that she would understand. So you dress
up your avatar, as you would in FL, with clothes that reflect your personality and behave as you would
in FL. Taking off your clothes to impress a girl would not be understood as a form of socializing but as
offensive.
In referring to my hypothesis and expectations, I am quite certain that the frequency of particular
types of behavior is affected by a change in realism in the SL environment. And the most important
reason why people display different behavior in different environments depends strongly on the
intention of the designer. Here I have not dealt with reasons why we almost automatically behave in
ways that seem to follow the intentions of the designed environment. But it is a related question to my
investigations. Therefore this research should be considered as a starting point for further research on
the affect of the intention of the designer on the behavior of virtual world visitors.
In the introduction I referred to the optimism and the utopian views about the internet and the
expectations associated with the coming of Second Life. From my own research where it seems that FL
behavior is shown more often in realistic SL environments, anti-social behaviour such as racism or
fraud, seems just as likely to happen. In SL the policing is determined by individual owners of
environments with some global SL restrictions, but unlike the utopian views of a safe SL world, I
7/9
propose that SL is an ‘unsafe’ world with the potential for crime and comparable to the boundaries as
we know it from our First Life.
References
[1] Nood, de D., Attema J., The Second Life of Virtual Reality, 2006
[2] Saelens B.E., Sallis J.F., Frank L.D. Environmental correlates of walking and cycling: findings from the
transportation, urban design, and planning literatures. Ann Behav Med. 2003;25:80–91.
[3] De Bourdeaudhuij I, James F. Sallis, Brian E. Saelens, Environmental Correlates of Physical Activity in
a Sample of Belgian Adults
[4] Passmore R., Durnin J.V.G.A., Human energy expenditure, 1955
[5] Fetscherin M., Lattemann C., User Acceptance of Virtual Worlds, 2007
[6] Yee N, Bailenson JN, Urbanek M, Chang F, Merget D, The Persistence of Nonverbal Social Norms in
Online Virtual Environments, 2007
Acknowledgements
I would like to thank Prof. Dr. B. Haring, Leiden University, for being the supervisor of this research
and Sonja van Kerkhoff, www.sonjavank.com, for proofreading the paper,
8/9
Appendix – Observations
Realistic
26 July - Bourton Village - 20:35 26 July - Phatland - 19:40 27 July - Japan Resort - 10:40
Time Number of Number Time Number of
(min) avatars Number of sitters Time (min) of avatars Number of sitters (min) avatars Number of sitters
0 9 5 0.555556 0 18 4 0.222222 0 6 5 0.833333
5 8 5 0.625 5 20 5 0.25 5 11 5 0.454545
10 9 6 0.666667 10 21 8 0.380952 10 6 3 0.5
15 11 5 0.454545 15 21 7 0.333333 15 10 5 0.5
20 14 4 0.285714 20 24 9 0.375 20 10 6 0.6
25 9 7 0.777778 25 27 9 0.333333 25 13 7 0.538462
30 11 7 0.636364 30 23 8 0.347826 30 15 9 0.6
average 0.57166 average 0.320381 average 0.575191
average total realistic 0.489078
Non-realistic
23 July - Luskwood - 20:00 26 July - Aiot's Toon Wonderland - 21:55 26 July - Northstar - 12:40
Time Number of Number Time Number of
(min) avatars Number of sitters Time (min) of avatars Number of sitters (min) avatars Number of sitters
0 17 2 0.117647 0 7 1 0.142857 0 5 1 0.2
5 19 2 0.105263 5 7 1 0.142857 5 6 1 0.166667
10 16 3 0.1875 10 9 0 0 10 8 1 0.125
15 16 3 0.1875 15 9 2 0.222222 15 7 1 0.142857
20 17 4 0.235294 20 8 3 0.375 20 11 2 0.181818
25 14 4 0.285714 25 10 2 0.2 25 10 3 0.3
30 15 4 0.266667 30 12 3 0.25 30 7 3 0.428571
average 0.197941 average 0.19042 average 0.220702
average total non-realistic 0.203021
9/9
