The following screenshots were derived during the Layout phase
of Social Network Fragments. All of these images focus on
one individual's email data - "Mike." As this structure
is meant to provide a portrait of Mike, it is easier to understand
the structure presented by having some background
information about Mike.
Below are a series of screenshots and descriptions intended to
provide a basic level of information about what data is being presented,
how, and why, and what this tells us about the subject.
The Layout phase allows us to see the evolution of the layout positioning,
so that we can visually see the spring system at work, repelling
and attracting nodes based on proximity and appropriateness. This
is most compelling by watching
the evolution as a movie. This movie shows the motion, but it
does not show the individual people that are attached at the end
of each line.
In the flat images below, we show the final positioning of Mike's
world, explaining what the different clusters mean.
In these images, lines represent links between two people and
baby names indicate a person. Normally one would see the email addresses
of the individuals, but we've used baby names to protect their privacy.
It is easy to underestimate the power of these graphs in this form,
because it is not personal and the connections do not have personal
relevance. This is the reason why Social Network Fragments
is a far more powerful tool for the subject than for a stranger.
In any case, to explain this technique, we have chosen to present
our character Mike.
The following color chart applies to all these images. We color
coded the email address by which Mike knew the individual; there
are five major colors representing the five major facets that Mike
felt his addresses encompassed. Within each major category, we assigned
slightly different colors to different addresses, thus you will
see some slight variation in the colors. (Bulk refers to mailing
After the spring system settles in Mike's worldview,
we are able to see Mike's worldview, as comprised by a set of
small clusters, indicating groups of people drawn towards one
This image represents everyone that Mike knows and is aware of.
There are approximately 4,000 different individuals spread between
some one hundred clusters. As can be seen, there are a few clusters
which contain the majority of the people that Mike knows and many
clusters that contain only a handful of people.
Note the colors of these clusters. By far and large, each cluster
is of a homogenous color - several clusters of blue, for example,
show us that Mike knows several clusters of people from school,
and that these clusters do not know of each other. In Mike's case,
we see that the fragments of his social networks are roughly corrolated
with the function by which they know Mike. This separation in Mike's
social network makes sense for an individual -- understand that
if we were to graph the email communcation patterns at a company,
we'd hope to see connections between these different clusters.
By focusing closer into the main region of the above image, we
can see a better view of the primary clusters.
Such closeup examination allows us to focus on particular clusters,
such as this relatively simple one, an interesting chain of clustered
Here we see four different clusters of people, each connected
by one link. This image is particularly revealing to Mike, who recognizes
the individuals who connect these chains of groups of connected
Note that the coloring for the majority of these nodes is consistent,
with a slight variation in shade occuring in the lower portion.
Again, we see that the people Mike knows tend to know each other
in the same way that they know Mike.
A complex cluster in Mike's network shows us what happens when
a lot of people are known in two different contexts that eventually
converge. In the center of this image, we see that Mike's friends
(blue) tend to know other of Mike's friends, but that there is some
overlap between people he knows from school (brown). We'll revisit
this portion of the image in later images.
The cluster in the lower left corner contains an excellent example
of "bridge" individuals: two people who know Mike in different ways
and who know each other, but do not necessarily know that the other
knows Mike. There is a link between the green cluster (people who
know Mike through his website work) and the brown cluster. Were
Mike needing to keep these identites separate, it would be extremely
important that the green and brown bridge individuals not become
aware that the other know Mike.
Looking closer at the complex cluster begins to reveal the strucuter
of Mike's social network at school and of post-school friends. Here
we see that there are twelve links connecting Mike's friends to
people he knows from school -- and that clearly there is a strong
connection between these two communities.
Color variations within a complex cluster show us where different
social settings are mixed together. Here, we see that some of Mike's
friends seem to be more aware of people from his school than of
his other friends. The occasional blue clusers inside the brown
(and the brown clusters inside the blue) are expected in this case.
A brown clump inside a blue region indicates that there is a circle
of people who all know each other and know Mike, but have ended
up knowing Mike's friends better than people from school.
Colors only denote how they know Mike, not how they know each
other, thus when we see four or five separate clusters that are
all the same color, this only means that those groups are not aware
of each other but still know Mike in the same social setting.
This series of images, along with the explanatory
movie and the description of Mike
are intended to provide the reader with an overview for the Layout
section of Social Network Fragments. This layout scheme is
the basis for the Visualization
phase. The images provided reveal the complexities of data that
can be conveyed, but yet these images are not without problems.
In particular, the layout fails to properly convey what happens
when two unrelated people with shared ties are visually close due
to similarities in connections. Given this and other problems, a
critique of this scheme
is necessary and thus part of danah's thesis.