Update: Since the creation of this document, there have been several
documented instances of conspiratorial "googlewashing."
A
Register article,
as mentioned in a
post
on
Slashdot,
which chronicles the
"
Googlewashing"
of the term
"
second superpower,"
is an excellent example of social network theory and its effect on language.
Unfortunately, the Register incorrectly assigns the blame on alluded
conspiratorial behavior of webloggers.
In fact, the phenomenon is not the result of a conspiracy,
but a natural consequence of the dense network that interconnects them.
For those unfamiliar with the article, it describes, how the phrase
"
second superpower,"
coined by Patrick Tyler in the New York Times
on the 17 February 2003,
was coopted by James F Moore in his online weblog a short time later.
Originally used to describe anti-war protests,
its meaning was altered in Moore's blog as a call to net users to
organize themselves as a
"
superpower."
What followed, was the slow but almost complete replacement
of the original definition with the new definition, as defined
by ranked returns from the
Google
search engine.
Social Network Theory
Members of a social circle benefit by being able to
draw on the resources provided by members of their network.
A network is a pattern of links or connections among a set of units.
In the case of
social networks, the connections are relationships among a group of social units.
These links or relationships are defined and maintained by persistent,
repeated actions between social units.
A network of units that share a large number of strong links relative
to other units form what is known as a social circle or clique.
Strong links are typically reinforced by homophily, the tendency
to interact with others, who are similar to oneself.
Within a social circle members benefit by being able to
draw upon the resources provided by other members of the network.
The availability of common resources provided by a network,
from which a node can draw, is called social capital.
The quantity of social capital a given node posesses
is a function of the interconnectivity of that node and
the extent and shape of the surrounding network.
Extremely well connected or "popular" nodes in a dense network are
known as broker nodes and possess tremendous social capital.
Although similar to heirarchies in that all nodes do not
have equal social capitial, it is important to emphasize,
that a social network has neither clear boundaries nor formal leadership.
Furthermore, social networks can reach such a high level of complexity,
that individual nodes can only perceive relationships that they
maintain directly and are often unaware of nodes further removed.
Although they share a network and are bound by homiphily,
nodes typically do not manipulate remote nodes directly.
Social Networks in the Register Article
It's not the number of nodes but the density and layout of the network
that provides social capital.
To return to the thesis of the article, the Register erroneously
accuses webloggers of spinning an
"
alternative, neutered definition"
of the phrase
"
second superpower."
This supplantation of the original definition is not the result
of a conspiracy but a natural consequence of the dense network
that interconnects the
"
A-list bloggers"
in conjunction with
Google's method for ranking pages.
It is stated in the article,
Although it took millions of people around the world to compel the
Gray Lady to describe the anti-war movement as a "Second Superpower",
it took only a handful of webloggers to spin the alternative meaning to
manufacture sufficient PageRank to flood Google with Moore's alternative,
neutered definition.
This paragraph defines two social circles, one which is large but sparse
and the other which is small but dense.
The small group, due to its density, possess greater social capital
than the larger group, which manifests itself through the rapid disemination
of information among those in the social circle.
Because Google ranks pages according to interconnectivity, otherwise
known as popularity, the new definition created by the small, dense
social circle supplanted the original definition.
The article goes on to say,
But the real marvel is that they did it with so few people. Pew Research
Center's latest research says the number of Internet users who look
at blogs is "so small that it is not possible to draw statistically
meaningful conclusions about who uses blogs." They peg it at about
four percent. But we're looking at a small sub-genre of blogdom, the
tech blogs, and specifically, we're looking at an 'A list' of that
sub-sub-genre.
This paragraph emphasizes the small size of the social circle, calling
out a
"
sub-sub-genre,"
which clearly represents the network brokers.
Two points are overlooked.
First, the described
"
A list"
is simply a group of network brokers.
There exist many other nodes within the social circle whose connectivity
to the broker nodes gives the network its density and social capital.
Second, the paragraph is overlooking an important tennant of social
network theory.
It's not the number of nodes but the density and layout of the network
that provides social capital.
The article goes on to repeat the same mistake.
Which means that Google is being "gamed" - and the language perverted -
by what in statistical terms in an extremely small fraction indeed.
Again, from the perspective of social network theory it is not
the number of nodes but the density of the network.
If those networks have similar searchable phrases with competing
definitions and, if the search engine ranks phrases according to
interconnectivity, the dense network's competing definition will
rank higer than the sparse's.
Perceived Collusion within Social Networks
The collective behavior of a network gives the illusion of
conspiracy but it is, in fact,
only a manifestation of network dynamics
As stated before, in complex social networks
individual nodes only perceive immediate relationships.
Although they share a network,
there is no way they can affect removed nodes in any direct manner.
Instead information is shared and linked.
Thus, in this instance, the collective behavior of a network
gives the illusion of conspiracy or collusion but it is, in fact,
a manifestation of the underlying network dynamics.
For example, the Register states,
All a strange coincidence, no doubt, but the picture darkens when you
look at a parallel conversation taking place elsewhere, whose hyperlinks
contributed to the redefinition, and help explain how this semantic
ethnic-cleansing took place so quickly.
This is a perfect example of how conspiratorial behavior can be
attributed to a complex network, over which no one node has control.
While the outcome is the same as if there were organized collusion,
the conspiracy is a manifestation of the circle's social capital.
Power Law
Social networks compete for relationships.
In order to build social capital it is in the benefit of a node
to form relationships with as many nodes as possible.
The nodes that are most desirable to form links with
are those that already posses much social capital.
Thus, nodes with large amounts of social capital
will accumulate social capital more rapidly than
those which posses less social capital.
This competition for a limited resource in conjunction
with the desire to improve social capital by associating
with nodes already with a lot of social capital leads
to a power-law distribution.
It has been shown that the
distribution
of links on the web does, in reality, scale according to a power law.
The inescapable ramification is that network brokers will
always
dominate the returns from a search engine that ranks pages according to
popularity and interconnectivity.
Solutions would be the redistribution of links within a
network or the modification of the search ranking criteria.
A redistribution of links would disperse the social capital of
network brokers among nodes with much less social capital.
For instance, the network broker, Slashdot,
would disperse its content to multiple unrelated
nodes, in order to share its links with other nodes.
Alternatively, search engines could employ different ranking criteria
than popularity to rank pages.
Such criteria could be readability, size, number of references, or
the level, with which the results please the writers at the Register.
