[ Pobierz całość w formacie PDF ]
1
A Plague of Viruses: Biological, Computer and Marketing
1
Jeffrey Boase and Barry Wellman
Sept 21, 2001
Current Sociology, forthcoming
Abstract
We analyze the transfer of biological, computer and marketing viruses. Despite differences
between these three types of viruses, network structure affects their spread in similar ways. We
distinguish between two forms of networks – densely knit and ramified – and show that biological,
computer and marketing viruses all behave in similar ways depending on the form of network.
Densely knit networks promote the quick dissemination of a virus, and increase the odds that
many of the members will become infected. Ramified networks allow a virus to disperse widely,
jumping between different milieus. In the end, the spread of viruses in the real world involves a
combination of both densely knit and ramified networks, which we call “glocalization”.
Viruses Thrive on Networks
As the twentieth century becomes the twenty-first, biological viruses have been
joined by two other types: computer viruses and viral marketing. Are there similarities
between biological viruses, computer viruses and viral marketing other than the common
term in their names?
All three depend on networks for their spread rather than growing
in situ
. A virus
spreads through contact, be it purposeful or unaware juxtaposition. Without networks,
viruses would lead lonely lives without affecting anyone except for the original hosts. For
example, biological viruses do not come from toilet seats: They are transmitted person-
to-person (or at least species-to-species). Computer viruses almost always flow between
human beings. Viral marketing, where Harry tells Sally what cool brand to buy or what
Internet petition to sign, is inherently and consciously person-to-person. Even at its
weakest form, viral marketing is by observation, as when one high school student
observes the star quarterback wearing Fubu (last year) or Ecko (last month) clothing.
Just as viruses differ, so too do the networks of biological, computer and
marketing viruses. One similarity is that frequent contact increases the likelihood of
"catching" a virus. Another is that densely knit groups accelerate the spread of a virus but
keep it within a confined population. By contrast, viruses spread more slowly in sparsely
knit networks but will jump to new milieus and ultimately spread more widely. We
examine here all three kinds of viruses – biological, computer, and marketing – their
similarities and their differences. We are especially interested in how different kinds of
networks affect how such viruses operate.
Before we compare these three kinds of viruses according to network structure,
we first compare them:
1
We appreciate the advice of Alden Klovdahl and Martina Morris with respect to biological viruses,
Danyel Fisher and Douglas Tygar with respect to computer viruses, and Eszter Hargittai, Valdis Krebs, Bill
Richards, Emmanuel Rosen, Patrick Thoburn and Matthew Stradiotto with respect to viral marketing.
2
Biological Viruses:
Biological viruses have been the best studied of the three
2
.
Please note that our comments here refer to bacteria as well as viruses, but we use the
term “biological virus” to preserve the literary conceit that organizes this issue.
The preponderance of biological viruses need contact with humans or other
animals to be transmitted, although some may be transmitted indirectly by human-aided
contact (biological warfare; tainted animal feed). Transmitter and receiver must be near
each other, but they need not have any social contact (see Table 1). The need for social
proximity or contact means that the speed by which these viruses spread is low. In many
cases, the young are the most vulnerable.
Biological viruses frequently mutate. Preventing their spread requires physical
isolation, tracing the spread of the viruses is hard, and eradication and healing is
accomplished with great difficulty by medical personnel. Widespread prevalence can
have important second-order consequences, with the debilitation of a population through
disease and the desire of apparently-healthy individuals to flee the infected area and
quarantine the ill.
Computer Viruses
: Computer viruses are not born but deliberately made by
either hackers or cyber-warfare experts. They are commonly transmitted by the Internet,
although file-sharing is another vector just as needle-sharing is for biological viruses.
Like biological viruses, transmission often occurs without deliberate human intent (after
initial creation).
Transmitter and receiver need not be physically proximate, but must have
connected communications, most typically the Internet. The transmitter must minimally
have awareness of the receiver's Internet address. In many cases, social ties are vectors.
Friends and acquaintances inadvertently infect friends by sending viruses that have been
hiding in one system as attachments to an existing file: Strongly tied close friends are
especially apt to do this because they have the most contact. In other cases, the virus
ransacks the host transmitter's Internet address book: The greater number of weak ties in
such books means that more acquaintances will be infected than close friends.
Children and teenagers are probably the most vulnerable because of their general
reluctance to take protective measures. Although one might think that corporate
executives, protected by well-organized information technology units, would be the least
vulnerable, large organizations have repeatedly suffered virus attacks because of their
reliance on Microsoft Outlook, a favorite target of hackers (Taylor, 1999).
Computer viruses transmit at hyper-speed. The Nimda worm spread so fast that
the New Brunswick (Canada) provincial government had to shut down their computer
systems for a day, September19, 2001 and use "old-fashioned" techniques. "We're using
phones; we're typing; we're conducting a lot of business face-to-face and by fax,"
reported Susan Shalala of the Supply and Services Department (Canadian Press, 2001).
Viruses mutate readily, as hackers obtain the original virus and use "script kits" to modify
it. Dealing with computer viruses has itself created sizable corporations providing anti-
virus solutions (e.g., www.symantec.com). Anti-virus measures include interception
programs to recognize and neutralize viruses as they enter a system and reverse path
tracing to identify a virus's origins.
2
See also Ellen Gee’s and Maticka-Tyndale’s papers in this issue.
3
The social consequences of a virus attack range from the annoyance of having to
maintain up-to-date anti-virus software and backups, through the personal difficulties of
losing information on one's hard disk, to serious organizational disruption. The
embedding of the Internet in the developed world's everyday life means that large-scale
or well-placed attacks might cause societal paralysis and the resulting socioeconomic
isolation of units within the society.
Viral Marketing:
Although the practice has been around for millennia, viral
marketing, sometimes called “buzz marketing” is the newest recognized type of virus.
Viral marketing refers to the marketing of a product or service by word of mouth. It dates
back to Biblical days, as when Jesus said to his disciples, "Go ye into all the world, and
preach the gospel to every creature" (Mark 16:15). Unlike biological and computer
viruses, many people welcome viral marketing because it gives them new information, a
chance of feeling socially accepted, being in the know, and following the latest fashion.
Although this form of marketing tends to be overlooked because it is not easily traced,
word of mouth may make or break any product, despite formal advertising (Rosen, 2000).
The idea of having commercial viral marketing campaigns was not introduced
until 1940’s (Rosen, 2000), and corporations rarely choose it as their primary mode of
advertising. Diffusion can be through strong ties who are trusted with their opinions, or
through weak ties who observe the tag lines appended to messages. But the virus can also
spread merely through observation of physically-proximate role models / “aspirational
leaders” – such as the stars of high school sports teams (Thorburn, personal
communication, 2001). It flexibly supplies needed and wanted information at low cost.
The interpersonal nature of viral marketing means that it conveys information more
precisely and efficiently than the mass media to people who have higher probabilities of
wanting the information.
In the 1950’s, Elihu Katz and Paul Lazarsfeld (1955) created the classic concept
of the “two-step flow of communication”: persuasive communication diffused through
the mass media (step one) as interpreted by influential people through their interpersonal
relationships (step two). In
Medical Innovation: A Diffusion Study
, James Coleman et al.
(1966) showed how knowledge about drugs spread informally among physicians
3
. He
found a snowball effect where well-connected doctors act as early adopters and then
influence non-adopters.
A research field soon developed, “the diffusion of information” (Rogers and
Kincaid, 1981; Rogers, 1995). Perhaps its greatest proponents were public health
officials, who soon became persuaded that health information about contraception use
was best given to Third Worlders through interpersonal ties (Valente, 1995; Valente and
Watkins, 1997). It might be said in such cases that viral marketing is used in such
situations to combat biological viruses. Yet, the most extensive use of viral marketing has
not been to prevent disease but to promote various forms of contraception.
The Internet has brought its own novel forms of viral marketing because the
technology makes it easy to pass on messages – to one or one hundred closest friends.
Computer-based viral marketing takes several forms:
1. There is the conscious circulation of petitions or similar appeals to do
something. For example, a petition for the United Nations to look into the possibility of
3
Also see Van den Bulte and Lilien’s critique (2001).
4
election fraud during the 2001 presidential election in the United States. We have
received similar petitions several times in the past month, and many times in the past
decade.
2. There is the forwarding of rumors and humorous stories. Many people, for
example, maintain “joke lists” which circulate purported humor for short or long periods.
For example, one of us has received the story about the Nieman Marcus cookie recipe
twenty-seven times in the past eight years.
4
3. There has been quite successful latent viral marketing by taglines to email. For
example, some Internet service providers such as Hotmail and Yahoo provide free email
to users who agree to have a phrase added to every mail telling the receiver that they can
get free email by going to
http://explorer.msn.com
. We will say more about this later.
4. There is a movement now in the computer game industry to use the Internet as
a vehicle for promotion. "Many of the new games are viral, meaning that they permit
players to spread the games by e-mail to friends." (Marriott, 2001: D1)
Viruses as Social Networks: Densely Knit Groups and Ramified Networks
The spread of viruses is shaped by both the nature of interpersonal relationships
and the composition and structure of the interpersonal networks of which these ties are a
part. Social network analysis has developed concepts and procedures for analyzing such
networks (Marsden and Laumann, 1984; Wasserman and Faust, 1994; Wellman, 1988;
Wellman 1997; Garton et al., 1997). We use the social network approach here to compare
the similarities and differences in the forms of ties and networks that shape the movement
of biological, computer and marketing viruses. In doing so, we show how the structure of
social networks – the pattern of relationships that connect people and their computers –
has important consequences for the spread of viruses: how much of the population is
infected and how fast this infection takes place.
There are two structural archetypes. In
densely knit groups,
most members know
each other, are in frequent contact with each other, but have little contact with outsiders
(Figure 1). In network analytic terms, such groups are densely knit and tightly-bounded
(Garton et al., 1997; Wellman 1997). By contrast, in
ramified networks,
few members
are in contact with each, and a large portion of interactions are with outsiders. Such
networks are sparsely knit and loosely-bounded. Reality, of course, often occurs in the
continuum between these archetypes. A common one in contemporary societies is
glocalization
: rather densely knit clusters of relationships (usually at home, at work, and
with kin) that also have many ramified ties to other people and groups (Wellman 1999;
Hampton 2001). For the sake of clarity, we concentrate here on two ideal types: densely
knit groups and ramified networks.
4
See www.neimanmarcus.com/about/cookie_recipe.jhtml; www.urbanlegends.com/food/two-fifty/neiman-
marcus.html.
5
Figure 1: Three Models of Network Structure.
Densely Knit Groups
A virus can move quickly within a densely knit network because almost every
member has frequent contact with every other member. This is true for all kinds of
viruses. Densely knit groups allow for synergy, in which a particular group member is
often exposed to a virus through frequent contact with other group members and, in turn,
exposes and re-exposes other group members to the same virus.
“Homophily,” birds of a feather flocking together (McPherson et al., 2001;
McPherson and Smith-Lovin, 1987), increases the likelihood that biological, computer
and marketing viruses will spread even more quickly through a densely knit group. Not
only are people in densely knit groups in direct contact with each other, they also tend to
share common characteristics, such as socioeconomic status, tastes, or attitudes
(Erickson, 1988; Feld 1981, 1982). They are also have considerable influence over each
other’s decision processes (Cross et at., 2001). This increases the likelihood that they will
have similar patterns of behavior and hence be more likely to be exposed to the same
virus.
Biological Viruses:
Consider how the AIDS-causing HIV biological virus may
have spread in the San Francisco Bay area by this kind of group synergy in the late 70’s
and early 80’s. During this time the gay community there was a closely knit group, with
frequent unprotected sex between members. The ideology of sexual freedom as having
multiple partners and the relatively bounded structure of the group meant there was often
overlap between sexual partners (Rotello, 1997). Overlap caused the virus to spread
quickly within the group.
Computer Viruses:
There is reason to believe that computer viruses spread within
densely knit groups of computers in much the same way. A computer virus is a file which
has the ability to attach itself to other files (Aunger, 2001). Once these files are modified ,
they will carry out a list of instructions called a protocol. If a virus lies latent until a
certain date or a trigger event, it is called a “trojan” after the Greek legend of the Trojan
Horse.
[ Pobierz całość w formacie PDF ]