Network Studies (part three)

Connected, or What It Means to Live in the Network Society
Steven Shaviro

Toward a Structural Theory of Action: Network Modes of Social Strcuture, Perception, and Action
Ronald S. Burt

Structural Holes: The Social Structure of Competition
Ronald S. Burt

Theories of Communication Networks
Peter R. Monge
Noshir S. Contractor

Small Worlds: The Dynamics of Networks between Order and Randomness
Duncan J. Watts

No editing done.

Connected, or What It Means to Live in the Network Society
Steven Shaviro

Shaviro is using science fiction as a way to describe network society in the same way Marxist theorists used realist novels to further their scholarly projects. He sees theory and science fiction connected since both types of writing “seek to grasp the social world not by representating it mimemticaly but by performing a kind of ‘cognitive estrangement; upon it…so that structures and assumptions that we take for granted, and that undergird our own social reality may be seen in their full contingency and historicity” (x).

The metaphor of network has been taken up by corporations since it provides the backdrop for both a perfect, self-regulating environment, but also touches on the plug-in which makes agressive predation and social Darwinism acceptable, and moreover, provides a “logical model capable of overcoming a contradiction (an impossible acheivement, as it happens, the contradition is real)” (Levi-Strauss qtd in Shaviro 4) that “reconciles the conflicting imperatives of aggressive predation on one hand, and unquestioning obedience and conformity on the other” (4).

Concepts evoking Latour on page nine, bottom of the page. Everything has the same ontological status and distance/perspective. This is not to say Shaviro is using Latour’s ideas about networks. In fact, Shaviro seems to be moving towards the sociology of the social.

Nice description using the text Noir to explain how viral marketing works. (14)

Continuing through the book it becomes apparent that Shaviro’s discussion on media is pretty important. Using McLuahan’s terms like “cool media,” Shaviro describes TV as an inviting but not altogether passive experience for users. While the user (a la Benkler) can not make anything and broadcast artifacts back on the same television signals, it allows the user to participate in a public discourse free of penalites. The user can yell, talk back, or turn off the machine, and yet it’s still there and often the touchstone source for inofrmation going on beyond the user’s immediate locale. It presents only certain points of view and a small number of choices to deal with the conflicts it often brings to the user; it indoctrinates the user into a worldview that becomes the user’s worldview–it’s a soft boot camp.

The Internet and the Web are also cool media, but they are so unfinished they become more involving and beckon the user to create content or add content or explore the frontiers of it (surf) to see the spectacle of the mediea frontier. The Internet is a “haptic” space, one where the user is constantly involved watching every minute detail and touching everything with her disembodied finger synched to her mouse. She must read, click, choose, wait for downloads, type quickly to maintain her own in a chat room discussion; in short she must focus on the machine and she must always be multitasking. This, of course, makes the Internet more inviting (addictive) than TV. (6-7)

The talk of control society on page 31-34 is interesting.  Here Shaviro is discussing the work of Deleuze; specifically his work on the control society.  The control society is different from Foucault’s disciplinary society because of the way it uses instant communication and the constant gathering of information to shape the behaviors of those existing within its bounds.  The process of control through networks is what Deleuze refers to as “modualtion” ( 32), a “self-transmuting moudling continually changing from one moment to the next, or like a sieve whose mesh varies from one point to another” (Deleuze qtd in Shaviro 32).

On 34 Shaviro echos Latour in claiming that all of society is now flat since it’s all part of one larger network.  There is no more doubling between the documents that signify a person since in an information based society that information equates to the person; that information is used as a mediator to press an individual actant into a particular action.

The subject no longer exists as an “empirico-transcendental doublet”; its structure has collapsed back onto a single plane.  All the familiar features of the network follow from this collapse into immanence: representation gives way to simulation, creation ex nihilio is displaced by mixing and sampling, murky depths give way to glittering surfaces.  (34)

Surveillance camera on 36 would be an instance of how the control society works.  In the next few pages is an example of resistance, the Surveillance Camera Players.  It’s a small hack when they perform different politcial charged pieces in front of the cameras; the infect it with the network wit their  ideas and force the surveillance network to become the ultimate audience at no overhead to them.  It’s a hack in the sense of harnessing a network and forcing to do what it does efficiently, but not for the purpose it was designed.  In this case, the cameras are setup to deter and record crime.

Digital technology promises access to everything, and this–according to Shaviro–is a lie.  The promise is given with market sensibilities as the protocol to decide who has access.  On pages 42 and 43, Shaviro recounts how hip hop has suffered from this arrangement: artists are punished for sampling, even small amounts within limits of traditional fair use laws, as they use the technologies of the digital age to remix and rewrite culture.  Free speech and the ability to create are not covered in the instances he (Shaviro) recounts.  I would suppose, though, this comes from the fact that the Constitution and the Bill of Rights protect you and your interactions with State and Federal governments, not corporations or private individuals wanting to sue you for copyright infringement.

47–an interting take on Digital Right Management (DRM).  The best way to control info is not to censor it, but embed within the info itself the means to stop an actant from accessing the information.

50-51 is the recounting of decisions through the Supreme Court to protect the First Amendment rights of corporations.   While I don’t doubt this happening,  I do find it appalling that corporations–entities made real through the filing of paper work to limit damage to primary investors in case their enterprise fails–should be given the rights of a living, breathing citizen.

Jeter’s Noiris built on a dislike for the gift economy and open software or copying.  It’s the antithesis of Shirky’s work concering how collectives get thigns done (think about Shirky’s example of the programmer who creates the fix for MS Outlook).

Unlimited copying is now techinically possible, but so is a system of tracking so precise, and so extensive, that not a single byte, in any machine, anywhere, will escape being identified and accounted for. (62)

Each individual may be a node in the network, but the corporation is the network itself.  The more corporations are recognized as persons, as has increasingly been the case under United States law, the less unincorporated indviduals are able to be so recognized.  (63)

Shaviro calls the “mediasphere” our “nature” since we know, understand, and connect to the products, logos, and Hollywood icons found only in the mass media.  “That is to say, the electronic media are the inescapable background against which we live our lives and from which we derive our references and meanings” (64).  On page 70 Shaviro disagrees with Debord in the concept of a grand spectacle which detaches people from a directly lived life.  For Shaviro, life has always been a time where images were taken from their original context and used to form the world people must navigate to survive.  Shaviro agrees with Warhol that television hides nothing, but makes viewer aware that everything “we see, hear, and feel is just a representation” (71).  Shaviro continues on to claim that life isn’t made up of one grand spectacle but several tiny spectacles, each a monad which is connected to the other through the network.

Shaviro’s concept of people having to make a small spectcle of themselves through their webcams (78-80) so others know their alive matches up with Latour’s talk about the worknet and the individual actor networks.  As Latour explains, the only way for anyone to know and indivudal actor network/actant is alive is by their ability to respond to the mediators which have are making them do something, ie, translate, transform, or produce an artifact with the information or materials given to them throug their connections to the worknet.  The difference here is Shaviro’s emphasis on the digital network and the Internet.  Still, it seems Shaviro is working towards the same point but staying with within the confines of the Internet and the Web.  Everything is a node, everything is information, and everything has some sort of agency in the network.  Also, evey small section that composes the book appears to be a monad that connects to the others through the network that is the book.

Each time we extend ourselves technologically, some part of the real gives way to the virutal.  This is why every cultural innovation is attended by an ambivalent sense of loss.  And this is also why we tend to equate virtual with disembodied, even though it would be accurate to use it as an equivalent for prosthetic.  In a certain sense, then, we ahve always been cyborgs, even if, by the strictest meaning of the term, the transformation only happened recently, when our prostheses became “electromechicanical devices.” (104)

The space of flows is an intersting way to conceptualize the world of virutal space.  In the space of flows nothing remains of traditional concepts of space; the space of flows if “exhilarating, disorientating, or oppresive” (131) but is still different than a traditionally defined room or building working with a spcecific design ethic to exude one sensation or mood.  Most importantly in the space of flows: there is no conswquence of duration, and therefore, no distance, and this makes all communciation instantaneous.  “Proximity is no longer deteremined by geographical location and by  face-to-face meetings, but rather by global flows of money and information,  The predominant form of human interaction in this space is networking…the develpment of eletroinic communictations terchnology frees this sort of networking from its dependency upon specific locales and specific time; now netowkring can flourish on a global scale” (131).

133 and 134 are outlined the three different layers of virtual space.  All are directly taken from Castells, and it’s all depressing as the virtual is dominated by those connected to the market, and conversely, everything considered worthwhile is connected to them through this network space (money, art, entertainment, services).  Also, the “medium is the message” slogan of McLuhan resonates with Latour on 133 through Shaviro’s explanation of it.

Film noir as the bulwark against postmodernist emptiness and inauthenticity on 142-148.  The male hero’s eventual betrayal is a recreation of the Christ story, and for the protagonist, is the onlyway to know he exists is thorugh emulating Christ’s eventually betrayal and death at the hands of the larger social-political system.  Shaviro explains how even the genre of noir is a product of postmodernist thought using Gravity’s Rainbow, and therefore a futile attempt to escape the postmodern world of connectivity and transnational capitalism.

The discussion of zombies as workers (best definition on 168-169) and slake moths as capitalists slurping human creativity, taking it as their’s, and then protecting the right to own the creativity under the banner of “intellectual property” they’ve acquired the rights to (170) is one of the better metaphorical explanations of late stage capitalism ever.  “But the crucial point…is that it is nothing more than a monstrous intensification of the “normal” functioning of the system…they are just capitalism with an (appropriately) inhuman face.  They are literally unthinkable, yet at the same time, they are entirely immanent to the society that they ravage” (170, 171).

Shaviro sees the network as akin to our galaxy–a large system with a black hole in the center of it (173).  All information falls in and disappears.  This would make the network monolithic, something that can not be resisted.  There is no chance at agency.

An interteresting critque of capitalim on pages 222-223.  The internal conflict of overproduction/underconsumption Marxists have continually hoped would force capitalism to break down have been continually used as a moment of crises that leads to some new territory to be found and conquered.  Shaviro hypothesizes capitalism’s goal is to oversatuarate the Earth, and then paraphrasing Ken Macleod, head into outerspace.

The drug metaphors and the crisis of capitalism appear to be leading to this revelation: in moment of crisis the network can be hacked.  There is the potential for change within the network since becoming something is traversing the lines between the nodes, and there is space between the nodes without anything but dark matter.  The overall concept of the book appears to be the network society is the new, postmodern form of capitalism.  When crisis occurs within the market system there are still those with the right skills, right worldview, and the right abilities to make the world as they see fit, a Nietzchian will to power in which the individual manipualtes the network to get what s/he needs.  Like Tron.

Or since these are all only little monads, that ‘s the just the conclusion of a string  of them.  For good chunk of the rest of the book there is a discussion about the “exubernce” of society built on capitalism, that is, the luxurious, wasteful spending and living (wasteful also in the sense of creating tons of waste).  From about 225 to 235 there is a long discussion concerning genes and memes, and for a bit it seems as if he’s trying to argue that memes are the reason for the exuberence of capitalism.  In the concept of waste, or going against a Darwinian version of life, celibacy is seen as a meme that is utterly detrimental.  The meme serves itself and does not care that if it is successful, it will not only run humanity out of existence but itself since there wouldnt be any more vehicles (humans) for the meme to travel, parasite like, and infect new minds.  This talk is eventually dismissed as if it’s silly that everything (meaning nonhumans) within the network could have agency.

This lack of agency seems to extend to humans as well.  The network is capitalism, and everything in the network is merely the coercive forces put into place to make the human nodes within the system consume.  Capitalism is monolithic.

“So this is what it means to live in the network society.  We have moved outof time and into space.  Anything you want is yours for the asking,  You can get it right here and right now.  All you have to do is pay the price” (249).

Toward a Structural Theory of Action: Network Modes of Social Strcuture, Perception, and Action
Ronald S. Burt

Burt is advocating a structural theory of action that will bridge the atomistic and normative action theories. Structural theory appears to be built on several small mathematical models that predict how inidividual actors make decisions in their own best interests within a given context, which can then be tested with empirical data collected using various methods germane to sociology (10).

Atomistic theory:As an individual, or social atom, an actor exists today in reference to his previous conditions and evaluates alternative future actions in reference to his current conditions.  An “atomistic” perspective assumes that alternative acitons are evaluated independently by sepersate actors so that evaluations are made without reference to other actors.  For purposes here, the atomistic perspective is defined by separate actors having exogenously formed interests, one actor’s interests, or preferences, being analytically independdent of another’s…Such a perspective lays the foundation for twentieth-century lilberal democratic theory as it is based on the property concept that Macpherson terms “possessive individualism”…This is the perspective that Smith develops in The Wealth of Nations to elaborate market mechanisms in terms of supply and demand.  (5)

Normative theory:As a member of society, an actor exists within a system of actors and evaluates alternative actions within that context.  A “normative” perspective assumes that actions are evaluated interdependently by separate actors as a function of socializing processes that integrate them withing a series of actors.  For the purposes of this discussion, the normative perspective is defined by separate actors within a system having interdependent interests as social norms generated by actors socializing one another. (5)

Empirical data back normative action theory, and yet this work compiled by experiments, surveys, and ethnographic studies can not be made clear conceptually.  Atomistic action theory works better in the realm of economics, and therefore, these social scientists have established deductive theories which do not match empirical studies coming out of other social sciences.  This causes a schism in the social sciences, and makes for two very different explanations of how actors work in society.  Burt is looking to remedy this cacophony of voices coming out of the social science by providing structural theory; structural theory is “deductively superior to normative action since its use of network models provides a rigorous algebraic representation of system stratification from which hypotheses can be derived,  It is descriptively superior to atomistic action since it explicitly takes into account the social context within which actors make evaluations” (8).

I have chose a deductive approach to theory construction combined with the strategic use of empiricial data.  I alternate conceptual and applied chapters,  in the former, three items are presented:

1. Initial ideas are introduced aspects of a component in Figure 1.1 to be captured.
2. These ideas are formalized as a mathematical model.
3. Some of the model’s empirical implications are then derived as hypotheses, and the model, together with it implications, is illustrated with heuristically hypothetical data. (11)

In short, I am making strategic use of empirical data.  Particularly, relevant data are used to empiricially inform each proposed model as well as to demonstrtate some way in which the model informs ongoing substantive research.  However, the data are in no sense used to justify proposed models as empirical generalizations.  (12)

Chapter 2

Network Structure: The Social Context

The models presented in this chapter are based on social topology.  For Burt, these models describe the materiality of experiential reality, and moreover, are not based on hypotheticals but empirical data.  In this case, Burt is describing the social differentiation among actos within a given system–which is important since he’ll be using these models to analyze and describe the social differentiation among actors in two different social networks.  The models in this chapter–according to Burt–serve as a connection between micro and macro theory “as well as an epistemic link between abstract concepts and empirical research” (19).

Chapter 3

Stratification in Elite Sociological Methodology

In this chapter Burt attempts to demonstrate the ties between those who create an “invisible college,” a “system of scientists tied to one another less b their common instituional affiliation than by their interpersonal relations of advice and collaboration” (95).  Once that’s accomplished, Burt’s purpose is to “describe the stratification within the invisible college of elite experts in sociological methodology” (95).  In a move which illustrates concept and empirical observation, Burt’s models (supplied with the correct data) prove “Mullins interpretation of speciality prominence” (127), a theory suggesting “that the prominence of a group of scientists is a fucntion of the colleague and teacher-student ties among scientists in the group.  Groups are prominent to the extent that the have dense colleague relations and to the extent that “founding fathers” in a group prolifically generate graduate students to continue their work” (127).  Similarity in training and topic of inquiry coupled with direct lines of descendency seem to be how those in the “first tier” of this field gain dominance and (at least in the scale of academia) fame.  Burt finds this troubling since some groups, like the social statistics elite, could maintain the glare of the spotlight based on the continuation of the status quo while “the mathematical sociology elite could continue to pursue myriad different substantive interests, generate new experts within each of these interests, and remain an invisible group: a group better known for its label than for its accomplishments and members” (128).

Burt appears to favor a more meritocratic system within sociology, and yet his findings would also link up with the work of Barabasi; Barabasi at certain points within his monograph Linked explains that the big stay big because they started within a network leveraging the social politics of said network to their advantage.

Chatper 4

Stratification in American Manufacturing

The purpose of this chapter was to describe how the different boards of various manufacturing companies were connected through sharing common members.  This chapter varies from the last one in that it uses data from a small group of actors.  This is do to the size of the American econonmy, the network topology in question.

The sharing of board members served the needs of particular corporations in various ways (most of them obvious if you think of oligarchies and monopolies), and also in the most interesting: the co-optive sense.  Co-optation is described as “the process of absorbing new elements into the leadership or policy-determining structure of an organization as a means of averting threats to its stability or existence” (Selznick qtd in Burt 132).  Burt, early in the chapter, gives the example of the Tennesse Valley Authority.  The TVA avoided potential showdowns with local groups by “appointing representatives of the organizations to positions in the TVA desicion-making structure” (132).  At the end of the chapter, Burt explains the firms most likely to use this strategy were those that were large, controlled by dispersed interest groups, and often unable to dominate their boards with either their own handpicked management or anyone with kinship ties to the CEO (think family firms).

Again, and since I’m horrible with the equations in the book, I’d have to read Burt’s work through the lens of Barabasi.  For Barabasi, nodes in the network interconnect due to efficiency and stability.  While it runs counter to the myth of meritocracy and egalitarianism, it also shows just that–both are a myth and the only thing governing these systems are the ability to thrive and survive in a specific environment.  This, of course, does not mean these nodes are shock proof, just well-adapted to the work of the network at that moment.

Chapter 5

Interest: The Perception of Utility

My purpose in this chapter is to propose a model of the way in which his perception of advantage is contingent on the context in which he makes the perception.  I begin by distingsuihing two aspects of actor interests–subjetive evaluation of concrete stimuli versus social context–and discuss algebraic representations of these aspects.  The two are then brough together in a structural model of perception.  Under the proposed model, acto interests are patterned by the positions of actors in social structure…Moving to  aless general level of abstraction, one with clearer empirical implications, I show how the dervied social norms and deprivation effects can be used to clarify some conceptual ambiguities in diffusion research while simultaneously extending that research to include new substantive results.  (173)

Chapter 6

Conformity and Deviance with Respect to Journal Nomrs in Elite Sociological Methodology

Journals are not judged as superior or inferior by the interactions of elites with said journals nor with the interactions with other elites through said journals; it is the relationship between elites and their arbitrarily defined and shared idea of what constitutes a  “good” journal that makes a specific journal “good.”  If they aren’t socially constructed as “good” they just don’t exist.

This is way to re-affirm the stratification in the invisible college.  The journals read are an indicator of status, or at least the status the reader aspires to.  There is no articulated standard of excellence for a journal, merely the folkloric concept that “everyone” of note reads this journal because it has the “best” articles by the “best” writers.  It all boils down to cultural capital bestowed by a small group of elites.

Chapter 7

Autonomy and Cooptation

The purpose of this chapter is to demonstrate a nodes structural autonomy within a network, “their ability to pursue and realize intyerests without constraint from other actors within the system” (265) while still recognizing there are relational patterns at play that individual actants have to navigate.  The model Burt creates is not designed to “capture all nuances of the ideas commonly referd to as oligolpoly and conflicting group affiliation” (287), and yet Burt does claim that if “a substantive area conforms to the model’s limitations, I believe the model can be a rewarding guide for empirical research” (287).

Chapter 8

Market Constraints and Directorate Ties with Respect to American Manufactoring Industries

In this chapter Burt attempts to illustrate what advantages come from shared directorate ties among corporations.  His conclusion:

Interindustry differences in successful cooptation do not add to the prediction of industry profit margins by market structure alone.  This does not rule out the possibility that directorate ties changed the nature of competitionin the economy, but it does show that the mere strategic placement of such ties did not distinguish unusually profitable manufacturing industries from those in which low profits werre typically obtained.  (324)

There are two possibilities with the information Burt does collect using his model.  First, that this arrangement helps the efficiency of production within the market.  The other, and the one I would ascribe to, predicts that there is a decrease in efficiency due to the a selective flow of information which only reaches favored trading partners “so as to suppress innovation [from upstart rivals] while ensuring markets for overpriced commodities” (325).

Chapter 9

Toward a Structural Theory of Action

I have worked toward a structural theory of action by focusing on its internal features in order to demonstrate that such a theory is plausible at a high level of coneptual rigor while maintaining considerable substantive promise.  Shorn of details, the logical structure of this enterprise has been very simple: stat e the premise, identify the most basic issues contained in the premise, and present empirically acceptable, mathematically simple models providing some resolution to these issues.  (329)

I have only scratched the surface.  Representing status/role-sets and structural interests within the social topology of a system as described in Chapters 2 and 5 provides rather general models within a structural theory of action.  My preceding remarks docus on these models. The structural autonomy models is not at all the same class of generality.  As I stressed in the conclusion to Chapter 7, the model is an arbitrary simplification, a plausible baseline midle for empirical research that describes how relational patterns constain the ability to act.  Not only are there many directions in which this model could be generalized, there is the concept of structural power to be considered, namely, the ability to act despite constraint (as opposed to the structural autonomy, which is the ability to act despite constraint).  Power and autonomy together would underlie transformational changes in social structure as described in the preceding.  (356)

Structural Holes: The Social Structure of Competition
Ronald S. Burt

Introduction

Structural holes is a theory about actants leveraging the information and resources they have that others don’t.  This ability to have info/resources within a network not available to everyone is the evidence of hole within the network; only certain actants have to information that would be commonplace if a node were present to either pass along these resources or make info available for leverage by others to perform mundane tasks specific to the system in question.

There are four characteristics jointly characteristic to the structural hole argument:

1. Competition is a matter of relation, not player attribution.
2. Competition is a realtion emergent, not observed.  The structural holes in which competition develops are invisible relations of nonredundancy.
3. Competition is a process, not just a result.
4. Imperfect competition (the competition of reality, not the bucolic competition of theory) is a matter of freedom, not just power.

The social structure of comeptitiotn is not about the structure of competitive relationships.  It is about the social structure of the relations for which players compete.  The structural hole argument is not a theory of competitve relationships.  It is a theory about competition for the benefits of relationships.  To explain variation in competitive success.  (5)

The distribution of structural holes around around the relations that intersect in a person or an orgnaization determines the player’s enterpreneurial opportunities and thus the player’s comeptitive advantage.  Holes create inequality between organizations as they create inequality between people.  (2)

Chapter 1 The Social Structure of Competition

Burt’s ideas and uses of the concept understood as network are different from Benkler, Shirky, Spinuzzi, Barabasi, or Galloway and Thacker.  While those writer-theorist see the network as a larger interconnected set of nodes getting work done through co-ercion and processing power, Burt defines networks as contacts developed around key, individual players and the manipulation of the network is for the betterment of the solitary entrepreneur looking to profit. Efficiency is gauged by how much energy it takes the entrepreneur to cultivate and sustain useful contacts within a given network and the energy it takes for the entrepreneur to leverage the holes within the network to his advantage.  The ideal relationship for the entrepreneur is the tertius gaudens–the third person between any two other actants who can leverage either goods and information that neither of the other two have to extract the most profit, or the person who can effectively play actant A against actant B and ensure the tertius in the scenario can create competition between A and B that somehow profits the tertius.  The goal is to create competition, leverage products from the structural hole to get the upper-hand in this competition, and at the same time use the contact within the given network which are not being played to speak well of the tertius so as to trade on reputation as trustworthy.  Burt calls this “referrals.”

The substance of information benefits are access, timing, and referrals .  The player’s network provides access to information well beyond what the player could process alone.  The network also provides that information early, which gives the player an advantage in acting on the information.  These benefits concern information coming to the player from contacts.  Referal benefits involve the opposite flow.  The network that filter information received by other abuot the player.  Referrals get the player’s interests represented in a positive light, at the right time, in the right places…The structural holes that generate information benefits also generate control benefits, giving certain players an advantage in negotiating their relationships.  (47)

Burt often uses the second person when describing how all of this works, and makes it clear the entrepreneur in the best position is the one who has no structural holes around him but access to various networks filled with structural holes.  In this state, the entrepreneur has structural autonomy in opposition to lowly structural equality.  Also, Burt is not concerned with tracing out how the network was created or what motivates a person to manipulate structural holes for profit.  He assumes, it seems from my reading, that all of these social practices common to a society based on the market and amenable to capitalism.  What is interesting and new for Burt “is the expression of competitive advantage–in economic, political, or social arenas– in terms of structural holes as an elemental unit clearly defined in theory and readily operationalized for empirical research” (49).

Chapter Two Formalizing the Argument

There are three points in the creation of a structural autonomy model.

1. The relations that span the control benefits of holes are teis of exclusive access.
2. Structural autonomy is a nonlinear function of constraint, decreasing most sharply at low levels  of constraint with the initial loss of structural holes.
3. The boundary around a competitive arena is an issue for players outside the arena.  To obtain the benefits, players outside the arena have to take on a strategic partner established in the arena.  (81)

The potential of a player’s network is summarized in three ways.

1. Effective size is the number of nonredundant contacts in the network.
2. Structural autonomy is an interval scale measuring the extent to which the player, relative to others in a study population, has unconstrained access to structural holes.
3. A hole signature summarizes the distribution of opportunity and constraint across each relationship in the network. (81)

Chapter Three Turning a Profit

The analysis performed in this chapter illustrates ways to demonstrate competition, and also could be used for the more “practical tasks of distinguishing customer segments for marketing strategies, understanding competitive pressures on potential buyers in each segment, and understanding the profit potential of a product” (82).

Three points are established within the chapter:

1. Profit margins are eroded by structural holes among producers and enhanced by structural holes among suppliers and customers.
2. Hole effects are nonlinear and multiplicative in the final structural autonomy model predicting profit margins.  Structural holes have their greatest effect as completely unconstrained action begins to be constrained.
3. The bulk of business for most producers is concentrated in no more than a handful of critical transactions.

Chapter Four Getting Ahead

“Managers with networks rich in structural holes get promoted faster and at a younger age than do their peers…The kind of analysis that follows is useful for the social science task of styudying competition and occupational achievement, as well as the more practical tasks of understanding how specific kinds of individuals rise in the firm, detecting barriers to achievement for kinds of individuals within the firm, assessing the leadership abilities of individuals or groups, and developing programs to enhance leadership abilites in target individuals or groups” (115).

Managers who want to take advantage of the structural holes need to play the role of the tertius and focus on developing networks that stress referrals from higher up in the hierarchical structure and highlight/enable the manager’s ability to leverage various network human and corporate resources to complete projects.

Five points should be taken away from this chapter.

1. Managers with networks rich  in structural holes tend to be promoted faster, and they tend to reach their current rank earlier.
2. Hole effects are most evident for managers operating on a social frontier.  A social frontier is any place where two social world meet, where people of one kind meet people of another kind.
3. The most serious frontier is the political boundary between top leadership and the rest of the firm.  Structural holes affect the early promotions of high-ranking men in a way different from their effect on the early promotions of women and entry-rank men.
4. On the other side of the political frontier, competition has a more personal flavor and it serves the climbing tertius to ensure they have a strategic partnership with “built around a strategic partner other than the immediate supervisor, reinforced with extensive socializing within the immediate work group” (165).
5. Although the reported differences between manager netowrks have clear implications for promotions, there are no differences among managers in their tendencies to have one network rather than another…In other words, a manager’s physical or functional position in the firm is less a cause or consequnece of the manager’s network than it is a context defining the manner in which the network contributes to promotion.(163-165)

Chapter Five Player-Structure Duality

“The unit of analysis in which structural holes have their casual effect is the same at a macro or micro level of analysis.  It is the network of relations that intersect in a player” (192).

“The distribution of structural holes around the relations that intersect in a person or organization determine the player’s entrepreneurial opportunities, and so the player’s competitive advantage.  Structural holes create inequality between organizations as they create inequality between people” (192).

Using this theory of structural holes allows Burt to create a method of analysis that uses the individual as the unit of analysis for understanding how networks engender or constrain power within its confines.  Since the network intersects in a player, there is the ability to analysis both levels (micro and macro) in one fell swoop.  It seems these models are ways to interpret empirical data often found to be unwieldy.

Chapter Seven Commit and Survive

Important to remember:  A “player” for Burt can be any entity.  It’s either a person or a company; hence, the ability for the structural hole theory to cut across micro and macro level analysis.

“The commit hypothesis is that low-autonomy players conform more closely, under threat of being excluded from relationships, to behavior characteristic to their location in the social structure…The commit hypothesis is an occasion to develop a bridge between the strucutral hoel argument and the interface model of markets…The lower the structural autonomy of players in a market, the greater their commitment to the market schedule characteristic of their market” (226).

“The corollary survival hypothesis is that higher rates of change, new players replacing old, occur where there is little structural autonomy precisely because there is little room for error…The survival hypothesis is an occasion to develop a bridge to population ecology analysis…Illustrative evidence on American markets shows that market leaders survive longer as leaders in more autonomous markets and that structural autonomy decreases the mortality of organizations new to the market.  The lower the structural autonomy of a market, the greater the odds of players being forced out of the market” (226-227).

Theories of Communication Networks
Peter R. Monge
Noshir S. Contractor

Preface

The field of network topics has no overarching framework for integrating conceptual, theoretical, and empirical work.  Currently, the field borrows from heavily from the social sciences to develop and and test network hypothesis, more often implicitly than explicitly.  This book is an attempt to provide that framework. (xii)

The impetus for the book is based on a few key problems the authors hope to remedy:

1. Problem: Few network studies utilize theories as the basis for formulating research hypothesis. Those that do only use single theories.  This, in turn, accounts for relatively small amounts of network variance.  Solution: The development of a multitheoretical perspective as a way to help compare and integrate diverse theories and to increase the explanatory power of research efforts.
2. Problem: Most research is conducted at a single level of analysis, typical the individual or dyad and rarely at the entire network level.  Networks are multi-leveled and with components and properties at each level.  To explain this there needs to be information contributions from all levels.  Solution: The framework for network topics developed in this book is multitheoretical as well as multilevel.  By multilevel, the authors mean “all the typical levels within a specific network at both a given point in time and at earlier points in time.”  Also, they include in the framework the other networks to which the focal network may be related, as well as the attributes of people who comprise these networks.
3. Problem: While many researches are exploring challening issues within emerging system properties, this has yet to occur in network research.  Solution: The introduction in this book of complex adaptive systems perspective.  This is done through an agent-based modeling framework.  Starting with a network and tracing out the actors that make up the network, the claim is these agents follow probabilistic rules that may be interconnected or dependent.  The agents observe other agents in the network they are connected to and respond to them.  As the agents follow the rules, network structure emerges.  If the rules and/or the interconnections are changed, then the structures change.
4. Problem: Most network analysis is static and cross-sectional.  Solution: The use of Blanche computer program to study co-evolutionary dynamics.  Using the program allows for the creation dynamic simulations of network evolutions, and then these simulations are used to “generate interesting hypotheses and to analyze research data” (xiii).
5. Problem: The ability to empirically test the ideas and framework presented in the book.  Solution: The use of the p* statistical framework and PSPAR computer programs.  In the space of this book, however, the authors only provide “illustrative examples rather than definitive results” (xiv).

Chapter breakdowns can be found on pages xiv-xv.

Chapter 1 Network and Flows in Organizational Communication

The opening is a rehash of the intro, but under the heading “Communications Networks and Flows in a Global World” is a great overview of the varying theories talking about the effects of the collapse of space and time due to instantaneous communication, the shift to a networked information society (due in part to communications technology), and the ability of actants within this system to create virtual organizations–groups no longer tied to specific locales, times, or events (4-7).

Emergent networks is the designation given to organic, actant formed networks that deal with a specific problem, project, or topic.

Mandated networks are the formal networks often found within corporations or similar workplaces.  Orders travel down these strict hierarchies, while (ideally) information flows up.

Empirical research shows that most problems in work situations are solved by emergent networks.  The social and power stratification of workers and the distance from those considered at the top reduces the incentive to communicate with those outside of the immediate work group–even with those designated as “problem solvers” in the upper echelons of the company.  At most, mid-level employees receive info from those working on the ground.

There has been much debate and research over the formal versus the emergent network, and more talk about a bias within these studies to champion emergent networks as the way things “really” get done since the actions of emergent networks are often at odds with the formal networks they reside in (that is counter to workplace goals).  M&C explain this may all be moot as workplaces are becoming less formal in their organization.  Shifts in management philosophy, team-based forms of organizing, “the adoption of matrix forms of organizational structure, and shifts to network forms of organizing” (10)–all under girded by the growth and adoption of communication technologies that support lateral communication–have dramatically changed the ways firms operate and conceptualize themselves.  Networks within this work setting are now “network organizational forms” (11) because they’re based on an “interactive form” and have interconnections which “span accross the entire organization, unimpeded by preordained formal strcutres and fluid enough to adapt to immediate technological demands.  These relations can be multiple and complex. But one characteristic the share is that they emerge in the organization, they are not preplanned” (Krackhardt qtd in Monge and Contractor 11).

Break down of chapter on pages 20-25.

Chapter 2 The Multitheoretical, Multilevel Framework

Most of the chapter is limited to the definitions of terms germane to network analysis.  The following are quotes from the chapter with the heading they appeared with leading off the quote so as to give context to the quote.

Network Concepts and Measures

“The relations, such as “communicates with” or”provides data to” are represented as lines connecting the various nodes” (35).

“The second property is strength, which indicates the quantity of the relation…Alternatively, it could represent the frequency with which they communicate, for example,, once a month, once a week, daily, or their satisfaction with the communication on a numerical scale” (35).

“When relations are studied one at a time, the are called uniplex.  Two or more relations studied together are considered multiplex” (35).

Measuring Network Properties

Wasserman and Faust suggest there are five distinct levels.  The individual actor level is the level of the participants represented by the nodes or points in the network, whehter indvidiuals, groups, or organizations…The dyad level examines pairs of network members together with their relations…The triad level examines three nodes at a time, focusing perhaps on the level of balance among all triads in the network.  The fourth level is the subgroup…The [fifth and] final global level is the network as a whole.  (37)

Individual Level of Analysis

Degree, Indegree, and Outdegree

The number of nondirectional ties associated with a node is simply called degree…In a directional communication network, a node’s outdegree coudlbe interpreted as “expansiveness” while the nodes indegree would signal its popularity…Nodes that have a degree of zero arereferred to as isolates; that is, the have no ties to others in the network. (38)

Betweeness

While degree metrics gauge the extent to which a node is directly connected to all other nodes in the network, betewwnness measures the extent to which a node is directly connected only to those other nodes that are not directyly connected to each other…In a communication network, a node with a high betweenness is often interpreted as deriving power by interpretation of that information. (38)

Closeness

[C]loseness measures the extent to which nodes are directly or indirectly connected to allother nodes in the network…Closeness is thereofre interpreted as a useful measure to assess a node’s ability to efficiently access information directly or indirectly “through the grapevine.” (39)

Dyadic (or Link or Tie) Level of Analysis

For valued relations, mutuality measures the similarity between the values of the links between two individuals.

Distance and Geodesics

For any pair of nodes, two types of links can exist: direct and indirect.  Direct links are connections between any pair of nodes that involve only those two nodes.  Indirect links occur between any two nodes by virtue of their connections with other nodes.  A direct link between two nodes is said to be a one-step connection.  The smallest indirect connection is two-step, which ties together three nodes with two direct links.  Here the first node is directly connected with the second node, the second is directly connected with the third, which leads the first and third nodes to be indirectly connected to each other with a two-step linkage or two degrees of separation. (41)

The shortest distance between two points is called a geodesic.  The largest distance is called the diameter…First, reachability is the shortest path (or the geodesic) that connects two individuals in a network…Second, redundancy measures the number of alternative shortest paths (or geodesics) that connect two individuals indirectly.  A high redundancy score would indicate a greater likelihood that information will flow from one individual to another via one of the multiple indirect paths.

Triadic Level of Analysis

Transitivity and cyclicality measure the extent to which every set of three actors, say, A, B, and C, in the network demonstrates certain structural patterns…A network is cyclical when A Directs a tie to B, B ties to C, and C in turn links to A, thereby completing the cycle.  (42)

Subgroup Level of Analysis

Components and Cliques

More [than] likely, the graph [AKA network] is unconnected or disconnected, meaning that is it not possible to get to all points in the graph from the other points.  This implies that there are subsets of points in the network that are connected to one another, call ed subgraphs; it also implies that the subgraphs are not connected to each other.  These conneted subgraphs of the network are called components of the network. (43)

[A] clique is defined as a maximally complete subgraph, that is, the maximum number of individuals in the network who are all directly connected to one another, but are not all directly connected to any additional individuals in the network…An n-clique includes the maximum number of individuals in the network who are all directly or indirectly connected to one another via no more than n links.  Further, the are not directly (or indirectly) connected via n or fewer links to any other additional individual network. (43)

While n-cliques relax the requirement of a direct link to all members to the clique, k-plex relaxes the requirement of a direct link to all members in the network.  A k-plex therefore includes the maximum number of individuals in the network who are directly connected with, at least, all but k of the individuals in the group.  (43)

All the different models in which the various theories can be used as the generative mechanism are be found 55-69.

Chapter Three Communication and Knowledge Networks as Complex Systems

There’s a general overview of various system models on pages 79-85.  Things get interesting on 89 with self-organizing complex systems.  These systems have four features:

1. At least one of the copnents in the system must exhibit autocatalsis, that is, self-referencing.
2. At least two of the comopnents in the system must be mutually causal.
3. The system must be open to the environment with respect to the exchange of energy and matter.
4. The system must operate in a far-from-equillibrium condition.  (89)

Essentially, these systems have to be self-organizing without intervention from a bureaucratic office nor the leadership of a charismatic, central leader.  The example given in the text the self-organized system call slugs (89-90), a carpooling system which organically sprang up among DC commuters.

Knowledge networks as complex systems might be useful for future projects.  This model envisions knowledge networks as interconnected computers, and counts human and nonhuman actants as part of the network.  This work seems to echo the ideas of Latour and Syverson.

Even more interesting: knowledge networks as complex self-organizing systems.

In knowledge networks that are formed on the basis of social networks for generating and sharing knowledge, there are two types of components in the system that exhibit autocatalysis, the knowledge per se and the people.  As discussed before, human cognizance plays a key role in the creation of new knowledge.  Through complicated cognitive processsess, new knowledge can be generated by the accumulation of new information and intergration with preexisting knowledge.  The social netowrk is self-generative in that the charisma or reputation of a person can by itself serve as a strong magnet to attract more particiapants to the network.  (95-96).

The chapter ends with Contractor calling for “system thinkers [moving] beyond a metaphorical fascination for definitions, conceptualizations, and collecting analogies.  In addition, schoalrs need to think about what new insights would be gained if the theoretical mechanisms of self-organization were to be used to study organizations” (97).

Chapter Four Computational Modeling of Networks

Computer simulations have often been used to predict the performance of systems with dynamic relatiionship among various elements of said system.  The problem, M&C point out, is “[t]heses characteristics are typically obtained from theory and then articualted in the simulation as difference or differential equations. The goal of engineering simulation is then to assess the dynamic performance of a system based on a priori knowledge of the dynamic relationships among the various elements of the system…while this approach has produced a considerable number of studies…many the results of these models have been criticized for specifying relationships that were at best untested and at worst flawed” (99).

An emerging trend is the use of computers to “augment and assist theory building” (100).  Computational organizational theory (COT) follows this research process path: Theory–>Formulate Logics of Emergence–>Run Dynamic Simulations–>Deduce Hypothesis from Simulation Data–>Empirical Validation. This is in contrast to traditional research process, which looks more like this: theory–>Verbally Deduce Hypothesis–>Empirical Validation (101).  C&M stress “the results of a computer simulation are not a surrogate for empirical data.  Rather, they help to identify the emergent process implied by the theory” (100).

One simulation program C&M discuss at length is Blanche.  Blanche attempts to take into account the evolution of an agent’s attributes and relations over time through the use of nonlinear differential equations.  Blanche is publically available and can be found at http://www.spcomm.uiuc.edu/Projects/TECLAB/BLANCHE/.  The ethical and responsible use of such program is linked to two other projects:

1. A research design to collect the empirical data of the phenomena being examined, and
2. The use of appropriate statistical techniques to specify and test the likelihood of obtaining the observed realization of the network from among the set of possible configurations.  (109)

M&C recommend stochastic modeling versus determinstic modeling (as the name suggests, determinstic isn’t very realistic).  Stochastic modeling seeks “to determine the probability rather than the the actual value of an agent’s attribute or the realtion between two agents…As one would exect, repeatedly executing sochastic computational models will not necessarily yield the same emergent outcomes. Instead, the outcomes will vary and can be summarily viewed as a distribution with some emergent outcomes more likely to occur than others” (110).

There is a difference between modeling for social phenomena and modeling for a network.  There are special issues involved in modeling for a network; according to M&C these “constraints arrive fromthe recognition that relations within a network are not independent from one another” (112).

Strategies of Empirical Validation of Computational Models (118).

Computational models can be validated through the comparing of empirical data from a longitudinal study on a network.  This is the most direct method, but also the most expensive and time consuming.  The work around is to create “virtual” experiments, where a network is created complete with virtual agents.  These studies are used to “examine the transient and long-term emergent charateristics of a set of theoretical propositions” (119).  Using these virtual experiments allows for the modification of hypotheses (which would include the setting bench marks for the degree of intensity certain theories work withing a network), as well as come to an understanding how various initial conditions could effect the emergent characteristics of the real network in question.  Changing the initial  conditions of the virtual network and re-running the experiment also allows for the creation of different hypotheses based on different initial conditions (again, this is all virtual.  A responsible use of this technique would be to develop the hypotheses and then apply to the version of the material version of the network in question).

Here are three other ways virtual experiments could be useful:

1. They can also be used to determine how emergent characteristics are altered by parameter changes, that is, the relative influence of the variables on one another.  The parameters that are used for modeling virtual experiments may be based on prior field and experimental empirical studies.
2. They can also be used to study transient and long-term influence of interventions (such as the introduction of a new technology) on an ongoing social system.
3. Virtual experiments are excellent approach to explore the conditions under which a system will transition from a state of chaos to a state of self-organization, or move from a stable equilibrium state or a self-organized far-from-equilibrium state into a state of chaos.

It is important that nay inferences made by the researcher derive entirely from the rules or generative mechanisms in a computational model.  These rules reflect the tenets of the specific theory.  To the extent that the rules implemented in the model fail to reflect the specific theory, the inferences drawn by the researcher are suspect.  (121)

Results from the field studies can be used to specify effect sizes that are used as parameters or weighting coefficients in the computational model.  Further, field studies can provide the initial data that can then be used in the computational model to consider various possible outcomes.  (122)

Chapter Five Theories of Self-Interest and Collective Action

The overall goal of this chapter is to define what the generative mechanism, or the prime mover, in theories of self-interest and collective action is.  For the large category “theories of self-interest,” the three theories under scrutiny are social capital, transaction cost economics, and network organization.

The generative mechanism in social capital is structural autonomy, the ability to exist within a network, be an active entrepreneur in said network, and have the ability to profit through unconstrained, strategic moves.  This generative mechanism relies heavily on Burt’s structural holes theory.

Transaction cost economics (TCE) attempts to examine how firms (traditional organizations using various levels of bureaucracy) profit within a market.  TCE asserts a firm’s ability to organize information and communication so as to find the best buys within a market offset the coordination costs associated with such a centralized organization like a corporation.  Exchange and reciprocity are the elements which make up the generative mechanisms in this theory.

An alternative form of organization, the network organization, can reduce both information search costs in markets and administrative costs of hierarchies.  Network organizations seek to maximize joint value of exchanges with the organizations to which they are linked.  Network organizations are themselves embedded in larger networks of organizational relations that make economic behavior neither over- nor undersocialized.  (159)

The collective action theories described in the second half of the chapter are the collective action and mobilization theory and the collective action and the adoption of innovations theory.  Both rely on the mechanism of the public good, that there is some public good which must be contributed to and developed so whatever the public good is, it survives through management and replenishment.  M & C stress these network work best in a centralized communication network, one where all actants are connected, informed, modeling, and communicating with one another through the center or through mutually strong ties to all nodes in the network.  M & C hypothesize this stops the abuse of the public good as well as stops the problem of “free riding,” where actants take from the public good but do nothing to maintain it.  Coupled with their emphasis on high/low reach and high/low selectivity, M & Cs ideas about collective action appears to be what Shirky is arguing against–it’s the old model of organizing.

Chapter Six Contagion, Semantic, and Cognitive Theories

Contagion theories seek to explain networks as conduits for “infectious” attitudes and behavior.  Semantic theories attempt explanations on the basis of networks that map similarities among individuals’ interpretations.  Theories of cognitive social structures examine cognitions regarding “who knows who” and “who know who knows who,” while theories of cognitive knowledge structures examines cognitions of “who knows what” and “who knows who knows what.”  Finally, cognitive consistency theories explain how metworks are understood on the basis of individuals’ cognitions of consistency or balance in their networks.  The remainder of this chapter discusses each of these areas and their extensions.  (173)

Chapter Seven Exchange and Dependency Theories

This chapter opens with a talk on network exchange theory, which is built on a calculus of exchange of material or information resources as the under-girding constant of all human interaction.  “Network exchange theory posits that the bargaining power of individuals is a function of the extent wot which they are vulnerable to exclusion from communication and other exchanges within the network” (209).  This theory is not the same as theories of self-interest as self-interest theories conceptualize the individual as maximizing their individual investments independent of its exchange value; in NET this is a major concern and is part of the equation when deciding with whom to ally with.  Due to this, there appears to be a big emphasis on separate entities (firms, corporations, individual actors) within a network and the executive ties between firms.  This concern is mollified by a newer concept that fall underneath the umbrella of exchange and dependency theories: network organizations.

Rather than being organized around market or hierarchical principles, network organizations are created out of complex webs of exchange and dependency relations among multiple organizations…network organizations differ from their predecessors (functional, multidivisional, and matrix forms) in four important ways.  First, rather than subsume all aspects of production within a single hierarchical organization they attempt to create a set of relations and communication networks among several firms, each of which contributers to the value of the product or service.  Second, networks are based on a combination of market mechanisms and informal communication relations…Third, members of networks are often assumed to take a proactive role in improving the final product or service, rather than merely fullfilling contractual obligations.  Finally, a number of industries are beginning to form network organizations along the lines of Japanese keiretsu, which links together producers, suppliers, and financial institutions into fairly stable patterns of relations. (219)

There are six essential qualities to network organizations:

1. The use of information technology to integrate across organizational functions.
2. Flexible, modular organizational structures which can be readily reconfigured as new projects, demands, or problems arise.
3. Use of information technology to coordinate geographically dispersed units and members.
4. Team-based work organization, which emphasizes autonomy and self-management.
5. Relatively flat hierarchies and reliance on horizontal coordination among units and personnel.
6. Use of intra-and inter-organizational markets to mediate transactions such as the assignment and hiring of personnel for projects and the formation of interorganizational networks. (Poole qtd in Monge and Contractor 220)

Through these qualities networked organizations become “boundary-less” (220)–a space where the point that one actant begins and another ends is difficult to tell.  Since several things are being shared (information, goals, resources, personnel, and finances through communication technology), so highly collaborative work arrangements must be established because it’s the only way to “transfer embedded knowledge” (220) integral to the project’s success.  According to Poole, there’s a high cost to this non-traditional set-up; there are problems “maintaining a sense of mission, committment, loyalty, and trust, and dealing with increased levels of work stress and burnout” (qtd in Monge and Contractor 221).

The rest of the book is more in-depth talk about the theories which can be used as generative models.  In chapter ten there’s a fairly complex summary (298-306).   The overall gist of the book is to use the models in chapter two in conjunction with the theories spoken about at length throughout the rest as a way to provide structure for different data sets–I think. Coming from the humanities this all seemed exceptionally convoluted.

Small Worlds: The Dynamics of Networks between Order and Randomness
Duncan J. Watts

“The small-world phenomenon formalises the anecdotal notion that ‘you are only ever six ‘degrees’ of separation’ away from anybody else on the planet'” (4).

Based on Miligram’s mail experiment, tons of work has been done considering the small world phenomenom. Most of the theoretical and empirical work has endeavored to determine for social groupings:

• The characteristic number of “handshakes” between members,
• The expected number of “friends” that each member has, and
• The structure of the group which relates one member’s “circle of friends” to those of other members. (4)

While Watts feel the work done has been interesting, he also feels it’s “suffered from a number of methodological and phenomenological difficulties” (5).  Namely:

1. Detailed data of “who knows whom” is extremely hard to come by for sufficiently large groups.
2. People are notoriously bad at estimating the number of “friends” the possess.
3. Some friendships are more important than others, and some people are vastly more significant than others in connecting a network (for example, Kevin Bacon owes much of his eminent connectbility to his appearances with superstars like Jack Nicholson and Rober DeNiro).
4. Friendships are not symmetric: that is, subordinates are more likely to regard themselves as connected to their superiors than vice versa.
5. The whole notion of “friendship is highly dependent upon both the social cnctext (Amish famers in rural Pennsylvania probably have quite different views of what a friendhsip necessitates than do Hollywood stars) and the nature of the question being asked (that is, friendship for the purpose of borrowing money is quite different from friendship for the purpose of spreading rumors). (5)

This may seem like nitpicking, but the issue becomes one of validity and reliabilty when work based on the Miligram study can be attacked on these grounds.  Studies coming from this linneage appear to “hang on arbitrary assumptions and so does not necessarily indicate much about the world in general” (5).

Chapter two is a brief review of what is understood about small world phenomenon, and is an attempt to answer Watts reformulated set of questions concerning small-world theory, ie, “Do we actually live in a small world?  What are the most general conditions under which the world can be small?”.  This is done through considering a broad range of netowrk structures and identifying where, if anywhere, in the family of possible “worlds” dramatic changes in global network characteristics occur.

Chapter three is a space where these questions are approached through the indtroduction of two classes of graph-theoretical models: relational graphs and spatial graphs.  “Despite apparent differences, both in motivation and construction , these two models exhibit an underlying structural simlarity that can be captured by the idea of random rewiring and that allows their statistical properties to be expressed in a model-dependent fashion.  This motivates the construction of a third model, which embodies the random rewiring concept explicitly and which unifies the properties of the alpha and beta models.  The main result is the identification of a class of graphs–small-world graphs– that appear to embody the defining characteristics of the small-world phenomenom” (6).

Chapter four works off of the intuitions gained from these simulations.  A heuristic is constructed that “yields analytic approximations for the relevant statistics of both spatial and relational graphs” (6).

Chapter five deals with length and clustering properties of three real world networks.

Part II asks “why should anyone care about the small world phenomenon?”.  Chapter six deals with a simple model of the spread of disease through a structured population.  Chapter seven examines the impact that random rewiring has upon the global computational capacity of a cellular auomata.  Chapter eight is a tentative exploration of the emergence and evolution of cooperation on graphs, using the iterated, multiplayer Prisoner’s Dilemma.  Chapter nine broadens the scope to include continuous dynamical systems, examining the conditions under which systems of coupled phase oscillators can spontaneously lock into macroscopic, mutually entrained clusters.

The interest in this book–contrary to many socially texts worried about entrepreneurs, promotions, effective marketing schemes, etc–is “how systems behave and how that behavior is affected by their connectivity” (7).

Chapter Two An Overview of the Small-World Phenomenon

The two important things from this chapter:

1. Weak ties, as defined by Granovetter, are important to the small-world phenomenon since they form the bridges between two densely knit clumps of friends, “these clumps would not, in fact, be connected to one another at all were it not for the existence of weak ties” (Granovetter qtd in Watts 15).
2. The creation of visual graphs to augment a network analysis  “enables the observer to gain more insight into the relationship between members than would be possible by staring at a large matrix of numbers (17).

[I]n terms of social networks, the only networks who statistical properties are analytically tractable are those that are either (1) completely ordered (for instance, a d-dimensional, hyper-cubic lattice) or (2) completely random (such as Rapoport’s random webs).  (20)

Because the methodological basis of measuring distances in the network sense, solely in terms of who is connected to whom, rests on much firmer ground, both theoretically and empirically, network distance wil be treated here as the sole measure of distance, at which point all talk of either non-Euclidean or nonmetric spaces instanly disappears.  A network does not necessarily exist in any particular space at all, but as all network distances must certainly conform to the triangle inequality, then…an embedding is guaranteed by an algorithm that is described briefly in Section 2.2.3.  (22)

Chapter Three Big Worlds and Small Worlds: Models of Graphs

Graphs of the networks in question are appropriate because they display “the nature of the elements of the ‘system’ is unimportant–all that matters is the fashion in which they are connected” (41).

There are two possible graphs possible for this project 1) relational and 2) spatial.  Relational graphs readily display small-world phenomenon, while it takes spatial graphs with more “exotic” (42) distributions to display small-world features.  Relational graphs, therefore, are the better choice.

Three successive graph models are presented in this chapter.  The third model, the phi model (the first two being alpha and beta), is:

motivated by a desire to unify the observed properties of a the [alpha] and [beta] models, in terms of a model independent parameter [phi], as a function of whicl all such models display the same characteristic transitions.  This result leads to a better understandingt of the small world-phenomenon through the introduction of a class of small-world graphs: highly clustered graphs with small characteristic path lengths.  (42)

Social space is the space where these graphs are considered to be occuring; the axioms needed to justify this concept are found on page 43 and are number one and two.

The two extremes in social space are Caveman world and Solaria world.  In Caveman, everyone “you know knows everyone else you know and no one else” (44).  In Solaria world “the influence of current friendships over new friendships to be so slight as to be indistinguishable from random chance” (44).

Chapter Five “It’s a Small World After All”: Three World Graphs

In a small world, everyone seems to be the center, because everyone is close to everyone else.  Kevin Bacon isn’t the center of the Hollywood universe, he just appears to be in a network which can be traced out with some certainty and is somewhat manageable when it comes to graphing it.  After comparing the Kevin Bacon network to the networks of the C. elegans nervous system and the Western States Power Grid, Watts exclaims “[S]omething interesting does appear to be going on. There does appear to be a common thread linking these systems together in terms of the qualitative arrangements of their connections, and regardless of it consequences, that is a remarkable thing in itself” (162, emphasis original).  The next part of the text should answer the question, “So what?”

Chapter Six The Spread of Infectious Disease in Structured Populations

The message here is that the highly clustered nature of small-world graphs can lead to the intuition that a given disease is “far away”, when ,on the contrary, it is very close.  The fact that so few short shortcuts may be required to achieve this small-world effect is important, because such tiny alterations to the network configuration could be impossible to detect from the perspective of an individual.  (175)

Connecting typologies of given networks are the key to linking.  Two things factors come into play

a) The nature of the attractor is determined by the coupling topology, or b) The time taken for the systems with different coupling topologies to reach the same attractor (characteristic transient time) is determined by the coupling topology.  Specifically, spreading occurs faster in systems with shorter characteristic path length. just as one might expect intuitively.   (180)

Chapter Seven Global Computation in Cellular Automata

Cellular automata (CA) are the descendants of John von Neumann’s self-reproducing auotmaton.  These are one dimensional cells which work with  a general algorithm (GA) to locally process info so as to complete a project global in nature (global here meaning within the confines of the system they exist in).  Often these tasks would be trivial if there existed a central processor within the CA, but there isn’t.  The GA forces them to evolve and create solutions as they evolve.  The “firing squad synchronisation [sic] problem” is an example and can be found on page 184.

Starting with CA, the chapter evolves into discussing how to solve global computational problems with locally connected systems by manipulating the architecture of the system rather than its rule base.  Watts concludes deciding the point may not be to affect the system using GAs or with systems that are only one dimensional.

But perhaps this is the point: natural computational systems are not strictly one-dimensional architectures.  In fact, Chapter 5 suggests that the  connectivity of many natural networks is better represented by small world graphs than by many other plausible models, including one-dimensional architectures.  Hence elegant and attractive though the one-dimensional GA/particle approach may be, the broader project of understanding the computational capabilities of real, locally connected computational systems should probably account for the dramatic effects of small world architectural topologies in which the traditional two-dimensional space-time diagrams cease to be meaningful.  Perhaps the most fruitful approach to solving these kinds of problems might be the application of genetic algorithms to the combined space of all possible rules and all possible connectivities.  (198)

Chapter Eight Cooperation in a Small World: Games on Graphs

Cooperation doesn’t do too well in random graphs since they’re “poorly clustered” (222).    Since cooperation in Prisoner’s Dilemma and Tit-for-Tat require trust and cooperation with a small clique against a hostile world, once a few defectors can use shortcuts to others without worrying about intermediaries, the project crumbles.  In small world graphs, however, cooperation seems to organically grow as a strategy–as long as the small world graph is not random.