Networks, Societies, Spheres: Reflections of an Actor-network theorist

Bruno Latour

Reported by Drew Margolin & Anna Li

Bruno Latour, born in 1947 in Beaune, Burgundy, from a wine grower family, was trained first as a philosopher and then an anthropologist. From 1982 to 2006, he has been professor at the Centre de sociologie de l’Innovation at the Ecole nationale supérieure des mines in Paris and, for various periods, visiting professor at UCSD, at the London School of Economics and in the history of science department of Harvard University. He is now professor at Sciences Po Paris where he is also the vice-president for research of that school.

Professor Latour’s lecture combines a discussion of the core themes in Actor Network Theory with insights regarding the enormous quantities of data that are now being produced and made available to researchers.  Using a broad array of examples, including Isaac Newton, the Space Shuttle Disaster, and a comparison of Marcel Proust’s childhood to the world faced by youth today, Latour explains and elaborates on the idea notion networks should replace objects as our locus of attention.  In particular, Latour recalls the insights of Gabriel Tarde and his criticism of the idea that there exists “a society” which is an object separate from individuals.

Latour argues that the notion of society was invented to compensate for the lack of data available to researchers in earlier eras.  Social theory, he suggests, is a function of the “datascape” — what we can record about behavior.  Today we have access to enormous amounts of data, and it is a mistake to try to fit our treatments of these data into traditional constructs such as “the individual” and “society.”  Instead of trying to understand individuals, who are irreducibly complex, we should focus attention on the networks through which they distribute action.  Unlike the notion of society, these networks are simplifications rather than aggregations.

Latour ends the lecture by pointing to two challenges that face researchers.  First, he argues that we must confront the technical and theoretical challenge posed by the new mass of data.  Second, he cites Walter Lippman and his concern with controversy and the fragility of public discourse.  Latour addresses these remarks in particular to the climate change controversy and the role that scholars could play in re-inventing the newspaper.

The Dark Side of Metcalfe’s Law: Multiple and Growing Costs of Network Exclusion

Ernest J. Wilson III & Rahul Tongia

Reported by: Drew Margolin & Cuihua Shen

Ernest James Wilson III, Ph.D., is Walter Annenberg Chair in Communication and dean of the Annenberg School for Communication and Journalism at the University of Southern California. He is also a professor of political science, a faculty fellow at the USC Center on Public Diplomacy at the Annenberg School and an adjunct fellow at the Pacific Council on International Policy. He was elected the first African-American chairman of the Corporation of Public Broadcasting in September 2009.

Rahul Tongia is a Program Director at the Center for Study of Science, Technology and Policy (CSTEP) in Bangalore. His research focuses on infrastructure and technology for sustainable development, especially for underserved regions such as India or Africa. His current work covers the broad areas of Digital Divide, ICT for Sustainable Development, Smart Metering for Electricity Networks, and Energy for Developing Regions. He has a doctorate in Engineering and Public Policy from Carnegie Mellon, and an Sc.B. in Electrical Engineering from Brown University.

Dean Wilson introduces this work as an attempt to focus on the “dark side” of networks.  Much research has emphasized the collective benefits of network size to the individuals who are included in a network.  There is little research on the consequences to those who are excluded from the network.  Wilson and Professor Tongia address this topic with the following questions:

• Why is “exclusion” excluded from so much work on technology and networks?
• Who benefits from inclusion and exclusion?
• What are the distribution and utility of network membership? In this work we propose that inclusion should be paired with the issue of exclusion.

The discussion centers on the digital divide and related policy concerns, but it is informed by work in China, India and Africa where inclusion and exclusion from technology and institutions can be applied more broadly.

Tongia reviews several approaches to valuing networks based on their size.  Laws proposed by Sarnoff, Odlyzko, Metcalfe, Nivi, and Reed each posit that as networks grow, the utility to those in network also grows.  These laws do not specify, however, the consequences in utility to those who are excluded.  More broadly, there is a question of whether to value exclusion with an “inclusion framing,” wherein the cost of exclusion is the the opportunity cost of foregone inclusion, or an “exclusion framing” in which those that are excluded bear additional cost as their numbers fall.  For example, being the “last person on earth” that is not in the network may bring unique and substantial disadvantages beyond the foregone benefits of inclusion.

Tongia proposes several ways that such exclusion costs might be calculated depending on the framing.  Of particular interest is the case where growth in value is exponential, as in this case the excluded (or even those who join late) may be permanently prevented from capturing equal value.  He points out that in the early phases of network formation and growth, the inclusion benefits must be substantial to get people to join.  Conversely, when the being part of the network is the norm, it may become “a necessity,” imposing additional costs on those that still remain outside.

What are the costs to society as a whole?  In many areas societies must keep parallel/dual networks. For digital and analog television; a health insurance system and an emergency room system; a cell phone network and a pay phone system.  An exclusion framing of network utility would suggest that as the concentration of usage shifts from 50-50 to 85-15 the dis-utility to those excluded from the dominant system may outweigh the benefits to those who are included.  The public phones fall in disrepair as cell phones become more popular.  In some cases, the included also bear the cost for the excluded such as when the unpaid emergency room bills for the uninsured force hospitals to pass costs along to the insured.

Dean Wilson concludes the talk asking why there is so little scholarly attention to the excluded.  Is it because there are too few of them?  Perhaps it is because it is difficult to gather data on the excluded.  Or perhaps it is because of homophily — the included are doing the analysis and assume everyone is like them.

Discussion

The discussion featured comments on a variety of issues, including the theoretical and methodological approaches in the paper as well as the larger social implications of Wilson and Tongia’s findings.

Carter Butts and Michael Macy suggest ways the model might be more clearly bounded and described.  Butts argues that the notion of value in network inclusion models are from the operator’s point of view, whereas the paper asks about the value to the user.  He also suggests that the function is polynomic, rather than exponential.  Michael Macy suggests this model captures average value to the user rather than marginal value and is concerned with usual, rather than extreme cases.

Jonathan Taplin asks how this argument can be used in a policy debate regarding.  He asks, for example, whether public money should be used to subsidize broadband access for the poor?  Tongia points out that the argument applies here but also in any scenario where there are two infrastructures, what he refers to as “dual networks.”  David Grewal points out that this dual networks problem has been in political economy for some time, for example, Adam Smith argued that for a boy to go shoeless in Scotland there was no shame, but to go shoeless in London there was shame.

Fritjof Capra argues that the most harm brought by network exclusion is that alternative technologies are being eliminated. e.g., no more tellers, no more payphones.  There is s simple solution — companies should support “lifelines” of basic services.  A small tax on cell phone usage could pay for public phones.  Woody Powell suggests that the paper shows how far we’ve moved, in terms of politics and policy, away from suggestions like Capra’s .  He reminds the group that during the Carter administration the FTC seriously considered giving out “cable TV stamps” to be sure everyone had access.  Now this idea is laughable.

Several participants considered the application of this work to network research.  Nosh Contractor wonders whether the assumption that all individuals are completely connected to the network makes sense in all cases.  What if individuals in a telephone network don’t have the social capital to knwo whom to call?  How can the notion of exclusion be extended to an incompletely connected network?  Peter Monge points out that fully connected networks are extremely rare.  For example, in the densification study by Kleinberg they showed that in most networks the exponent of growth was very low — 1.45 or so in co-citation networks and 1.1 in e-mail networks.  Given these empirical findings, the fully connected network (as implied in the study) is not a very good generalization

Some scholars raise questions about the generalizability of this approach.  Macy argues that the effects for telecom networks may not generalize to other kinds of networks, such as terrorist networks.  Janet Fulk provides an example of a case where those included in the network experience costs because of those who refuse to join the network.  She describes the findings of a study she did of law enforcement agencies trying to pool data about their operations. The failure of the network to recruit key municipalities undermined the strength of the network as a whole. Butt suggests there might be a competency trap here. Those excluded may be doing very well. For example, the early elites cannot use email and the switching costs are high. So they have their assistants print out their e-mails for them.  People who are the last to adopt may be doing very well and they may well be the elites rather than the poor.