Gad, It looks like your view differs a bit from Steve's. Whereas you seem to warn against an artifact of data collection, a flaw in how the network is generated, Steve seems to be calling attention to a social process (which need not involve Amazon) thath amplifies small differences into mutually exclusive factions. I find Steve's observation interesting because I've been fooling around with Jose Luis Molina's NetMirror concept in which (in the context of a consulting engagement) you repeatedly feed back to a group a picture of their own network structure. Jose Luis and I have largely assumed (but will be testing) that the effect of this will be to create a less differentiated, more integrated structure as peripheral individuals scramble to achieve less exposed positions. But Steve's observation suggests that NetMirror could actually create more differentiation and factionalization. Clearly, what's needed is a theory of when it goes one way or the other. Steve gives a path-dependent kind of explanation where any tendency toward factions at the beginning tends to amplify, but would that imply that feedback always leads to factions, since every group probably contains such proto-splits?
steve (b)
----- Original Message -----
From: [log in to unmask] href="mailto:[log in to unmask]">Gad Yair
To: [log in to unmask] href="mailto:[log in to unmask]">[log in to unmask]
Sent: Thursday, March 06, 2003 1:43 AM
Subject: Re: A window in the American souls -- politics and book co-purchasing patterns

After showing my graduate students how neat network analysis is – and admitting that the idea to do this analysis is indeed bright – I, like Steve Corman, feel that it is important to keep in mind the way data are generated.  While network models try to show that people are not independent (and in that sense, there is a family resemblance to HLM models), in the case that Valdis beautifully shows there IS the problem of artificial generation of "buddies", and therefore the mechanical generation of data-dependence.  In essence, buyers' de-facto tastes were generated by the computerized persuasion system, and we cannot assume that these tastes are independent of the machine-generated-buddy-list. Steve is accurate in suggesting that Amazon at least partly manufactures these data, capitalizing on the first purchasers' tastes – so we might be seeing the logic of capitalistic amplifying machiner*ies more than individuals' independent tastes or political views.  This again tells us the 'be-cautious' tale that measurement and theory are linked in important ways, and that a theory about social behavior should "talk" with the appropriate theory of data generation.


Despite this minor critic of a beautifully undertaken project of a bright idea – Valdis's project can teach the 'manufacturers of consent' how to tinker with their 'buddying machine' so as to increase sales.  After all, network analysis can become a powerful strategic tool in a powerful mode of production.


Either way – this neat study does tell a story of an American Soul…


Bravo for the ingenuity, Valdis!


Gad Yair

Department of Sociology & Anthropology

School of Education

Rothberg School of Foreign Students

Hebrew University of Jerusalem

Jerusalem 91905



[log in to unmask]

voice: 972-25883333

fax: 972-25324339

cell: 051669391


-----Original Message-----
From: Social Networks Discussion Forum [mailto:[log in to unmask]] On Behalf Of Steven Corman
Sent: Wednesday, March 05, 2003 4:34 PM
To: [log in to unmask]
Subject: Re: A window in the American souls -- politics and book co-purcha sing patterns


Wow, that's pretty interesting.  Validis's writeup explains this structure as a reflection of political tendencies in the readers.  However, I can think of at least one other generative mechanism.  If Amazon's "buddy book" marketing is effective (and I'm pretty sure it is, because it regularly "gets" me), then the system tends to reproduce its own links.  If there was any polarization in preferences when the system was launched, then new buyers would have been steered toward these same choices (and given limited budgets, therefore away from others).  You could spin this out into friendship networks and back too: Discussing books with like-minded friends leads to recommendations for other books from a similar political perspective, which are then purchased and contribute to Amazon's data.  Amazon's buddy book system may actually cause political polarization, and would appear to at least represent a source of intertia in that phenomenon.

Steven R. (Steve) Corman
Professor, Hugh Downs School of Human Communication
Arizona State University

Vice-Chair, Organizational Communication Division
International Communication Association


-----Original Message-----
From: Steve Borgatti [mailto:[log in to unmask]]
Sent: Tuesday, March 04, 2003 12:53 PM
To: [log in to unmask]
Subject: A window in the American souls -- politics and book
co-purchasing patterns


This page by Valdis has been making the rounds. It's pretty interesting ...


>  <<Political Patterns on the WWW.url>>