Complexity and Social Networks Blog

This is mostly a stylistic point regarding the impressive framework Anna presents, but hopefully is relevant to many of us seeking to present a complex model to a diverse audience, which can be a bit much to take in quickly.

In Anna's presentation, there were some enthusiastic exchanges probably based on misunderstandings.

It seems that given a comprehensive framework, it should be possible to reproduce more pedestrian relationships, associations & dynamics as special cases. (e.g., Properties of financial flows while setting the influence of relationship levels to zero.)

If this is the case, then the comprehensive nature of the model may be demonstrated to a diverse audience by presenting some examples of these special case scenarios that reproduce some of the favorite phenomena specific to some disciplines (e.g., finance, economics, marketing, etc.). Such anchoring may help ease disciplinary objections (although there is always the risk of opening new ones), while explicitly demonstrating the dynamics that are being integrated and augmented.

I agree with Brian's point about illustrative examples. I'm curious baout the boundary conditions under which total social utility (in terms of pareto exchanges) increases with social interactions?

Two questions insprired the fascinating talk:

* I was very intrigued by the attempt to integrate social network theory with more conventional graph theory problems, most of which are dependent on flows. What's the flow? Could be information, spilling into the info diffusion model world. Could be trust: PGP key rings have been studied for transitive trust. Organizationally, can various actors control resources that are distributed through interactions? Is the representation of a dynamic social network itself improved with the inclusion of flows?

* Methodologically, I've been trained to value parsimony in modeling. I was thus a little uncomfortable with the number of parameters in the integrated model presented. While the optimization problem wasn't overly complex, how many parameters have to estimated for (21) and (45) to be identified? Anna suggested that, if one is testing specific theories, having a problem with this many dimensions is not problematic; does anyone else have thoughts?

Brian,

Thank you for being the first to comment/post.

If one sets, say, the relationship levels to zero in the model, the model will collapse to, as you may have inferred, to a financial model that has already appeared in the literature. I will be posting the talk when I get back to Amherst this weekend and in the ppt there are citations to journal articles on which the more general model with the inclusion of the social network component is built.

Brian,

Thank you for being the first to comment/post.

If one sets, say, the relationship levels to zero in the model, the model will collapse to, as you may have inferred, to a financial model that has already appeared in the literature. I will be posting the talk when I get back to Amherst this weekend and in the ppt there are citations to journal articles on which the more general model with the inclusion of the social network component is built.

As a co-author of the papers "The Evolution and Emergence of Integrated Social and Financial Networks with Electronic Transactions: A Dynamic Supernetwork Theory for the Modeling, Analysis, and Computation of Financial Flows and Relationship Levels"(with A. Nagurney and L. Zhao), to appear in Computational Economics, 2006, and “Dynamic Supernetworks for the Integration of Social Networks and Supply Chains with Electronic Commerce: Modeling and Analysis of Buyer-Seller Relationships with Computations� (with A. Nagurney) (2004), Netnomics 6, 153-185, I would like to thank Brian Rubineau and Allan Friedman for their comments.

I agree that presenting examples of special case scenarios that reproduce some of the favorite phenomena specific to some disciplines is definitely a good idea to reduce disciplinary objections.

We believe that it is important to include people's behavior and interaction with other network systems into a model that tries to explain how social networks evolve. With our framework that is inspired by transportation research we try to make a first step into this direction. Specifically, we want to show how changes in people's behavior influence the development of different network systems and how these networks influence each other. Contributions and insights from different disciplines can greatly enrich this framework. Therefore, the interdisciplinary discussion made possible by the Institute for Quantitative Social Science is of high value and I would like to thank everyone for their contributions. Furthermore, I would like to thank the colloquium organizer, Professor David Lazer, and the Kennedy School for their support.