Complexity and Social Networks Blog

Interesting lunch-time talk.

I'm still not sure I sense how much of these free-scale network findings relate to social networks since it is much easier to scale a WWW router to be 10x its current size than multiply the hours of socializing of an individual by 10.

In some places human interaction can be free-scale (once one has e-mail distribution lists) you could send the same message as easily to 1000 folks as 10 folks, although not clear that it has the same effect at the receiving end.

And in some contexts, like business hierarchies, you could have power laws in effect because people at the top of a business hierarchy may indirectly have thousands reporting to them.

I'm fascinated by what the social ramifications are of being a *power hub* in a social network (the Swede who has 1000 sexual partners) or the person who has 1000 e-mail contacts in a month. Does that person disseminate information more effectively? He or she comes in *contact* with more people (possibly, see note below) but there might be a negative relation between whether people trust the person and how much of a *power hub* he/she is.

Very interesting to think about what the relationship is between strong and weak ties in such networks. Are the same people who are power hubs in weak ties also power hubs in strong ties or is there is a negative relationship. (Sounded from Laszlo like there was some work that his colleagues were doing, didn't get a cite, at least on what happens to the structure of cellphone networks, for example, as one removes the strong links as opposed to the weak links).

You could imagine that there is a mathematical relationship in the spread of ideas between links and levels of trust (where strong ties might be a rough proxy for trust). So, assuming that a person X passed on some piece of information to person Y (rumor about a Hungarian power plant, a recommendation about product X, etc.) the likelihood that it would be acted upon by person Y or passed on in turn to person Z would be a function of the strength of the tie between X and Y. If this were the case, and if strong ties and weak ties were not positively correlated, the weak hub person A (with lots of weak ties) would face an advantage over strong hub person B (with mostly strong ties and a high number of strong ties, but way fewer ties than A) in being able to spread the word to lots more people, but how effectively this spread would be limited if the relationships from A were all relatively low-trust relationships. I'm sure that whether a recipient passes on information depends on lots of situation-specific factors (what the information is, the likelihood that the person passing on the info. might be acting in self-interest, how time urgent the info. is, etc.).

As people turn more to monitoring e-mail networks, cellphone networks, etc. because it is easily to get large-scale data, has anyone done any testing to see whether these technological-heavy modes of communicating are representative of low-tech social networks as well. In principle, the same factors that attracted someone to cellphones (for example) might also attract one to different kinds of social relationships (e.g., heavy e-mail users might be people who like F2F communication less or value productivity over strong ties, etc.). One couldn't really tell unless one did some comparison of the structure of such technological networks with more old-line methods or saw for instance whether heavy cellphone users are also the heaviest F2F socializers. (In principle, some of these modes could be substitutes for another so the most signficant cellphone user may not net-net be a higher socializer than a low-frequency cellphone users but simply have a higher % of his/her socializing happen over the cellphone).

Anyway, interesting fodder for thought.

Best.

Tom

Thomas Sander
Executive Director
Saguaro Seminar: Civic Engagement in America
John F. Kennedy School of Government
Harvard University
http://www.ksg.harvard.edu/saguaro