Joe Ganley
Writing code since 1979
I have been a professional software engineer for over 10 years. I have written many kinds of software, but my particular strengths are interactive graphics applications, compilers and interpreters, and algorithms.

I also enjoy writing, woodworking, and home improvement. Also this.


Email Me

Atom Feed

not from a security camera



Monday, April 26, 2004

A lot of recent
study has examined so-called Small World networks, such as the 'Kevin Bacon' network of actors, linked by appearing in movies together. I suspect it has already been (or is being) done, but it would be interesting to consider the difference between the networks stored in online social networking services such as Orkut and Friendster and the absolute networks they attempt to represent; in this case, networks of who knows whom. I believe that a couple of key phenomena would make these networks very different from one another. The first is that the online networks are unlikely to contain people who aren't particularly technical; that is, such people are unlikely to be interested in joining such an online service, and their friends, knowing this, are less likely to invite them. The second is that in graph-theoretical terms, the degree of people-as-vertices is limited. If you join a service, and you know ten people, chances are you would invite all of them to join. But if you know 10,000 people, presumably you'd only invite your closest friends and those most likely to be interested in such things; typically, perhaps, a few dozen people. This greatly limits the power of people to act as 'bridges' between otherwise disconnected portions of the graph (e.g. Rod Steiger is widely cited as the strongest bridge in the Kevin Bacon network), and I expect this network consists much more of large but disconnected cliques of people than does the 'real' who-knows-whom network. I've never been invited to join one of these services; if we can stipulate that I have friends, and I hope that we can, then I think we see these phenomena in action. Most of my friends are relatively nontechnical, and among those friends who are more technical, we probably aren't close enough friends for me to be one of those few dozen people they think of when they sit down to send their invitations. If it could be done in a statistically valid way, it would be interesting to have people rank their friends in order of closeness, and then to form a graph connecting each person only with their N closest friends, and then to see how the topology of the network varies as you change N. (Related humorous aside, via Bob Congdon.)

| permalink

Comments: Post a Comment

This page is powered by Blogger. Isn't yours?

The Long Now Foundation
The Long Now Foundation

Man-Bag Buying Guide
Man-Bag Buying Guide

Copyright (c) 1988-2004 by Joseph L. Ganley. All rights reserved except where otherwise noted.