Barabasi on Mobiles as Social Dashboards

Alex and I are spending the week in Budapest at the Hungarian Academy of Sciences, which is sponsoring a conference called "Towards a Philosophy of Telecommunications Convergence". It's organized by our new friend Kristof Nyiri, who wrote one of the most interesting takes on information technology ever - "Thinking with a word processor".

Alberto Lazlo Barabasi, one of the fathers of social network analysis, gave a fascinating paper (abstract PDF) on "Time and Motion in Mobile Communication", in which he described a project that uses large datasets of mobile phone subscribers to describe analyze how people move about cities. As his abstract states "cell phone usage offers access to patterns of human dynamics and mobility at a level and detail unimaginable before.

He began by explaining power law distributions - the mathematical curve that describes the distribution of many of the things we do. Put simply, human activity is bursty. This shows up in everything from Einstein's well-documented post correspondence to how often we visit the library. Barabasi was interested in whether this applied as well to our mobility patterns.

Animals have been shown to move in what are called Levy flights - a variant on the random walk distribution. Essentially, animal motion is fairly random. We can also see it in studies of currency movements like Where's George. Sociologists have looked at these data sets as measures of human movement, assuming that people are carrying these dollar bills around in their pockets.

Barabasi set out to do this for human activity using a huge dataset that recorded the time of mobile phone calls and the cell site ID - which allowed them to geolocate the callers to approximately the neighborhood level. They can then create metrics for each individual that describe their movement patterns in space and time - essentially a mobility signature for each person. Then you can look at the overall patterns, and as it turns out, we are not doing Levy flights - there are significant differences between how different people move.

This kind of analysis has many applications, but on of the most interesting he raised is using it during an epidemic to look at real-time flux of people across parts of a city. (one of Barabasi's appointments is at the Harvard Medical School) Using this data, you could develop a metric of how many people are flowing across every cell site boundary. I imagine you could also develop predictive models of this flow.

The next step in their research, something I have through extensively about from an urban studies perspective, is the effect of synchronization (using mobiles of course) affects the randomness of these mobility patterns.

This paper caught my attention because there are a number of interesting research projects that are starting to leverage the mobile as a real-time sensor, to create some very interesting pictures of emergent social phenomena. We looked at Nathan Eagle's Reality Mining project a number of years ago, and Carlo Ratti's Real Time Rome perhaps takes this to its logical conclusion. But Barabasi's paper has pushed the mathematical basis for these investigations even further... it's as if we're moving into a new world of social physics in which massive data sets enable algorithmic mining. Interestingly enough, it seems that Boston is emerging as the global epicenter for this new research niche.

Barabasi's research site is the Center for Complex Network Research.

I missed his 2002 book Linked: How Everything is Connected to Everything Else and What It Means (what a title!), but it's probably time to go back and take a look.

Alex is blogging the rest of the conference over at the End of Cyberspace.

Technorati Tags: Hungary, information technology, social networks

Comments

[Peter]

Well, sure following epidemics is very important. But, and here I'm feeling bleak and a tiny bit snarky....

I would guess that the main use for this kind of technology is surveillance and anti-terrorist activities. And it's rather chilling, not at all fascinating or really cool (this is actually a similar tone to an earlier post which was more tongue-in-cheek - isn't there a technological determinism-sans-ethics that's worrisome here?)

I attended a talk by Marieke de Goede this past summer, where she's talked about governments are using realtime network data in finance to 'track' terrorist networks.

From her description of the project:

Data Wars: New Spaces of Governing in the European Fight Against Terrorist Financing

In the aftermath of the events of 9/11, political debates and public investigations emphasised the problem of locating information about risk and threat in advance of a terrorist attack. The idea that data offer the solution to problems of security risk has quickly established itself at the forefront of what has come to be known as the war on terror. Financial information occupies a special place within data-led security practices, because it is assumed to be able to reveal 'blueprints' of terrorist networks. The goal of this research project is to map and analyse the reconfigured spaces of governing that are emerging through the deployment of financial information in the European war on terror.

She had really interesting prelim findings, including a chilling 'if you're not guilty, you have nothing to worry about' kind of mentality from States, and a specific example of how even the lack of activity becomes suspect: a family in UK (Muslim, natch) whose flat was raided in the name of anti-terror, and as evidence was cited that they had a big stash of cash under a mattress someplace. So lack of regular, surveille-able financial transactions becomes its own suspicion (they didn't put it in a bank apparently because of religious restrictions on interest).

Social physics is an apt metaphor, one wonders if lessons of physical physics during development of super-cool tools for, you know, splitting atoms and stuff, are still thought of as important..