All items about social graph


At Web2Expo this week, I chatted with Kevin Weil, who runs analytics at Twitter. We’ll publish the interview in audio form here soon, but one of the most interesting things I learned from him was how Twitter infers a relationship. Not all followers are equal, apparently.

  • If there’s a symmetric follow between two people, Twitter assumes they know one another and the relationship is conversational.
  • On the other hand, if it’s a one-way relationship — I follow the Bay Bridge, but it doesn’t follow me back — then Twitter assumes the relationship is around subject matter (in other words, I want the traffic updates from the bridge.)

Does this mean that if you’re the kind of person who follows everyone back, you’re not thought of as a subject matter expert? If so, expect a rash of unfollowing as users tweak their online profiles.

As Twitter’s opinion of us becomes increasingly important for things like advertising effectiveness, people are bound to try and game the system. Analysis of social graphs will be part of an arms race similar to that seen in Search Engine Optimization, where unscrupulous marketers try to convince Google to list them in search results. Call it Social Graph Optimization (SGO).


The social graph land grab

The map of how humans know one another will be a tremendously valuable thing. Internet giants like Facebook and Google know how valuable the data will be — it will govern everything from how we advertise to how we give people security clearances. We’re in the middle of the biggest social graph land grab in history.

In the absence of clear guidelines or legislation with teeth, however, the industry is taking an ask-for-forgiveness-not-permission approach to social mapping.

Google has had to pivot quickly in recent months from mapping the web to mapping its users. At least it’s transparent about what it knows: the company publishes its social network mapping, showing who it knows you relate to and how.

Others aren’t so up-front. A Facebook gaffe installed applications when you visited other sites. As this piece points out, when apps install themselves it’s called malware; but on Facebook it’s a feature.

There’s little to discourage Internet giants from building this map, and if you’re online in any way, you probably can’t hide. On today’s Internet, everybody knows you’re a dog.


Here’s a piece I just wrote for O’Reilly, which outlines many of the themes we’ll be discussing at this year’s Gov 2.0 Expo in Washington, DC at the end of May.  It was fascinating to write, and had input from several of the panelists and organizers; this is a hot topic in an era of national security concerns and unclear privacy legislation.

Peer-reviewed identity in the era of open social graphs is a game changer. Consider, for example, the work involved in creating a false identity today: Photoshopping childhood pictures, friending complete strangers, maintaining multiple distinct Twitter feeds, and checking in from several cities. It’s enough to make Bond retire.

Powered by WordPress, based on Mina theme.