Tuesday, January 22, 2008

The power of crowds: wis.dm

It seems to me that there are more social networks than users. It's a joke, of course, but it's also partially true.
Indeed, each social network specializes on some kind of "environment". I am a passionate product manager with a focus on online communications, I like sailing and windsurfing, and I love torturing my family and neighbors with my clarinet.
That means that I belong to three environments, to the least:
  • Online techno geeks and community passionates
  • Sailing
  • Classical music
For each of these environment, I feel I belong to a network. It'l like the "networks" of Facebook, or the "communities" of Orkut, or the "tags" of Spock. Sometimes people (elements of the network) overlap, sometimes they don't. But they are, and will be, mostly the same elements. It's just a collection of people I won't improve my relationship with by using the social networking tool.

But then, how can I socialize with people that share an interest in common with me? How can I do this without pretending to add people to my network in the hope that he or she will be a good "network element" to be friend with?

The wise answer is: ask.
In wis.dm you ask lots of questions. People ask lots of personal, political, religious questions.
Everybody has an opinion, you have to answer a simple "yes" or "no". It is addicting.
You can make a choice by asking the crowds, they will give you the results: like flocks protect themselves from predators using their collective behavior, we humans can use our common knowledge to empower our own decisions with the power of wisdom of crowds.

Moreover, the magic that's under the hood of wis.dm computes an affinity score. People who think alike are matched and grouped. Clusters and "sub-flocks" form, and new groups with a strong cohesion on certain topics arise.

It's like the principle of cross-selling, isn't it?
What a pleasure for a marketing manager, all these profiles available... Even Google, with its immense archive of personal taster would be interested in this kind of data...