facebook & predicting behaviour

by Hugh

All this data we are putting into the web – say, into our blogs and into facebook and elsewhere, could be used for much more than just figuring out what kind of sneaker ads we’re likely to want to see.

If you have a big enough and powerful enough database, and you felt like tracking more complex things than ad click-thru rates, you could start figuring out how different cues actually affect decisions, actions and opinions of specific people or groups of people. Let’s say you had a captive portal where all sorts of data about an individual (oh, say, interests, education, religion, location) and relationships (the people in that person’s network) and actions (causes they support, pictures they comment on), AND further more sophisticated content started going in there (say, blog posts – even a feed from a blog), then you could (and probably would) start analyzing what a person actually does.

And you could start correlating profiles with actions. Which in effect is what pollsters do, but with tiny bits of bad data from surveys and focus groups, from a discrete moment in time, and without any way to measure how responses correlate with actions. Whereas the net – and places like facebook – are enormous databases of detailed info about specific people, tracking not just static points of what they like or don’t like, but also some of the things they actually do on the net, over weeks, months or years.

And that means, if you have access to that database, you might have a good platform to craft a strategy to make people do what you want them to do.

While that’s a bit creepy when you think about Facebook and Myspace etc, let’s face it: as the semantic web evolves (linking content with context), as bandwidth and database processing power grows, the web itself might as well be considered a captive portal, and we *will* continue to put all this data into the system. Much that you might wish to know about me, including my opinions on various political issues and the news that inspires my outrage or laughter, can be found on my personal blog, for all to see, catalog, measure and track. Probably it’s a bit early for databases and modeling systems powerful enough to manage this kind of complex data – but it’s coming.

So the worries about privacy and your data are much bigger than just targeted marketing … there are much more sophisticated uses for our information.

Oh, and while I’ve been thinking this for a while, the spur that kicked me to think about it again, was the “about” page on Justin Hall’s website that says:

The web and video games are merging. All of information space is a shared multiplayer adventure. I am working to make that merging happen faster by developing “Passively Multiplayer Online Games” where your history of web browsing defines your online character.

Check the website of the research project: passively multiplayer.

Cool for vid games. But useful, and worrying, for all sorts of other applications.

UPDATE: check also wefeelfine.org and imagine the other sorts of text-strings, say, governments would like to track with such a pretty tool. (For instance: “I hate Bush”).