Sharad Goel writes:
We just launched an experiment about online privacy, and I was wondering if you could post this on your blog.
In a nutshell, people upload their browsing history, which we then fingerprint and compare to the profiles of 100s of millions of Twitter users to find a match. Browsing history is something ad networks and others can collect without your explicit consent, so we’re trying to understand how much information leak there is. The experiment is a little creepy but also kind of fun I think….!
Sharad and I have collaborated on some projects (though I’m not involved in this one) and just about everything he does is great. He’s full of brilliant ideas.
Just because I might face the risk of accidentally getting run over by a truck on the street one day this is akin to jumping in front of one on a test track just to see how it feels. :)
Like parachuting. Some see it as fun, others can’t fathom why anyone would want to jump out of a perfectly good airplane.
Won’t there be severe selection problems with how this study has been designed? The people likely to participate will probably not be a representative sample of anything remotely like the population (of … anything)? Of course, the stated purpose is to examine the degree of “information leak” by trying to identify users from their search behavior (at least I think that what it says) and this is a way to do that. But I suspect the resulting research will then be presented as a more general finding when it really only applies to the type of person who would agree to participate – a type that will not include me, or Rahul apparently.