13th
Google Flu Correlates Its Flu Data With CDC Figures
Google.org is about to publish its methodology for monitoring flu seasons for the US in the journal, Nature. This is a really interesting use of search data, but I wonder what happens when the method is published? When the method is known, then anyone (who has a large enough incentive) might try to game the system. So how resistant is the method to attack? I assume the method can be altered to be more resistant to potential false data attacks, and if so, then similar trend-detecting methods could be really valuable as well. I can imagine VCs using search trend data to help determine which startups are hot and investment-worthy. Stock prices might also be predicted (as well as elections)…. I think this could be potentially better than “prediction markets” — due to the anonymity of aggregated data.
