Tuesday, November 11, 2008

Surveilling Sickness

Drudge has a flash up about Google volunteering to provide a new tool to help federal officials determine where searches for "flu" are coming from geographically. I've included the full text of the Drudge flash below.

I think there are some really valuable uses for Google's aggregated hive mind. Perhaps this is one of them. Is this sort of thing easy to subvert? If you wanted to make it appear that there's more interest in "flu" somewhere, you would have to generate more searches from that area. I don't think it would be hard for a nefarious group to drive around to various local wireless hotspots and do that. It sounds like Google's using a cluster of search terms, so maybe it is not easily subverted.

And how do they deal with local media inadvertently driving the searches? For example, if a local news report causes a spike in the searches for "flu", does the CDC respond before figuring out that there are no more flu cases than usual? Are these sorts of false positives common enough to obscure the signal in a lot of noise? I guess we will find out.

Update: Slashdot has more including a link to the new Google Flu Trends which includes information on how it works.

Update 2: Instapundit has a couple of links covering the privacy angle. In one, Orin Kerr promises to write an article on a Search Engine Privacy Act. In the other, the Vodkapundit's Stephen Green, reminds us of Google's moto: don't be evil. I still think Google's Flu Trends is possible without infringing on privacy, but I understand the concern. Drudge's headline "Sick Surveillance" is much more inflammatory than it needs to be. That said, I'm very interested in Orin Kerr's Search Engine Privacy Act.

Full text on Drudge:
Tue Nov 11 2008 15:34:50 ET

GOOGLE will launch a new tool that will help federal officials "track sickness".

"Flu Trends" uses search terms that people put into the web giant to figure out where influenza is heating up, and will notify the Centers for Disease Control and Prevention in real time!

GOOGLE, continuing to work closely with government, claims it would keep individual user data confidential: "GOOGLE FLU TRENDS can never be used to identify individual users because we rely on anonymized, aggregated counts of how often certain search queries occur each week."

Engineers will capture keywords and phrases related to the flu, including thermometer, flu symptoms, muscle aches, chest congestion and others.

Dr. Lyn Finelli, chief of influenza surveillance at CDC: "One thing we found last year when we validated this model is it tended to predict surveillance data. The data are really, really timely. They were able to tell us on a day-to-day basis the relative direction of flu activity for a given area. They were about a week ahead of us. They could be used... as early warning signal for flu activity."

Eric Schmidt, GOOGLE's chief executive vows: "From a technological perspective, it is the beginning."

Thomas Malone, professor at M.I.T.: "I think we are just scratching the surface of what's possible with collective intelligence."


No comments: