Thursday, May 1, 2014

Google Mail and the Limits of Data Mining

Data mining has its limits, as Google decided today. At CNET:

Google has always stored abstracted linguistic summaries of the data in e-mail messages it hosts. It uses this information to help select the advertising that users see, mainly on other sites. It may also use it for other purposes we don’t know about. But the mere collection of this data, from people who have not individually signed up for the Gmail service, may violate California wiretapping law. And so today, Google decided it would stop collecting the text of email messages sent and received by students who use its Apps for Education suite. It will go on to make similar adjustments to commercial versions of Google Apps.

Data mining has limits when it involves data that you don’t quite have the right to read. Google had held that its process of abstracting the text of email messages and other messages from one person to another insulated it from its duty not to intrude on people’s private communication, but now has decided that view might have been pushing things a bit too far.