Hiding in Plain Sight, Google Seeks More Power - New York Times

There's a really good story on the New York Times about the incredible computing facility that Google is constructing in Washington State. It begs the question of what Google actually wants all that computing power for - but then you read something like the quote that Nicholas Carr picks out of an interview with Google's Hunter Walk, where he claims that "when we say we want all the world’s video, we really do want all the world’s video." And they want it at high quality, too.

Nick can't see what the end of this is, and simply puts it down to Google hubris. But that's missing the long-term point: fast access to all the world's information, including video, gives immense power. The ability to use that data to teach AI's, as a pool for learning about how humans connect things together, is all incredibly valuable - in some ways, more valuable than the content itself.

Google is attempting to solve the oldest problem in AI: how to model the ability that humans have to put things into context. The process of contextualising information is at the heart of a lot of the incredibly smart things that people can do, from what we normally call "common sense" to our ability to make predictions about how things behave.

But why would a commercial organisation like Google be interested in doing something like this? Because context is at the heart of its main business: selling advertising. Put an ad in the right context, and you're much more likely to sell something from it. So understanding how humans put things into context, how they make connections, is the core of Google's technology.

All of Google's efforts - from the search engine itself through Blogger and Writely to the new Google spreadsheet - are about providing it with a bigger pool of data that it can use to create something that understands context, whether you're talking about words, numbers, audio or video. A document that lives on your hard drive is dead data to Google, unless you're using Google Desktop Search of course. A piece of data that lives on its servers is live, and available for its use. Welcome to the Googleplex.