In his article, The Google PC generation, Jon Udell suggests some interesting ideas for an imagined Google PC operating system that could capture and index information from user tasks. In response, David Weinberger suggests a somewhat different approach in his Semantic Behavior Index post. (I left a comment on David's post wherein I mention that some of my current experiments on the iCite net might do something like "semantic behavior indexing").

The thing that I wanted to suggest though is that Google is not PC.

Jon's piece frames the discussion around the PC, personal computer. In other words, it isn't about the way Google looks at information as much as the way Google might look at the information tied to the legacy software and operating system structures of your PC.

The PC is a legacy of information structures developed with little or no connectivity in mind. The PC is more about information in isolation than about connected / collaborative information. But, Google begins with connected / collaborative information.

It's not that Google couldn't index the data / info that lives on a PC—I'm sure they could do a good job at it. It's just that what's great about Google is that it comes out of and enhances an environment of connected / collaborative information (the web).

While we might want and need to get our isolated PC apps to talk with each other and, oh yeah, maybe talk with other PCs, we could also imagine an information environment designed for connectivity / collaboration, and also personalized: and that is where I think Google's role would be especially interesting and useful.

I think what Ross Mayfield calls Site-Flavored Yummy, i.e., Google's new Site-Flavored Search, is another early sign of the personalization features that we'll see coming out of Google. Eventually, Google is something more like the OS, and the PC is more where the "play" and "record" buttons get pushed.