Today Google aired a webcast where they whipped the curtains all the way off Google Chrome OS for the first time. It's about what most people expected: Chrome OS running on top of a thin layer of Linux, designed for netbooks -- and designed for people whose sole computing experience is the web. It's Google's netbook answer to Android.
Right at the start of the 'cast the Google folks tried to temper any expectations. They had no actual, branded devices to demo. They were, by their own admission, about a year away from any such thing. What they did have was a technology demo and some bits about what they were trying to accomplish and why -- most of which people have already pieced together on their own from what's been already floating around. But there were still a few key revelations.
First, the obvious stuff. The reason it's called Chrome OS is because it consists of the Chrome browser and very little else. This is by design -- the browser is used as the application-execution framework, including things like keeping processes isolated from each other. This isn't that difficult, since Chrome OS isn't designed to run arbitrary binaries from any old place. It runs the web, period, as a security measure and as a way to guarantee uniformity of behavior and experience. User data is synced seamlessly into the cloud -- local storage is basically used as a cache, and everything stored locally is encrypted.
The Chrome browser, Google said, was written to satisfy three needs: speed, simplicity and security. The system (browser and OS together) should just work, should just turn right on and work, and should stay working. All three are well in evidence with the current versions of Chrome on PCs, and those things are meant to carry over directly into Chrome OS.
The other big thing right now is the release of the source code -- which I'm sure will be glommed on by a number of people and built into something that people can run locally or in a VM, as a first toe in the water. But what Google wants most is to partner with hardware makers to get this on branded devices. The devices in question have to meet a certain list of hardware specifications, and the devices themselves are months away from even being announced. Tight integration with the target hardware also cuts out another raft of problems; if you know what hardware you're running on, you don't have to probe for things that might never be there. (Sounds like Apple's approach, frankly.)
On a technical level, all of this is impressive. Google has taken the Linux kernel, their own open source projects, and a slew of other associated work and ganged it together into a very elegant way. What's tricky is how Chrome OS's success with the world at large is going to be tied into the hardware devices it's packaged with, especially since no hardware pricing was announced.
Something else to remember: the gap between a netbook-class computer and a full-blown notebook keeps closing. If people are given a choice between spending $300 for a netbook and $400 for a Windows notebook (or an Ubuntu one, for fairness' sake), those $100 mean that much more functionality.
The Google folks were fairly blunt about the role of Chrome OS machines: "Most people who buy this device will have another machine; this is a companion device." They were, from what I can tell, not entertaining fantasies about people ditching their fat-client desktops for a Chrome OS machine -- as seemed the case with a fellow in the audience who asked a question about video editing in Chrome OS.
In many ways, Chrome OS -- even after it's released with its companion hardware -- is going to be a first salvo, a trial run. Google's trying to determine a bunch of things at once: the viability of a web-only device tightly coupled to its hardware; the willingness for the open source community to rally around such a project; the ways an operating system and its applications can be engineered to be that much more complementary.
I have no delusions, though, that they're also trying to see how much bigger a slice of the computing pie they can bite into. And seeing how many more ways they can get people using Google's services.
InformationWeek has published an in-depth report on new software models. Download the report here (registration required).