The details of Google’s MVNO in the US have now been released - in the modern Googlish way - by a blog post. The rumours and leakages we reported yesterday (see - Google is preparing to launch its MVNO - no balloons involved) were about right (difficult to keep a communications service preparation secret when, by definition, you have to involve all those partners) and now there’s more flesh for those bones.

Project Fi (taking the Why? out of WiFi perhaps?) is described by Google as a way to “push the boundaries of what's possible.”

Turns out it’s been ‘possible’ to convince Sprint and T-Mobile that partnering up with the Cookie Monster will, if the project succeeds, hurt AT&T and Verizon far more than it might strategically hurt Sprint and T-Mobile.

Google says it has the technology to intelligently connect to the fastest available network whether it’s WiFi or one of its partners’ LTE networks. To reinforce this WiFi first approach, it’s going to refund unused mobile data.

The trap with this sort of WiFi first approach has traditionally been the complexity of the price-plans designed to incentivise WiFi usage. Google think it’s avoided this with a simple formula. Each GB will cost $10 per month up front (not especially cheap by European standards). When the user gets to the end of the month the unused data (if there is any) is refunded on a pro-rata basis. Thus the user is incentivised to lock onto WiFi as much as possible and really start to save money. Voice calls will be enabled over both LTE and WiFi.

Google says the user’s phone number lives in the cloud, so services can be accessed on “just about any phone, tablet or laptop - so the next time you misplace your phone, you can stay connected using another screen.”  This approach fits in with Google’s Chrome-based apps and services, where once logged in from any device, all the services are the same, bookmarks synced and so on.

It’s a sign up for the service and get invited deal to start with and it will only be available Motorola’s Nexus 6, described as “the first smartphone that supports the hardware and software to work with our service.”

Back at the beginning of the year (see - Why might Google want to go wireless?) we wrote of the Google MVNO rumours and pondered the motivation of T-Mobile and Sprint: “What amounts to a joint MVNO with Google (should it take place) is [for T-Mobile and Sprint] almost a network sharing deal in its own right and might help the two networks load up extra customers (albeit at one remove) as what most agree is likely to be a price war with Verizon and AT&T shapes up over the next year or two

In support of this idea there is evidence that the services that Google wants to deploy will be app-based (presumably on Android), so the handset could effectively make a network selection based on which of the two carriers is available and which is the least loaded, thus making the capacity and availability story stronger from a customer point of view. For Google Mobile [Project Fi] users, that’s a virtual merger.”

It’s not as if this isn’t what mobile services already do. My iPhone spends most of its time falling back to GPRS, way down at the 2G sub-1GHz end of the spectrum: it’s slow to the point of unusable, but it’s always there.

So Google is simply taking network selection a bit further and enabling it across network operator boundaries. The real innovation here is the network aggregation that Google has managed to negotiate for the service with its partner telcos. Is it a game-changer?

A strong challenger using some sort of WiFi first model was bound to arrive eventually - it happens to have been Google, so no surprise there. The interesting bit will be how Google fleshes out the services and how, if it all goes swimmingly well, the big carriers respond.