Ian Betteridge:
The real problem, though is that “Internet on the TV” is not where TV watchers are going. Instead, most TV watching is trending towards being a two-screen experience: you watch the show on the big screen, and chat about it on Twitter or Facebook using a mobile, laptop or tablet. The idea that you do everything on the same screen is just too ’90s.
That’s one solid theory, but I wouldn’t completely discount doing more than one thing on your TV screen from ever happening. Before we get to that point, though, there are much bigger challenges to overcome.
A key problem right now is that living room interfaces are terrible, and Google isn’t anywhere near the the top 10 list of companies which have a chance at cracking that nut. No one wants a keyboard, a mouse, or this hideous Frankensteinian hybrid anywhere near their living room:

With how hot and flexible smartphones and tablets the iPad have become, people seem happier, at least by comparison, to use well-designed apps on those to control other devices in their living room. But the group of people who want to do that still seems to be pretty small.
The real problem with “internet on the TV”—besides dual screens or bad hardware and software interfaces—is one of chickens versus eggs and content out of context.
Few devices and services have achieved anything that you could call success in this space—the Apple TV, Roku, Xbox 360, and maybe PS3 top this short list. But even then, look at their success in the category of “internet on the TV”: it’s all packaged, proprietary services like Netflix, Hulu, and the iTunes Store.
The Boxee Box is probably the one decent “internet on the TV” device that preceded Google TV and gotten anywhere, but even it hasn’t gone very far. Again, its best highlights are prepackaged services which Boxee had to cut deals for—Netflix, MLB TV, some Vudu movie service, and Pandora.
The internet was never designed for the living room, so scraping its content for a device it was never meant to be displayed on has so far proven to be extremely difficult at best, but generally futile. If the content is barely there to attract users, businesses don’t have much motivation to listen to that one guy down in the design department screaming “experience matters!”