I don't know that there is anything wrong with Chrome (the browser), other than that there is no real version of it for GNU/Linux or Mac OS X, and its aesthetics are questionable (as well as Google's political act of muddying the water by introducing yet another browser). Chrome OS is a lot more than the Chrome browser and the commonality of name is a marketing affair.
AFAICT, Chrome OS is a Google'ized stripped-down and customised GNU/ Linux with a UI that is built around the browser and "cloud" (network) based applications running within the browser.
There are a few things wrong with this idea:
If the idea is that the experience of GNU/Linux is intended to be more user-friendly, there is already an effort underway: Canonical's Ubuntu distribution.
But the real issue is this:
Google has bet on the idea of "cloud" applications i.e., the idea that you are always connected to the Internet and are better off running all your applications at some web site, where your data also resides. The presumed advantages: accessible from anywhere, any computer. No local tool install, upgrade, etc. Cheaper. Your data is secure and backed up. And if you are using a low-powered "netbook" (tiny laptops at very cheap prices) then this may be all you want from your computer.
Google's motivation is obvious: as long as they are dependent on operating systems and browsers made by others, those others (Microsoft, in particular) can not only limit their cut of the pie but can also use their platform advantage to promote competing products.
So, Google greatly benefits from users adopting the cloud model. But do users benefit?
I believe not, both in the short and long term.
First, short term: cloud based apps suffer various limitations that are a natural result of their lack of access to all local resources as well as their lack of a long UI development history. Regarding the latter, as one of the commentors on Fake Steve's blog noted, web apps today rely on the hideous kludge that is JavaScript + AJAX (not to forget, HTML/CSS). On top of that, Google has (IMHO) a poor track record when it comes to UI. And the assumption that near-real-time always-on Internet access is hardly justifiable.
Second, long term: I am not at all sure that sacrificing all your data to a corporate entity, locked into their storage format and accessible via their application (albeit with some sort of "export" option) is NOT a regression. Open storage formats, separation of (a) application, (b) data, and (c) communication/protocol, was one of the greatest strengths of the model that some of these services (e.g: Gmail) are trying to replace.
--ravi