Google's always approached design as an objective engineering challenge and less as an artistic endeavor. As a result, their designs are minimalist, utilitarian, and functional, but they sometimes seem to give up the ghost when they attempt to build complex or unique software like Buzz or Wave that could benefit badly from someone with a passion for aesthetics and UX.
Regarding Android, I'm encouraged by news that Google picked up WebOS's lead designer, (as reported here: http://www.mobilecrunch.com/2010/05/27/palm-loses-their-lead...) as WebOS is often lauded as an intuitive and innovative user interface in a manner Android simply isn't.
When Douglas Bowman (Google's former lead designer) left the organization, he cited, among other things, a fastidious, over-engineered, design-by-committee-and-hard-numbers approach that drove him crazy. He mentioned an example where Google had A/B tested 41 shades of blue to determine which to go with. That's demonstrative of the greater issue their culture seems to suffer from: if you want to bring about creative or subjective change, you often need objective facts and figures—metrics lacking by artistic design's very nature.
I wrote briefly about Google's methods in a bit of a rant on Facebook's whiplash-inducing approach to design here: http://www.htmlist.com/design/google-vs-facebook-interface-d..., but in the end, it comes down to balance. I have high hopes for Android 3.0 because I'm hoping that they just shipped with what they had and told themselves they'd worry about making it pretty later. Later is now, evidently, and it will be really interesting to see what Google does in this next round, especially if they've removed the handcuffs and decided to trust their designers.