A while back a blog post popped up here arguing that Chrome is a repackaging of a new Google bot: that Chrome was developed first as a crawler, then later repurposed as a desktop browser.
There's no real proof for this of course, but it makes this change to the Google bot's behaviour make sense and explains Google's massive investment of programmer effort into Chrome and everything surrounding it (e.g. WebKit/Chromium, V8, Chrome's update mechanism).
This was all back in Oct - Dec of '11. Basically we learned that Googlebot handles JavaScript and AJAX pretty much like a browser.
When it comes to AJAX, it appears to index the content under the destination URL of the XHR in some cases, while indexing it as part of the page making the XHR in other instances. Something about the way the AJAX request is made causes Google to treat it like a 302 redirect at times.
Standard JS window.location redirects also appear to be treated as equivalent to 302 redirects.
@dsl - I suspect you're correct. The Google Toolbar, Chrome's Opt-In Program, The Search Quality Program, and now Google Analytics Data (since the TOS change) are probably all being used to train the behavior of Googlebot when interacting with elements on a page.
Google also has plenty of patents related to computer vision, and their self-driving car is road-worthy... so processing DOM renders of the page ala Firefox's 3D View/Tilt is probably small potatoes for them.
http://ipullrank.com/googlebot-is-chrome/
There's no real proof for this of course, but it makes this change to the Google bot's behaviour make sense and explains Google's massive investment of programmer effort into Chrome and everything surrounding it (e.g. WebKit/Chromium, V8, Chrome's update mechanism).