Tedster, WebmasterWorld's Administrator, posted an excellent thread at WebmasterWorld asking how do you think Google handles the canonical issues they find on their own? By that he means, if webmasters don't use a 301 redirect or use Google's canonical header tag to instruct Google on how to handle that URL, how would Google handle it?
Would they cosmetically clean up the search results so that there does not appear to be any canonical (duplicate) URLs in the results? Or do they actually decide to implement a 301 on your behalf and pass all the 'link juice' from one canonical URL to the parent URL?
First, take my poll and then I will give you my thoughts on it:
I really think for the most part, where Google is not confident on which should be the main URL, Google will only apply this cosmetically to the search results. I remember when the new canonical tag came out and Google warned to use this carefully, because it is as powerful as a 301, but without actually being physically redirected. For Google to apply their own 301s, hidden be that, is extremely dangerous for both Google and the webmaster. I would assume, in certain cases, Google does do this, but I am not sure if they do this in most cases. Of course, this is just my thoughts - I have no hard evidence, since I never really tested it myself.
I would assume Google would want to be right 100% of the time on this. I would think that would be a goal. And when they are 100% or even 99% right, implementing this would make sense.
Forum discussion at WebmasterWorld.