My experience with the use of the canonical tag is limited, and I've always thought of it as a tool to minimize rogue duplicate content within a site.
Lately, in light of Panda, I've been trying to get proactive about identifying and addressing "thin" pages. (All of this is in the context of an ecommerce store.) By "thin" pages, I'm mostly talking about individual product pages where there is little content on the page beyond a basic description. I would agree that there is very little for googlebot on those pages, and they probably aren't necessarily appropriate to be included in the index. However, many of them have inbound links and are an important part of rounding out that category of products. It seems a shame to discard their importance.
Mostly, what I'm hearing is that the best practice is to apply a no-index tag to these pages. While that is a viable option, I hate to completely lose any importance the Google has attributed to them.
In one of the Matt Cutts videos, he made a short reference to using the canonical tag to group together similar pages, not necessarily duplicate. He even made a brief mention of using canonical to roll a page up to a category type page.
Is this a legitimate use of the canonical tag in an ecommerce environment? For a thin product page, would it make sense to link it to a similar product page, or even the appropriate category page? If so, that's great, though it seems like a radically different interpretation of the canonical tag than its original purpose.