When Google announced support for the canonical tag just about a year ago, webmasters were excited for the possibilities of a serving a 301 redirect to spiders but not users. But when should you not use it?
A Google Webmaster Help thread has Google's JohnMu explaining some situations as to when you should or should not use it.
In summary, think of the canonical tag as a real 301 redirect for spiders. If a spider is redirected away from new content, then that is an issue. When it comes to paginating content, such as product category pages or article archives, you really need to give a way for search engines to find that content. If you set the canonical tag to redirect spiders from page 2, 3, 4, etc of your product category pages to page one, then the spiders might never be able to index the products on page 2, 3, and 4.
John explained this well, saying:
Pagination: this is complicated, I personally would be careful when using with rel=canonical with paginated lists. The important part is that we should be able to find all products listed, so at the very least those lists should provide a default sort order where we can access (and index) all pages. Since this is somewhat difficult unless you really, really know what you are doing, I would personally avoid adding rel=canonical for these pages. One possible solution could be to use JavaScript for paginated lists with different sort orders, for example, that way you would have a single URL which lists all products.
This is something to keep in mind when building out the canonical tag within your content management system.
Forum discussion at Google Webmaster Help.
