incrediBILL - 7:29 pm on Feb 6, 2013 (gmt 0)

OK, I do SEO as well and that's a completely invalid use of the canonical tag and it will bite you in the butt. The canonical tag tells Googlebot which page should be the proper page to be indexed and not all pages should refer back to the main index page.

I think tedster mentioned that you were using it improperly in the second post.

Here's an example of the valid use of the canonical tag if you have URLs like these:

http://example.com/mypage.html?a=1
http://example.com/mypage.html?b=1
http://example.com/mypage.html?a=1&b=1

Google thinks those are 3 different pages instead of one page.

Having all of those pages use the following canonical tag solves the problem of bleeding PR per each parameter.

<link rel="canonical" href="http://example.com/mypage.html"/>

What you're suggesting puts all pages back to the root and while you're right, it might give the home page a minor PR boost, if can cause all the other pages to not be indexed or appear in the SERPs because you've plainly told it that the page they landed on isn't that page.

I have a feeling Google has code that stops people from accidentally shooting themselves in the foot if all pages claim to be the root because they clearly state that the canonical tag is just a suggestion, it can be ignored which it probably is in the way you're implementing it which is 100% wrong.