Page ranking

UPDATE:

By Cyril:

For me, this (nice) story should bring to us two conclusions:

1. PageRank and Google are not that good
Should'nt a decent tool be able to recognize by itself that these pages, the one without and the one with the "www", are the same, and that there is no need to index both of them?

2. People should be aware of the problem...
...and choose by themselves to check if the website they want to link to *really* need an extra (and quite cumbersome) "www".

On the contrary, the solution proposed by the author (in short, that all webmasters should communicate on their "www" address, even if they don't need to) is not the good one. Why bother with something longer and unuseful?

By Jesse :

You can't just assume they are the same site, because it is perfectly possible to have different content at www.mysite.com and mysite.com. I suppose google could implement a CRC32 or MD5 checksum, and then check it against the other pages in the DB and then do a full text compare if it finds a match. But, this would add to the processing time and this isn't exactly foolproof, because it takes quite a bit of time to re-index, so it is possible that the pages will change between when www.mysite.com is indexed and mysite.com is indexed. You could minimize the chances of this happening by doing a sort on the domain for your indexing order, but it still isn't foolproof.

By Paul :

Pagerank isn't quite as important as it's made out to be, it's keyword-to-content relevancy.

There are sites with lower pagerank that score higher than others for searches - and this trend is only going to get stronger, as Google get better at working out what you're looking for.

It should also be pointed out that the Pagerank you see in your Google toolbar atm is highly inaccurate at the moment... :)