Somebody at Google had a great idea that I didn't even think about: marking links that shouldn't be considered when calculating PageRank. Actually I shouldn't be surprised that they thought of it; after all, PageRank is their crown jewel.
The magic is simple: add the attribute/value pair rel="nofollow" to any link. This means I can now set nofollow
on a link-by-link basis, rather than on a whole page (which has been possible for years). That's why this is new and cool.
Google recommends using the rel="nofollow" attribute for all links, trackbacks, etc., but as a commenter said on Shelley Powers' post on the subject,
Too bad that legitimate visitors will really be the ones who suffer from this. The spammers will just find other ways. Meanwhile, the folks that I am too happy to promote by allowing a live link to their site won't get the benefit they deserve.
He's right, and if I use this attribute, it's not going to be without a mechanism for filtering spam vs. ham. A blanket application is unfair to legitimate commenters, and in any case my site has so few comments that I have so far been able to delete them manually when they appear.
Robert Scoble has a different reason for being excited about rel="nofollow"; he wants to be able to link to sleazeballs in a blog entry without catapulting them straight to the top of search results. I like his idea (my PageRank is not as high as his but still fairly good).
rel="nofollow" is a neat new tool, but it's still just a tool. What matters is the application.
pjm says:
I noted back in August that Ian Hicks was talking about this. I guess Google decided not to wait for the next spec?
(Actually, from the Slashdot thread it looks like the rel attribute is XHTML-safe; the value is not a recognized one, but any value string is valid. So this doesn't actually tromp on existing specs.)