Another cynical take on the nofollow tag

July 1, 2012

What is now a relatively long time ago, I wrote a cynical take on the nofollow tag. I've recently come up with another take on this issue, one that's even more cynical than before. As before, the basic premise is that nofollow is really a clever way to improve search engine results (which in practice means Google's results); however, this time I think the mechanism is different.

Google's original PageRank algorithm assumes that a link from site A to site B is a meaningful signal; it conveys some sort of authority or deliberate interconnection from A to B. This is true if the author of a site put in the links themselves, but it is not necessarily true if the links are added by someone else. For example, if the link is from the name of someone leaving a comment on a blog to the commenter's blog or website. This is true even if (and in fact especially if) the comment (and the links that come with it) or other user-contributed content is not spam.

The original Internet that Google indexed did not have very many blogs and other sites with user-contributed content, and so PageRank's assumptions worked well. As blogs and comments and so on started becoming more and more popular, this assumption started to break down. To deal with this, Google came up with the nofollow tag as a hack; PageRank could use the nofollow tag as a sign that, basically, a particular link didn't mean what PageRank assumed it did and so should be ignored.

(As in the original entry, Google got the most benefit out of nofollow from having the top blogs and other sources adopt it, because those are the blogs where outbound links influenced Google search results the most.)

I believe that Google has long since moved away from straightforward PageRank, so I suspect that its current algorithms don't really need the help that the nofollow tag might give them.

(As before, I am not necessarily a cynic and I like the nofollow tag for its other effects, at least when search engines respect it properly.)

Written on 01 July 2012.
« What Linux distributions should do to help their Python 3 transitions
The well behaved Unix program and job control signals »

Page tools: View Source, Add Comment.
Login: Password:
Atom Syndication: Recent Comments.

Last modified: Sun Jul 1 23:49:16 2012
This dinky wiki is brought to you by the Insane Hackers Guild, Python sub-branch.