Matt Cutts: That’s a good question. Just as a refresher on the NOFOLLOW, it’s a general mechanism introduced in 2005. Essentially when we see a NOFOLLOW link going from one page to another page we say, “This page won’t flow PageRank and it’s dropped out of our link graph so it doesn’t flow any anchor text as well.” So essentially it doesn’t contribute anything in search engine rankings.
Now I think that if you look at what Wikipedia did a few years ago they switched from having FOLLOW links to NOFOLLOW links. A lot of people were sort of doing spammy edits of Wikipedia. You would see the same sort of stuff that you saw on the open directory project when people would try to subvert the process and get links without the editors or the people reviewing those pages really noticing. So, Wikipedia added no follow to a lot of those external links and as a result people sort of stopped spamming Wikipedia at least for PageRank purposes.
Now the interesting thing is that I think Wikipedia could probably follow a slightly more nuanced policy. For example; if you have a Wikipedia editor who has made a lot of edits or have been a member for a long time and the edits haven’t been reverted very much or they are vouched for by other members, or whatever the trust mechanism is, there is some reason that you believe this person should be allowed to flow PageRank. Then you could maybe make it so that any links that particular person adds to Wikipedia would flow PageRank.
I think you can also imagine things like Word Press, by default you might have a third party commenter not being able to flow PageRank but if you saw that they were leaving really good comments or maybe over time you trusted their comments a little more the ability to remove that NOFOLLOW so that it does flow PageRank I think would be great.
So NOFOLLOW was fantastic in terms of stopping a lot of the effectiveness in spam. A lot of people spray out a lot of links but they don’t necessarily have a lot of effect in our search result rankings. But sites can still be a little more nuanced. I think it would be great for example if Wikipedia would want to explore if it’s possible to allow some links that they trusted a little bit more and maybe some links that they think are still in a probationary period or don’t completely trust.
We’ll see if that evolves over time but that’s the general idea of why it might or might not make sense to have NOFOLLOW on a link.