RE: rel="nofollow" attribute


> I look at it a bit differently--markup as a way of serializing intent.

Indeed -- 'intent' not procedure.

> Most of the links this will be applied to have no preexisting rel 
> attributes. This in itself expresses a kind of endorsement, as Google 
> have noticed and incorporated into their algorithms. 
> Being able 
> to interpret this intent has been one of the factors in Google's 
> success, and it's changed how people express themselves in HTML.

But the links are provided by the person who made the comment, not the
person who wrote the blog. If I have a blog about cameras, and in it you
write a comment that says you think my view on camera A is wrong, and you
think camera B is the best, you can put a link in. This proposal modifies
*your* link, not *my* link. True, I could endorse (or otherwise) you the
poster, but if I am going to say anything about camera B I should do that in
separate metadata -- *not* by modifying your metadata.

The fact that Google currently draws a false conclusion from your
endorsement for camera B appearing on my blog (it assumes it's an
endorsement from me) is completely understandable, but doesn't mean it can't
be solved.

> Now comes rel="nofollow"
> In your blog you wrote: "its presence tells a spider not to 
> navigate the 
> link" but that's not exactly right. its presence indicates that the 
> author's intent in providing the link doesn't count as an 
> endorsement, 
> but mere information. Spiders, once they are programmed to recognize 
> this intent, can do with it as they please.

That's not the case, though. The proposal is quite explicit that in the
first instance the blog software developers are to put rel="nofollow" on all
links in the comments area on your blog. You have no say in it. So as I said
before, it's not about making a value judgement on the link, it's about the
blog software companies providing some metadata to the search engine
companies -- and they've done it in an unhelpful way.

> So I agree with folks who say that "nofollow" is a crummy name. Maybe 
> something like "noendorse" or "fyi"??

That would be far, far worse. That would imply that I don't endorse the
link. That's not what I want, and I'm sure many bloggers don't want either.
As I keep saying, all that is on the table here is a way for Google to stop
crawling -- it's Google that doesn't endorse the link, not me.

> But despite the name, it is 
> already influencing how people express themselves in HTML. (How many 
> times have you read a web page that says 'this person said something 
> really [stupid|shocking|untrue] but I don't even want to link 
> to it'? Or 
> sites that refuse to link to competitors?)

But that is completely different! What you have described is authors taking
a decision whether or not to link to information. As an aside, my personal
opinion is that if you refer to something you should link to it. If I say
that something is "stupid|shocking|untrue" then I should link to it so that
people can make their own minds up. Otherwise I should have the good grace
not to refer to it.

Back to the point though, the proposal from Google is *not* about editorial
control, it's the opposite. They are saying that any links in comments that
appear on my blog are not first-class links in the web world. Now, I don't
mind if they don't index those links, or if they ignore them for the
purposes of search rankings -- it's their business, their software, their
service. What I object to though, is the fact that they are unilaterally
adding some gibberish metadata, and presenting it as progress.

> Getting people to express themselves better, and writing more 
> intentional markup is a win to me, even if it's not 100% 
> architecturally 
> pure.

Ahh ... The old "purity" argument. Unfortunately Micah, it doesn't cut it in
this discussion. As I've said, there is no proposal for authors to "express
themselves better", there is only a proposal to allow one set of computers
to talk to another set of computers. The computers that deliver blogs to the
world will have some extra information hard-code into their pages so that
the computers that index the web can know when to stop crawling.

... with not an author in sight.



Mark Birbeck
CEO Ltd.

t: +44 (0) 20 7689 9232

Download our XForms processor from

Received on Friday, 21 January 2005 14:43:28 UTC