On 21 Jan, 2005, at 2:26 PM, Lachlan Hunt wrote:
> ...
> Indeed, nofollow is an extremely poor choice and I consider the 
> relationship to be quite harmful [1].  However, "no-credit" is not 
> much better because, although it more accurately describes its 
> function, it says nothing about the relationship between the 
> resources, nothing about what the linked-resource is, nor the purpose 
> of the link.  ie. It does not express any semantics whatsoever.
> ...
> Better choices would express something like the link originated as a 
> result of a user contribution and/or the document author/owner does 
> not implicitly endorse the link.
> ...

<a href="http://example.com/ " rel="competitor" endorsement="0">
<!-- Search engines would follow this link, just like they follow any 
link they can find, but they would not assume any endorsement. It could 
not act punitively; it would be equivalent to no link at all. -->

<a href="http://example.org/foopy " endorsement="0">
<!-- CMSes may insert endorsement="0" in markup for a link posted by an 
untrusted contributor. -->

<a href="http://example.org/foopy " endorsement="0" rel="disagreement">
<!-- Because endorsements are orthogonal to relationships, CMSes may 
even allow commenters to insert their own rel= attributes without 
affecting any endorsement. (They may possibly prohibit well-known 
values such as "up" and "home" that may give a false impression about 
site structure.) -->

<a href="http://example.org/foopy " endorsement="0.6">
<!-- Advanced CMSes may allow non-zero endorsement levels using their 
own values of trust for a contributor -- for example, how long the 
contributor has been registered with the site. They may even let 
contributors set their own endorsement= value for a link, multiplying 
it by the site's trust level for the contributor to produce the 
endorsement= value in the final markup. -->

<a href="http://example.net/barpy " endorsement="75%">
<!-- UAs should accept percentage values for endorsement= as well as 
decimal fractions. -->

I suggest this only as an example of what could have been -- a more 
honest and flexible alternative to rel="nofollow". I do not necessarily 
think it is useful, though, because it would not make any noticable 
dent in comment spam, just like rel="nofollow" won't. (Just as with 
e-mail, spammers will post in ever-larger quantities so as to find the 
unfiltered recipients, and eventually so as to achieve actual 
readership rather than just search ranking.) Nor do I necessarily think 
the barn door can be closed after Google and its fellows have bolted.

-- 
Matthew Thomas
http://mpt.net.nz/