[uf-discuss] Level of rel=contact

Tantek Çelik tantek at cs.stanford.edu
Tue Apr 6 18:01:26 PDT 2010

On Tue, Apr 6, 2010 at 5:02 PM, Sarven Capadisli <info at csarven.ca> wrote:
> On Tue, 2010-04-06 at 22:48 +0000, Brian Suda wrote:
>> On Tue, Apr 6, 2010 at 9:23 PM, Sarven Capadisli <info at csarven.ca> wrote:
>> > I'm thinking that rel="contact" is generally attributed to someone that
>> > we have at least a "lowest form of friendship" with.  [...]

It is not.

The reason we added "contact" in XFN 1.1 was to indicate literally
just what it says - that you have some sort of contact info about the
person - this could be just a URL or email address for example.

You can't assume any other semantics.

It doesn't imply anything positive or negative.  Thus you cannot
conclude at least a "lowest form of friendship" - you cannot conclude
anything about friendship from rel="contact".

If you want to imply a "lowest form of friendship", then
rel="acquaintance" is what you're looking for:


  Someone who you have exchanged greetings and not much (if any) more
— maybe a short conversation or two. Often symmetric.

>> > Additionally,
>> > if the user doesn't have control over the declaration of such
>> > relationship, wouldn't it be more meaningful and safer to exclude this
>> > bit of information in the output?
>> --- You lost me on "exclude this", exclude what exactly?
> My bad. My reference was to the example.

Again, this is precisely one of the reasons why "contact" does not
imply anything about friendship, so that you don't have to worry about
"excluding" it due to  an imagined implication.

>> > The example I had in mind was 'Subscribers list' at
>> > http://identi.ca/csarven
>> --- if you are subscribing to someone, then it probably at minimum
>> meets the definition of: someone that we have at least a "lowest form
>> of friendship"

I don't think you can make that assumption. subscription !=
friendship.  You might be subscribing to an automated summary
aggregate feed for example.

For things like "subscribing" e.g. Identi.ca or Twitter followers or
followings, there's been quite a bit of brainstorming over the past
few years:


which appears to have converged on:

rel="follower" (for indicating someone who is a follower of yours)

rel="following" (for indicating someone who you are following)

On such services that also permit "direct messaging" - these URLs also
serve as a form of contact information, and thus rel="contact" could
also be used:

rel="follower contact"

rel="following contact"

Since direct messaging is not necessarily symmetric (neither is rel
contact), it might make sense for such a service to label a link to
another profile as a contact only if you do actually have the ability
to contact (dm) them - though that might also be asking too much of
the semantic of "contact" (since "has contact info" and "can contact"
are two different things.)

>> Are you suggesting it isn't and we should exclude it?
> No, I'll clarify. What I was trying to say was that, if I have a profile
> page where it lists a bunch of people that are subscribed to me, I
> wouldn't necessarily call them my contact since I don't really know
> them. Hence, in my example at http://identi.ca/csarven , rel=contact
> should be removed from Subscribers list. I agree that rev="contact"
> makes more sense here, but, I'm focused on the incorrect use of
> rel="contact".

Why is it incorrect? It's only incorrect based on the assumptions
you've presented about what rel="contact" means - which is basically a

> rel=contact is/should be reserved for people that meets the basic
> requirement of that "lowest form of friendship".

Why? We already have rel="acquaintance" for that semantic.

Would it help to add any of this discussion to the FAQ?




More information about the microformats-discuss mailing list