Clay Shirky
( Archive | Home )

Liz Lawley
( Archive | Home )

Ross Mayfield
( Archive | Home )

Sébastien Paquet
( Archive | Home )

David Weinberger
( Archive | Home )

danah boyd
( Archive | Home )

Guest Authors
Recent Comments

pet rescue saga cheats level 42 on My book. Let me show you it.

Affenspiele on My book. Let me show you it.

Affenspiele on My book. Let me Amazon show you it.

Donte on My book. Let me show you it.

telecharger subway surfers on My book. Let me show you it.

Ask Fm Anonymous Finder on My book. Let me show you it.

Site Search
Monthly Archives
RSS 1.0
RSS 2.0
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline


« backchannel modes | Main | Jonathan Abrams at SXSW »

March 16, 2004

RELATIONSHIP: A vocabulary for describing relationships between people

Email This Entry

Posted by Clay Shirky

Behold RELATIONSHIP, a vocabulary for describing relationships between people. I don't know if I'm the one to shoot these particular fish in this particular barrel, since both mme. boyd and Herr Weinberger are more eloquent than I on the subject of of making the tacit explicit, but this thing is self-critiquing. Here, just in case you were wondering, is how you should be characterizing your relationships with one another:
friendOf, acquaintanceOf, parentOf, siblingOf, childOf, grandchildOf, spouseOf, enemyOf, antagonistOf, ambivalentOf, lostContactWith, knowsOf, wouldLikeToKnow, knowsInPassing, knowsByReputation, closeFriendOf, hasMet, worksWith, colleagueOf, collaboratesWith, employerOf, employedBy, mentorOf, apprenticeTo, livesWith, neighborOf, grandparentOf, lifePartnerOf, engagedTo, ancestorOf, descendantOf, participantIn, participant
Describing relationships with a controlled vocabulary can sound credible right up to the moment you see the vocabulary, but this thing is a mess. Consider the world of work: They've given us worksWith, colleagueOf, collaboratesWith, employerOf, employedBy, mentorOf, and apprenticeTo, but take any moderately complex real-world work relationship of yours and try to fit it here. We start off with employerOf/employedBy, models of clarity, but what if you are employed by a colleague you collaborate with? Or, more realistically, what if you would call someone your boss in one setting, colleague in another, collaborator in a third, and say they are someone you work with in a fourth? And what if you cannot specify in advance which of those words you would use in any given situation and yet, _mirabile dictu_, you are not at a loss for words when those situations arise. The whole list is like that -- we get friendOf, then for a semantic richness bonus, closeFriendOf. But if we're going that route, where's veryCloseFriendOf? sleepsWith? usedToSleepWith? Where's wentToHighSchoolWith? Take the relationship closePersonalFriendOf. The designers of this list somehow overlooked it, possibly on the grounds that it's tautological, and only of use on talk shows. ("Oh yes, Julia Roberts is a close personal friend of mine.") But it is nevertheless informative -- you would only use closePersonalFriendOf if the person in question was someone of relatively high fame or station. In addition, anyone claiming to be a "close personal friend" of someone else is talking about a domain where a high degree of social interaction is the norm, e.g. show business. By extension, the seemingly oxymoronic friendYouDontLike is also a valid category, as anyone in highly social environments can tell you. (You often run into friendsYouDontLike at partiesYouHaveToGoTo.) So here's the dilemma -- the point of a controlled vocabulary is to de-thesaurisize. Instead of one doctor offering a diagnosis of manic depression and a second opinion of bi-polar disorder, a controlled vocabulary says "If everyone uses bi-polar, linguistic interoperability increases, and we'll get better sharing of diagnoses." No such luck with human relations, however. Manic depression and bi-polar disorder are both labels for an external condition. Human relations have the additional and curious property of changing the relationship through the act of labeling, and anyone who has ever said "I love you" can attest. So either we add closePersonalFriendOf and friendYouDontLike (and sleepsWith, usedToSleepWith, wentToHighSchoolWith, gotArrestedWith, etc.), or we say "No, you can't say wentToHighSchoolWith, that's stupid." But once we go the route of disallowing certain characterizations, we are destroying real-world meaning in the name of controlled vocabulary, a strategy that is internally consistent but useless. The RELATIONSHIP list should make it obvious that explicit linguistic clarity in human relations is a pipe dream. It probably won't though -- the madness of the age is to assume that people can spell out, in explicit detail, the messiest aspects of their lives, and that they will eagerly do so, in order to provide better inputs to cool new software.

Comments (24) + TrackBacks (0) | Category: social software


1. Peter Caputa IV on March 16, 2004 11:05 AM writes...

My gut says to agree with you on this one. However, for a slightly different reason. And I still am not sold either way on relationship taxonomy.

I still have these questions:
What is the value to anyone to explicitly define a relationship with another person? What is the value of making it public to that person? Is there value to the network operator for data mining/marketing purposes? Is there value to the greater network?

Although taxonomy is difficult to get right, for the aforementioned reasons stated perfectly by Clay, it could be worthwhile for some people.

Permalink to Comment

2. Seth Gordon on March 16, 2004 1:49 PM writes...

If only the vocabulary included hasSleptWith and hasSwappedSpitWith, one could use it to create sex charts.

Permalink to Comment

3. Adam Rice on March 16, 2004 3:58 PM writes...

I thought this was a joke until I clicked through the link.

Then I was certain it is a joke, only the teller doesn't realize it. I like Seth's idea though. It would be nice to literally map out the old truism "when you have sex with someone, you're having sex with everyone they've had sex with."

Permalink to Comment

4. Danny on March 16, 2004 4:11 PM writes...

A lot of your "...what if you are employed by a colleague you collaborate with?" type situations can be handled by the RDF/OWL models (in this example you'd just state all three). Such statements could potentially get pretty complex, especially when you start bringing in material from other vocabularies to provide context.

But certainly it is messy and difficult, and I think the best we can hope for is an approximation that is good enough to work with "cool new software".

But what exactly are you offering as an alternative? If you're saying that human relationships simply can't be described in a machine-readable form, doesn't that make "Social Software" the oxymoron?

Permalink to Comment

5. Clay Shirky on March 16, 2004 4:47 PM writes...

To paraphrase jwz: "A lot of people, when they have a problem, say 'I know, I'll use an ontology.' Now they have two problems."

OWL doesn't help because no ontology fits here -- the emotionlanguage mapping is imprecise.

And there's no need to describe things in a machine-readable form for them to work -- much social software is in fact merely a conduit for sense-making between people at the edges.

Consider the lowly photo -- it is, so far as Friendster or Orkut are concerned, merely a piece of binary data attached to one user profile and shown to others. And yet the things we can glean from a photo make it an essential piece of those services. (Cue Lawley: It's the faces, stupid.)

The kind of value in photos -- value that passes through the system from user to user, without the intermediary software understanding anything -- is a key part of much social software. Its an amplification of, rather than a replacement for human social cognition.

Permalink to Comment

6. Seth Finkelstein on March 16, 2004 5:28 PM writes...

Oh, come on - "ambivalentOf"? That's got to be a joke.

But we do need "enemyOf" and "antagonistOf" in Social Software implementations.

Permalink to Comment

7. Ian Davis on March 16, 2004 6:53 PM writes...

For some reason trackbacks don't show up here. Anyway I have a short response at

Permalink to Comment

8. Ross Mayfield on March 16, 2004 7:19 PM writes...

That does it, Clay! You are no longer my friend.

Permalink to Comment

9. Seth Finkelstein on March 16, 2004 8:26 PM writes...

Waitaminnit ... it wasn't a joke?!

Where's "hasSleptWith"?

Yes, of course, you don't need "hasAttendedAtLeastOneConferenceInThePastDecadeWith".

But the levels of granularity are very strange if you were serious.

Permalink to Comment

10. bardia on March 16, 2004 8:59 PM writes...

Clay's post is interesting in another aspect. I like Clay and have read a number of his posts and know that his opinion counts in many circles. That brings me to my point about trust and reputation in social networks. Just an example of future problems with relying on reputable people. Reputation is dangerous. Everyone is human, after all.

I have been lurking around the FOAF list for a while and know these guys have been at it for a long while. The questions raised by Clay have been brought up and debated to death by a number of smart people (OK, maybe not as smart as Clay :) )

Then here comes a reputable guy, who probably has not been involved in the effort, who without really knowing the issues, "fires a rocket at it". This rocket causes a lot of waste of energy because it is just a basic instinct kind of a rocket. Not really thought through.

I wish we could get to a place, sooner, where the reputation and credibility of one person would not have that much effect.

In one of earlier Clay's posts about the Howard Dean movement (or maybe one of the replies), a person was quoted as saying: I like everything about the Howard Dean movement, but Howard Dean. Back to the problem of one person being the single point of failure.

Permalink to Comment

11. Tim on March 16, 2004 11:39 PM writes...

Reading the previous comments makes me believe there should be a taxonomy for 'newsgroup attitude'.

It isn't the idea of having a vocabulary to describe the relationships between people that is flawed. It's the implementation.

Perhaps what's required to fix the apparent problems in the implementation we are ranting at here is to develop interoperable ontologies to take related classes such as emotion, relation, location, etc. to create nodes based on the moment in time the 'state' of the relationship existed.

Of course I could just be whacked from eating too much pizza, watching too many movies too late into the night and sitting with the blood pooling in my thighs for far far far too long...

What was that you said about relationships?


Permalink to Comment

12. stavrosthewonderchicken on March 17, 2004 12:43 AM writes...

Although it's perhaps stronger on kinship relationships than otherwise, the Korean language is richer than any other I've ever been exposed to in terms of relationship words. It's worth investigating, for those interested, to get an idea of the sort of nuance that in English is so difficult to express.

A reference work is mentioned here.

Permalink to Comment

13. William on March 17, 2004 1:59 AM writes...

Perhaps this is somewhat simplistic and indeed I'm assuming the ontology was created with this in mind, but where are the statistics to show the most important (read useful) relationships?

Obviously certain relationships, childOf, worksWith, spouseOf, are very common, while others, wasArrestedWith, are not. I personally can say my wasArrestedWith would have value 0, whereas my wantsToSleepWith would be very high. But seriously, applying probability over a population would bring the useful relationships out in the open while pushing the minor ones toward the catchall, hasRelationshipWith. Members of this last group would naturally migrate out into other relationships or be deleted over time.

Permalink to Comment

14. Jay Fienberg on March 17, 2004 2:01 AM writes...

OK, so maybe you missed my post on this from three days ago:

I think, once knowsBiblical (defined as "knows in the biblical sense") is added, it's all good.

Permalink to Comment

15. leo on March 17, 2004 6:29 AM writes...

@Bardia | about: FOAF

actually, the rocket is quite ok. It looks like Clay is not involved in doing the vocab, or ?

But actually the vocab is a child-vocabulary of FOAF. So the authors respected the FOAF results and made their stuff compatible with it.

if it is such a big success as FOAF, time will show.

Permalink to Comment

16. Danny on March 17, 2004 7:22 AM writes...

You say "OWL doesn’t help because no ontology fits here — the emotionlanguage mapping is imprecise." - but human languages are imprecise when describing relationships, yet we can still talk about them. Just because something is imprecise doesn't mean it isn't useful. A sketch map is usually better than no map at all.

Let's say I'm in Orkut, and browsing through friends-of-friends, and there all of a sudden I see a photo of the man that ran over my dog/screwed my wife/stole my car. Result - one broken monitor. Wouldn't it have made sense to use a machine representation of my relationship to that person (e.g. enemyOf) to prevent this?

I take your point about the value of software systems as a conduit, but this doesn't exclude their use as *computers*. There are other ways ways in which human social cognition can be augmented.

Going back to the photo example, there are already systems in place to embed metadata in digital images (e.g. all Adobe's tools now insert RDF into the files), why not use these tools? The photo gets stamped with the name of my arch rival, my FOAF profile expresses our mutual loathing, the social software prevents conflict.

Permalink to Comment

17. relaxedguy on March 17, 2004 10:24 AM writes...

That is a good start. Don't get bogged down in the semantics otherwise nothing will ever get done. The system should not care what the relationship characterization is, LooksLike, FriendOf, etc. It should find the commonalities such as LooksLike, FriendOf or whatever and act on those. Stuff it all in your FOAF, and so often roll out extentions to the vocabulary. People go to a website to update their FOAF, get an alert that there are new relationships to choose from, and choose to integrate into their FOAF. If they don't, the system shouldn't care.

Most people won't care who I'm related to, and I certainly don't care who they have slept with. Focus on extending the uselessness of "I have 287,673 friends" and add some finer grained relationship context as you go along.

Valdis Krebs has done some interesting entity relationship modeling that might shed some light on this discussion. I'm a huge fan of graphically representing relationships, hurry up and get this stuff figured out so I can build my front end.

Permalink to Comment

18. Jeremy on March 17, 2004 12:30 PM writes...

"the madness of the age is to assume that people can spell out, in explicit detail, the messiest aspects of their lives, and that they will eagerly do so, in order to provide better inputs to cool new software."

I think the assumption that people would (if they could) is actually an increasingly safe assumption. The madness may be not the assumption that people will do it, but the fact that people will do it.

Permalink to Comment

19. Tim on March 17, 2004 1:42 PM writes...

Well, considering how many times I have read about wanna be relationships such as wantsToSleepWith I wonder if a relationship isn't really defined relativistically.

How do you actually benchmark what is the subjective nature of a relationship to an individual who doesn't share the same emotional quotient as you do?

Perhaps mapping relationships will help people to understand and place appropriate attention on them. I'm put in mind of 'Gosford Park', a Robert Altman film containing a murder but really about the relationship between people of different classes.

In this upstair/downstairs drama the victim is the central to all manner of 'feelings' by virtue of his impact on members of both classes. His presence is felt in the after effects of his previous actions.

This only tells me that timing is important in relationships along with what happens during the course of them.

If you have no emotional quotient and are contented to be walked on then your relationship with the person I might find offensive is going to be different than yours or someone elses.

However, my instinct may be to keep this person in some relation to myself there may be external forces at work that would be prudent to acknowledge. Such recognition would then have to be given to conditions such as traitor, excommunicated, parentCan'tStand, etc.

So if the mapping itself is subjective how do you arrive at translating that to an interface to the language and them represent that language to the reader?

Will mapping enforce some sort of moral code as well especially if the language contains only a subset of what would represent the entire spectrum of human to human relationships?

Perhaps the absense of language to describe subtleties will leave room for the heart to grow fonder. But maybe only if you let the software do the thinking for you.

Of course I am thinking here of an advanced language implemented to 'manage' relationship mapping that would also be dependent on persons being mapped to record the events of their lives in real time and have those encapsulated events interact with the relationship mapping I maintain on you, for instance.

Wow. Creation.

Permalink to Comment

20. Seth Finkelstein on March 17, 2004 1:47 PM writes...

stavrosthewonderchicken: Looking at that Korean blog post, it reminded me of the English words "sugar daddy". This expresses a relationship, but it's business, not family - it just plays off the family word. The woman's "husand" can also be her "sugar daddy", this just expresses two different connections to the same person.

Anyway, even though human relationships are complex, so is real-world physics, but frictionless pulleys and massless ropes still have their place in modelling.

A problem though, is that so far, most of these models aren't very useful.

Permalink to Comment

21. johne on March 17, 2004 2:58 PM writes...

“the madness of the age is to assume that people can spell out, in explicit detail, the messiest aspects of their lives, and that they will eagerly do so, in order to provide better inputs to cool new software.”

The madness goes deeper than that. One example is the legal system as it's presently implemented in America, where it is assumed in trial that a narrative constructed of the memories of appearances and verbal exchanges, with externalities such as e-mails and and surrounding circumstances, can be relied upon to settle momentous relationship questions.

Permalink to Comment

22. William on March 17, 2004 2:58 PM writes...

Since relationships are two way streets, how about some logic in place that can merge relationship mappings, when both sides are known.

If I have a wantsToSleepWith relationship with X, but X has a willNeverEverSleepWith relationship with me (and I know about that relationship), then our two way relationship is wantsToButWillNeverSleepWith. Similaryly, a childOf who has been disowned would be a formerChildOf.

Permalink to Comment

23. Bill Seitz on March 19, 2004 10:09 AM writes...

What about having a single relationship container, then having for each relationship:

* some sort of date-range info

* free-text description of the relationship (perhaps including some sort of Rating info for ReputationManagement? Again, it doesn't have to be numeric...)

Permalink to Comment

24. Don Park on March 24, 2004 4:49 PM writes...

I tell ya. It must be the water. Another good pile of geeky nonsense.

Permalink to Comment


TrackBack URL:

Listed below are links to weblogs that reference RELATIONSHIP: A vocabulary for describing relationships between people:


Email this entry to:

Your email address:

Message (optional):

Spolsky on Blog Comments: Scale matters
"The internet's output is data, but its product is freedom"
Andrew Keen: Rescuing 'Luddite' from the Luddites
knowledge access as a public good
viewing American class divisions through Facebook and MySpace
Gorman, redux: The Siren Song of the Internet
Mis-understanding Fred Wilson's 'Age and Entrepreneurship' argument
The Future Belongs to Those Who Take The Present For Granted: A return to Fred Wilson's "age question"