1. Liz Lawley on October 9, 2003 8:48 AM writes...

It's not that I want all information duplicated across all systems. I want *core* information duplicated, though. I don't want to have to tell each system my login or contact information, for example, or where I live, or where my web site is.

In the face-to-face world we often present different facets of our "selves." I'm different in the classroom than I am at a club. But I don't have to repeat the introductions if I meet people in both settings.

2. Dan Brickley on October 9, 2003 11:17 AM writes...

Useful article. You're absolutely right this is one of the big design issues facing FOAF. I'm looking at a few ways to address this.

One issue for individuals to consider is whether different views/roles etc. of a person is adequate, versus having an absolute disconnect between various FOAF-described aspects of their lives. This varies between individuals and therefore isn't something that can be imposed as a one-size-fits-all approach across all of FOAF.

For the former, we could go into a space that, for example, offers alternatives to properties like foaf:img (which relates a person to their photo), such as something like foaf:work_image, foaf:fun_image. If a total disconnect is needed, things are actually a bit easier, since you just make sure there is no common identifying info between the two profiles/descriptions. So for eg., many Japanese FOAF files use a nickname (foaf:nick). The meinbild.ch site (http://rdfweb.org/topic/DataSources for links) also uses nicknames (sometimes a person's name, sometimes more creative) and doesn't expose any ID-ing fields such as mbox, homepage, mbox_sha1sum etc.

A 3rd design space (1st being addition of more contextually-oriented properties, 2nd being secrecy or at least diminished information to offset data-merging by aggregators) would be to treat FOAF documents as first class things that people could be related to. So you might say that I'm related to a doc foaf1.rdf by the relation foaf:professionalinfo and to foaf2.rdf by the relation foaf:nonworkinfo. That would involve taxonomising contexts, a non-trivial piece of design ('work' vs 'non-work' doesn't quite cut it...).

The tradeoffs between these different options aren't entirely clear yet. The secrets(and lies) option is deployable now. For the rest, we don't currently have differentiation by context in FOAF, despite (or because of) its importance. Your foaf:img is just a picture of you. There is no FOAF dating vocab, no 'foaf:hates', no FOAF CV stuff yet deployed, so the pain of this worlds-colliding effect isn't yet so urgent as it will become. Since FOAF documents were designed for independent extension, it's only a matter of time. Had interesting discussion w/ Marc Canter at lunch yesterday about all this...

Another important tangent here is characterisation of expected audience when publishing FOAF-based data. I'm happy sharing much with everyone, but
eg currently my cell phone info isn't world-visible, and my online banking details are very hidden
(in a PGP-encrypted FOAF file somewhere). Similarly, I happen to have political views which I don't mind sharing with the world (but don't currently in FOAF as the vocab isn't expressive enough). Others are not in this situation, either due to the context they live in or the nature of their views. How would different architectural approaches to context and facet description affect this? We don't know yet. Or same question, substituting sex for politics (not that the two are disjoint topics...).

If FOAF hasn't gone far in this direction yet, it is because the issues and options aren't clear, and there's a high cost associated with screwing it up. Suggestions welcomed!

3. Scott Rafer on October 9, 2003 12:31 PM writes...

I am doing some work with Feedster, who is adding a FOAF harvester to its existing scaled RSS gathering mechanisms. I have been wondering about "faceting," but I have been calling it "firewalling." More than not referring party pals to colleagues and vice versa, wouldn't we all rather not make it easy to surf in both directions? I think that visibility between facets needs to have traverse options that include 2-way, 1-way, and no!-way. How do you expect faceting to work in an open systems context where crawlers and search engines are out there gathering our (encrypted) FOAF data?

4. Seb on October 9, 2003 6:00 PM writes...

I think Scott is asking the right question. This wave is going to put increasing pressure on people to assume a single identity. Not only bad things may happen when the worlds connect.

5. Seb on October 9, 2003 6:59 PM writes...

Uh, that was a little unclear. I meant "Good Things may also happen when the worlds connect."

6. adrian chan on October 9, 2003 8:15 PM writes...

Goffman on his head, now that's presentation of self! This thread and comments go somewhere very interesting. But in Goffman's defense, I think he'd paint the picture with less contrast. Goffman's insights spoke in favor of a highly dynamic construction and presentation of self, one that unfolds in context and situation, framed by interactions and capable of numerous layers of embedding, reference, etc.
Oddly, faceting gets at this to some degree.
But i dont think that social software sites are each a fixed context. In fact I think they suffer from an inability to provide context. Rich context ought to allow us the chance to interact and play with the codes, know how things are going, test different interactions (flirt with the interviewer)... The trouble seems to me to provide context rich enough to facilitate that kind of interaction.
We dont own our identities, is Goffman's arg, as much as practice them.
So yes, we need more diverse ways of presenting and engaging ourselves, but we need more context too.
One site would be good enough if it were sensitive to modes of interaction/presentation...
Isnt that the reason Ryze throws parties? Because business networking benefits from a RL happy hour? (Or did I miss the point?!)

7. zephoria on October 10, 2003 4:23 AM writes...

Seb - i definitely agree that good things happen when worlds collide. But it's also important to consider who gets left out of that, who gets further marginalized by not fitting into normative assumptions of identity. Thus, balance.

Adrian - i adore Goffman and i definitely agree with your reading of him wrt ownership. The problem with the articulated version of identity is that it becomes disembodied, owned externally, not practiced. Also, while i think that the sites don't understand what their context is, people have created a context about the site, regardless of the designer's intention. (Do you know how many people i've interviewed who have no idea that Friendster is a dating site!?!?!?)

I don't necessarily believe that sites need more context; in fact, i think that they need less. I think that sites need more flexibility so that people can generate the context, not rely on the site. Think about this in terms of physical locations. I regularly visit a bar called "The Luna Lounge." Its vibe is different depending on the hour of the day or the day of the week; the people bring the social context. This private space gets transformed regularly by those who organize different nights.

More generally, take a public space that brings in a diverse group of people. For example, i doubt many of us have the same social assumptions about "Golden Gate Park" because it provides so much flexibility in terms of the kinds of social events that can occur. The trick is that the places that bring in the most diverse types of people are those that allow the people to define what the place means to them.

In order for sites to be effective at drawing in everyone, people need to be able to easily recontextualize the site for them and their friends.