SXSW 2005 Panel: Leveraging Decentraized Social Networks

Tuesday, March 15, 2005

[This is a (mostly complete) transcript of Tantek Celik, Jonas M Luster, Joyce Park, Ernie Hsiung, and Danah Boyd talking about leveraging decentralized social networks. Official details about the panel here.]


Does “social networks” mean anything when pretty much everything that goes out of the computer is a part of a social network in a sense?

Does “why” matter? Why do people put so much information about their lives online to possibly be archived for ever. Maybe it doesn’t matter why. Maybe we can just build on their behaviors and build products that meet their needs.

Who owns your data? AIM claims to own everything you write through their chat program, for example. [He only alludes to this with a cough.]

Can technology be too simple? Or are we making it simple enough as it is?

Luster: Let me start with a story. In 1993 I was in Somalia. I asked an old woman there who else was coming out there. She said “experts.” Turns out they don’t have a word for “expert.” So they use the word to mean someone with a suitcase who talks a lot and doesn’t know anything. I used to be an academic, but I’m not at “expert.” I’m looking at the question, “Can technology be too simple?” I’ve been looking at social networks since 1992, 1993, before anyone else was thinking about it in relation to a computer. [Shows screen with six words that define what makes a social network.] My first though looking at Friendster is that it’s not a social network — it’s a community site. When we talk about social networks, we’re not looking at the attributes of the person — we’re looking at the attributes of the relationship. My three favorite books are not a social network analysis. The group of people who like those books is. Proximity. It’s a factor. […] Time. On Friendster you can find people who are friends with lots of other people. The question is when and has that relationship changed over time? These are not answered by looking at a modern social network application [like Friendster]. Direction. Orkut is bi-directional — requires both sides to consent to a connection. In “old school” social networks it can be uni-directional. Density. Type. How did I meet the other person? What kind of social network? So what’s in a social network? More than just one variable. The networks are transient. We look at Orkut and see people are coming in and going out. Do they use well-definined vocabularies or free declarations? These are interesting to me because they are overlay networks, something much deeper than normal networks. Trans-communal areas.

Celik: [Graphic on the screen.] Here’s a graphic of the XFN web on Flickr.

Park: I built Friendster and then got fired from Friendster then worked on ePinions. I’m interested in the tension betweeen openness and privacy. Tantek asked why do people put things on blogs. I’ve been interested in whether people understand the consequences of putting something into a fully public realm. One big difference between the genders I’ve seen is that women have greater needs and desires for safety [and a few other things that she listed quickly]. For instance I was talking to someone about She said, “I’d never use that — it’s a stalkers wet dream.” And I think the best use would be for call girls. Another example. I was talking to a man about what social software product he’d like to see. He’d like Outlook for his social life so friends can schedule to have dinner with me, for example. So he his wife and she thought it was crazy — half the time she lies to people about why they can’t get together. The little white lie would go away. One last anecdote. I was talking to a woman whose uncle I know. She didn’t want to use social networking sites until she knew her uncle, work friends, and real friends couldn’t see the same data about her. Most social network sites have been built for young men by young men. And what do young men like to do? Look and lots of photos of good looking women whom they don’t know but might like to know. And to stalk them. So can we use this open, public, un-private system to fit these real needs.

Boyd: Hi. I’m a PhD student and an ethnographic engineer at Google. I’m interested in the questions of “why” and “who.” How people engage with networks. How they’re transformed by technology. What does it mean to be “open?” Put in the public. Open. Transparent. Utopian ideal that transparency will solve all of the world’s problems. But not everyone has equal access. There is freedom in walled gardens for many poeple. 1995 I created a website for Ani DiFranco fans with lyrics. She attracts 13-15 year-old queer girls who are often victims. So I created the site and I attracted those girls and they wrote about these troubles. And when I created a weblog, they went there. One problem with having a large audience is that they’re not now safe. On a big site, people sometimes attack them. They’re not as free. And even services we don’t think of as closed can be. Gay men thought everyone on Friendster was gay. But it’s not. In a homogenous group, you’re okay. So who’s values are being served? If you go to Golden Gate Park, you’re in the public. But we still have a notion of audience if I’m talking to someone in the park. I can talk to you about Friendster without explaining. So what’s lost when you’re dealing with diverse audiences that don’t know what Friendster is. So when we build these systems we have to think about whose values are being served. TechnoSocial problems: Social awkwardness. We don’t know how to say “no” to “are you my friend,” for example. Articulation problems. Try describing yourself. Does that match my description of you? No. We don’t know how to articulate this. Favorite music? Hard. The problem with the public. I heard TypePad is 50% private blogs. Not everyone wants to speak to the world-at-large. I’m an academic. I deal with groupware and that stuff. We’ve not actually built anything that new technology-wise, but we’ve built something socially new. And one of the best way it’s changing things is that people are throwing technology out there and is makes social change. The best social software evolves along with the social group. Friendster didn’t evolve, now it’s not really used. MySpace does. So does Flickr.

Hsiung: I’m probably the least academic of the people on this panel. But I do have a little experience. Had a blog in 2000. And am now a web developer at Yahoo! in their communities group. I look at the user types that get into these services and things like that. So I’ll come from a user-advocate perspective. Personas. Who are the types of users who would use different sorts of services. Yahoo! Photos. Used mostly by people like new mothers. So. XFM. Stands for XHTML Friends Network.

Celik: Parodying the Psychic Friends Network.

Hsiung: So the idea is to tak eht eida of social linking out of sites like Orkut and putting them on your own site or, in this case, a weblog. It comes with assumptions. You needa weblog and a blogroll. If you know HTML and have the blogroll, you can use a “rel” attribute in your “a href” to indicate this. So if I have a blogroll, I would like to Tantek’s blog with a “rel” attribute that says something like “met friend muse” etc. [Pulls up a screen showing pre-defined terms like “colleage” and “crush.”] For additional functionality developers would have to make tools to use it. For now it’s just a framework. There are some basic apps like In the same way like Friendster. The difference is that you’re on your own site and that where the whole idea of decentralization comes from. So bloggers are all about ego. When you invest all of this time in your blogroll and throw in all of these keywords from your friends. Then nothing happens. You go to RubHub and see your friends. Great.

Luster: I use a little CSS-fu to put an asterisks next to any name on my site that uses these XFN tags.

Hsiung: So. It has a lot of potential, obviously. But there’s not that much functionality. Right now it’s a meme. There’s not that much functionality built ot do stuff that’s cool — er, superfunctional.

Celik: Matt in Metafilter added XFN support. So then when you’re looking at a story, you can look at comments made by people in your social network.

Hsiung: It’s not difficult to make applications that use this. Right now, it’s just a need thing. If more people use it, it will build.

Luster: One thing I do and you’ll be able to do with Mac OS X Tiger. I’m extracting all the rel= attributes. If the person’s not in my address book, it’s add them with their URL or whatever. Goes into Dana’s argument — how much do I actually want to expose to others?

Question: If there was something you could do with verifyable anonymity… How would that affect social networks?

Luster: There’s a difference between implicit and explicit knowledge. Most social networks are not very explicit. It’s implicit. What does a hug mean? It’s implied. If we look only in the declarative space and trust everyone to be explicit, this might work. In the implicit place we’re in, this wouldn’t work.

Boyd: Anonymity is not possible. You can be short-term anonymous, but the first time someone points to you and says, “that’s Dana,” your anonymity is broken forever. And you can fingerprint people based on how they write.

Luster: Maybe we could make something that changes how you write. Like the Swedish chef. Bork Bork Bork!

Boyd: There are a lot of bloggers that claim to be public, but also use a lot of layered, coded information. My blog. People think it’s a professional blog. But my friends get other things out of it.

Celik: That’s a good point about anonymity. There’s a broad spectrum of privacy. And that’s totally natural. Where you feel you fit on that spectrum is totally up to you. Sometimes you’re not going to be able to make a swiss army knife.

Luster: One quick point. The question of anonymity. It’s come up about three weeks ago. Someone wrote how the word “folksonomies” sucks. So I tracked how the word spread from one paper to all around the world. Perpetuated over and over and over. By looking at little designators like this with a clear, distinct meaning I can see who reads who, who adopts from who. I can read these relationships and a few other indicators to very effectively break anonymity. And I can break information blocks put up against me.

Boyd: You’ve heard the magic number 150. When monkeys groom, they build social networks. That what we do by gossiping. Keeping up with people’s likes. 150 is the maximum number of people who can cognitivily keep track of. More is overloading. And the people ebb and flow.

Question: These are great tools to show connectivity, but what about reputation? This person is smart. This person is full of crap. And such.

Celik: Take a look at VoteLinks.

Boyd: Look at eBay. Very small percentage of negatives. They have to really piss you off. Most people arent willing to say “I don’t like you” publically. So when are we willing to be negative and what are the consequences.

Question: I’m curious about internationalization, such as the Brazilian network of Orkut. They seem to be using it differently.

Hsiung: From what I know, talking to Randy Farmer, one of the reasons Orkut is so focussed on Brazilians is because it’s still invite-only. The early techie adopters got tired of it. But the chain of friends began to focus on Brazil. As more and more came, they kind of crowded out the other members.

Boyd: How many of you know why? The dumbest thing. For the longest time on Orkut, they had a bunch of country flags. Looked like the World Cup. What do you do? Beat the other countries! So that’s what they did. Sign up your friends, we’ll beat the Americans. That sort of thing. And they had fun with it and it met their needs. Only recently did people begin to really move from city to city so people needed new ways of creating social networks. So it made sense. They were already doing it and they made sense with it. Culture and society coming together. And Google keeps trying to meet their needs.

Luster: There’s an amusing book called the Cowpatty Ecosphere. Cows graze. They crap. More grass. More cows. Loop. Same with social networks.

[Screen: Brazil: 65%. US: 10%. Iran: 7%. Pakistan: 3%.]

Question: Speak to unintended social consequences.

Celik: One word.

Hsiung: Research.

Boyd: Hegemony.

Park: Subpoena.

Luster: I had the same word.

[Note: I might alter this transcript in the coming days to remove typos and clarify the content. I post it now just to make it available as soon as possible.]