Running notes from Revenge of the User: Lessons from Creator/User Battles danah boyd, U.C. Berkeley http://conferences.oreillynet.com/cs/et2004/view/e_sess/4948 at the O'Reilly Emerging Technology Conference: http://conferences.oreilly.com/etech/ 11 February, 2004 San Diego, CA by Cory Doctorow doctorow@craphound.com -- What are users doing in Social Software networks? I've been looking at what users actually do. I'm a computer scientist who got involved in visualizing social networks and then got into studying Friendster and then YASNSes (Yet Another Social Network Service -- Shirky). Researchers have been looking at meatspace social networks for a long time, and have a well-developed theory of how people get jobs, dates, etc. Geeks tried to systematize this, make it more efficient, faster and easier to realize an opportunity. The models don't map 1-to-1 onto meatspace, and the disconnects are interesting. I'm looking at explicit networks, where you say who your friends are and connect directly. Mostly focusing on Friendster, since I've spoken to > 400 users at this point. When sociologists ask people to private disclose their social networks, they do something different from what technologsts do when they build YASNSes. Users took the work of geeks and changed it to suit their needs. -- Social theory: we assume that face-to-face architecture underlies YASNS. Kapor: Architecture is politics, Lessig: Code is law, code is architecture. When we model new architectures, we don't directly translate social theory. -- There are real-life spammers (telemarketers), and we can structure our lives to avoid them. We walk around the city to avoid people not like us: don't want to see a porn shop? Stay off Market St. When we scale up social networks, we put all the people -- including the people we don't like, into the same room as us. -- Some fallacies: Six Degreess of Separation: Millgram was talking about small worlds -- how many hops to get a pakcage from NE to MA? This didn't lead to the conclusion that everyone in the world can be connected in six degrees. This isn't a top-down model. When you need to get a package from NE to MA, you don't know who is closer than you to your end-point. We don't know how many degrees separate us. -- 150 people in our networks: Dunbar said monkeys had finite networks of monkeys they groomed. Dunbar said people gossip about 150 people -- doesn't mean we only know 150 people, but that that's the number of people you know at most. -- Vandevetter: Weak times. People find out about jobs through weak ties (friends), not strong ties (people you'd turn to in a major crisis). People you peripherally know. A FOAF is not necessarily a weak tie. A friend of a weak tie isn't necessarily a friend at all. Your Rolodex is weak ties, not THEIR Rolodexen. Vandevetter: People find jobs through weak jobs, but they don't get jobs through weak ties. You get jobs through context, not just strength. Say you party with someone who's dating someone who runs a tech company with a job opening. He'll happily tell you about it, b/c it gives him power as a bridge. But that doesn't mean he'll get you the job: he won't risk his rep with his boyfriend in order to get you the job -- he'll just say, "He's a club kid I know who might be good." Strangers aren't good people to set you up on dates. -- The weirder and fringier you are, the tighter your friendships with co-weirds will be. Your best friends' best friends might be your best friends. Ties don't always transition: knowing techy-Bob's hairdresser Anne doesn't tell you much. -- Compare the theory to the sites: Friends on YASNSes aren't close ties: they're acquintances with things in common. You can't tell if there's transitive trust there. Why would I want to date my hairdresser's brother's drug dealer's second cousin. Asking favors is different from offering them. Offering a favor gives you power. Being pressured to perform the favor makes you uncomfortable. On LinkedIn, where you know that person a is your conduit to person b, it puts pressure on person a (which makes sense if the relationship is transitive, but not if the next hop connection isn't work-related). The people on the YASNS aren't your friends, they're your para-friends, your "friendsters." -- There are significant consequences to articulating your social network. Try rejecting your boss's Friendster overture of friendship. Your friends say, "I thought you hated your boss, why's she listed as your friendster." People are bad at rating and systemizing their friendships. Ranking a friend's sexyness from 1-3 is hard. Who's a 2? Relationships aren't bi-directional. If I know everything about Angelina Jolie, it doesn't mean she knows anything about me. Your shrink knows about you, but it's not OK to ask her about her life. -- These networks don't feel real. They are socially awkward. They expose us to people we wouldn't know in real life. You might invite your jazz-obsessed work buddy to a jazz club some night, but you get to choose when that happens. Why are my drinking buddies asking my boss out for a date? -- A teacher's 16-y-o students popped up on friendster and found her clubbing buddies, one of whom had a ton of testimonials saying, "You like little girls," (he dates 22-y-os), and the teacher's students wanted to know why she's pals with paedophiles. She can change her profile, but her friends all still look like freaks. Contexts allow us to selectively display facets of our personality, but YASNSes take that away. -- Who do you link to? Friends? People you recognize? Anyone who invites you? How do you present yourself: if the people who invite you on is from Burning Man, you might present as a burner. But when non-burners sign on, you're in trouble. -- The Fakesters came on. Not just bad people. Profs killed their existing Friendster accounts and created fake accounts to link to their pals. Or they tried to create focii -- an account for "Burning Man," etc. When Friendster killed the fakesters, it turned into a whack-a-mole against people who wouldn't give up, who had a manifesto. "Why did I have to pretend it was real? Why couldn't I do it for fun?" People even started dating their fakester buddies, fulfilling the purpose of the site. -- Neonazis have had a grand time on Friendster, contacting people of color over and over and telling them to get out. Coke dealers use Friendster to connect up with buyers. -- The response is an attempt to "configure the users" -- constrain behavior to acceptable behavior with messaaging, kicking people off, etc. This won't work: you can't tell a hacker not to hack. These kids are social hackers. You can stop some bad behavior, but you chase off your best users, too. Dating doesn't happen because you're in a dating context. Dating arises out of real contexts. Taking away fakesters didn't make Frienster more real. Friendster is unreal because people never remove their friends, even if they never see them (the exception is when you break up, ironic, because ex-lovers are strong ties!). -- We're not learning from out mistakes. Orkut repeats Friendster's mistakes, such as ranking people by their number of friends, which creates the game of accumulating friends. -- How do we fix this? 1. We can pretend that these networks model real networks and keep the VCs happy. 2. We can model actual social networks and sacrifice some business models Social behavior doesn't have a technical solution. -- Call to action: 1. How do we create a way to allow nuanced social interaction without unbearable awkwardness? These sites are meant to make life easier, but they're making life harder. It's a hassle to organize and negotiate your YASNSes. People are getting YASNS-exhaustion. 2. How do we let people show the appropriate face to the public? We talk differently to co-workers, friends, etc. LiveJournal solves this best by letting you determine who sees which posts. Kids do this all the time, excluding their parents from private posts. 3. How can we let people gracefully opt out of this? How do we protect people from hate attacks? 4. Architecture should allow regulation through social norms. MUDs and MOOs tried to do this and they failed. Social norms don't always work in everyday life, but the systems constrain the way we interact. eof