Friendster and Publicly Articulated Social Networking danah michele boyd University of California, Berkeley - Information Management & Systems 102 South Hall, Berkeley, CA 94720-4600 [email protected] ABSTRACT This paper presents ethnographic fieldwork on Friendster, an online dating site utilizing social networks to encourage friend-of-friend connections. I discuss how Friendster applies social theory, how users react to the site, and the tensions that emerge between creator and users when the latter fails to conform to the expectations of the former. By offering this ethnographic piece as an example, I suggest how the HCI community should consider the co-evolution of the social community and the underlying technology. Author Keywords Friendster, social networks, configuring users, identity ACM Classification Keywords H.5.4 Group and Organization Interfaces INTRODUCTION Many social applications have social networking embedded both implicitly and explicitly in their design. Through features such as Buddylists and Blogrolls, developers and users have recognized the value of social networks. Recently, this implicit networking has evolved into explicit effort as entrepreneurs seek to capitalize on the social networking theory. Although explicit social networking sites have existed for years (SixDegrees.com), recent commercial interest has resulted in the emergence of multiple new sites dedicated to helping people capitalize on their social networks for jobs (Ryze.com, LinkedIn.com), dating (Friendster.com), recommendations and listings (Tribe.net). While all of these sites are valuable in this domain, Friendster’s popularity, press coverage, and diverse usage make it an ideal candidate for studying the value and implications of this phenomenon on the HCI community. In this paper, I present portions of my ethnographic work on Friendster in order to consider the tensions that emerge between the architect and the site’s population. In particular, I emphasize how users have repurposed the technology to present their identity and connect in personally meaningful ways while the architect works to define and regulate acceptable models of use. REFLEXIVE CONSIDERATIONS It is important to note that I have been an active participant-observer amidst both Friendster users and the social networking software creators. I have tracked Friendster through the media and through the viral discussions on mailing lists, blogs and IRC channels I have organized six focus groups of various relevant social groups and have interviewed or surveyed over 200 users on various aspects of their Friendster experience. Via access to 1/3 of the Profiles, I have analyzed thousands of Profiles and run queries on the visible data. While the vast majority of these accounts are located abroad (predominantly in Asia), the majority of my informants have been in the United States and Canada. Furthermore, the majority of my interviews took place before October 2003, although the site continues to grow and attract new relevant social groups. I have also consulted with or advised many competing companies and I have regularly informed the press of my findings. In studying Friendster, my primary ethnographic goals are to: 1) understand how people negotiate context when presenting themselves; 2) examine how the network structure of a meme spreads and connects people; 3) determine the issues involved in articulating one’s social network as compared to a behavior-driven network. WHAT IS FRIENDSTER? Friendster is a website that allows people to explicitly articulate their social network, present themselves through a Profile (interests and demographics), post public Testimonials about one another, and browse a network of people. Friendster is built on the assumption that friends-of-friends are more likely to be good dates than strangers. The site was built to compete with Match.com and other online dating sites, with social networks as an additional feature. While Milgram argues that everyone is connected through a countable number of connections [5], Friendster only allows you to access those within four degrees. Unlike most dating sites, Friendster encourages users to join even if they are not looking for dates, under the assumption that they probably know a wide variety of friends who are looking and, thus, would serve as a meaningful connector and recommender. Friendster launched its public beta in the fall of 2002. As of early January 2004, the site is still in beta and has amassed Copyright is held by the author/owner(s). CHI 2004, April 24–29, 2004, Vienna, Austria. ACM 1-58113-703-6/04/0004. CHI 2004 ׀ Late Breaking Results Paper 24-29 April ׀ Vienna, Austria 1279over 5 million registered accounts and is still growing. Both mainstream and alternative press have covered the site, yet word of mouth is the dominant entry point for most people. It is important to note that users have a selfish motivation in spreading the meme, as their network grows by doing so. Friendster’s population is primarily cluster-driven and users often convince their entire friend group to participate. While Friendster users are typically 20-something, educated city dwellers, their social and sexual interests are quite diverse. As such, they bring vastly different intentions and expectations to the site. THE VALUE OF THE NETWORK Friendster assumes that users will authentically define their identity via their Profile so as to ensure more meaningful connections. Embedded in this is the assumption that users will see the value in connecting to actual Friends. This is a critical assumption because the desired theoretical results rely on the accuracy of this. Unfortunately, Friendster fails to recognize that publicly articulated social networks and identities are not identical to the private articulation gathered by sociologists. Furthermore, while sociologists have employed various techniques to categorize and weight relationships, people are often unable to do this individually. As shown in Friendster, this architectural difference results in behavior not predicted by the sociological analysis developed on top of observed behavior and protected informants. Relationship indicators in Friendster are binary: Friend or not. When traversing the network, there is no way to determine what metric was used or what the role or weight of the relationship is. While some people are willing to indicate anyone as Friends, and others stick