Thomas Erickson, N. Sadat Shami , Wendy A. Kellogg, David W. Levine

1. Synchronous Interaction Among Hundreds: An Evaluation of a Conference in an Avatar-based Virtual Environment Thomas Erickson, N. Sadat Shami, Wendy A. Kellogg, David W. Levine

2. OUTLINE • Background • IBM • Second Life • AGM Virtual Conference Center • Methodsand Data • Findings • Perceptions • ExperiencingtheConference • Keynotes • SocialEvents • Poster Sessions • Discussion

3. Background • Most studies come out of areas like tourism research or theeventmanagementtradepress. • An exception is Brecht, who provides a high level view ofconference activity [2], but does not address the details ofhuman interaction of interest to those in HCI. • Thispaper is the first to address thiscall with a mixed method evaluation of a virtual conference. • There is also little work on the use of virtual worlds in businesscontexts. In this paper we address this gap, offering an in-depth study of a large virtual conference.

4. IBM • IBM is a global information technology firm with over300,000 employees. Its businesses range from hardware tosoftware toconsulting. • IBM’s Academy of Technology • The Academy is anorganization of IBM’s technical leaders that has existed forover20 years. • Academy membersparticipate in company-wide studies, workshops, and anannual conference, the Academy General Meeting (AGM)

5. Second Life • immersive 3D environmentthatrepresentsusers as avatars. • Userscontroltheiravatarstocommunicate with other avatars via textual chatand speech, and to create, modify andetc. • Users wearheadsets and as they spoke their in-worldvoices came from the correct direction, and varied involume according to the distance between speaker andhearer.

6. AGM Virtual Conference Center • consisted of several spaces: • conference halls for the Keynotes • areas forsocial events • posters area

7. AGM Virtual Conference Center(Cont’d) • Prior to the conference, wereheld to acquaint users with the virtual world: how to moveand navigate, customize their avatars, and interact withothers While often such • training is not widely used. • the survey we will discuss laterindicated that 73% of attendees reported spending 2 ormore hours in preparation.

8. Methodsand Data • Attendance was logged allowing determinationof the identity and home countries of attendees. • Theyiteratively developed the survey by reviewing onlinediscussions of prior virtual meetings held at IBM. • The final survey consisted of about 40questions. • 1095 Academy members and guests (e.g.,presenters) who were invited to the meeting, but did notnecessarily attend it. It received 444 responses (40%),though not everyone responded to every question.

9. Findings • 502 people attended from around the world, • most returning for 2 or 3 days (Figure 5). • About 350 peopleattended each day for the first 2 days. • most participants appeared to use voice. • Posters were not crowded.

10. Problems • Only20% of the survey respondents reported that theyexperiencednoproblems. • audio problems (52%) to server lag. • machine crashes (35% and 31%) • difficulties logging on (27%).

11. Perceptions

12. Avatars • Many were interested in being “appropriate,” either out ofconcern for making a good impression or beingapproachable. • «most hadn’t spent much time on theirappearance and looked eerily similar. He’d tried to makehis avatar look like himself» • Informants generally agreed that they couldn’t recognizeothersbytheiravatars. • «I just prefer to be in formal dress» • «thatwasfunnybecause a lot of people were in shorts and stuff; and thatwas kind of fun» • «My avatar has blue hair and sparklyboots. it looksdifferent and unique. That’s the kind of look I was goingfor»

13. ExperiencingtheConference: Keynotes • Keynotes were not worth theoverhead, and that video streaming would have been better.They cited problems with lag, slide resolution, and sound.

14. ExperiencingtheConference: SocialEvents • Most had more mixed reactions; several mentioned a highlystructured Social that contrasted with the more informallyorganizedSocials. • oneinformantfeltthatthe structured Social was “stupid” and “very artificial” • a concern that one’s conversationwould be overheard or would interrupt others.

15. ExperiencingtheConference: Poster Sessions • poster sessions never appearedcrowded. • informants repeatedly told that «the posterswere the most successful aspect of the virtual conference.» • peoplewalkingthroughtheposters area could see who was present, but could not hearconversationsin thebooths.

16. Discussion • each event had active participation. • majorities of survey respondents agreed that the virtualAGM was «a good experience» • Second Life «workedwell» as referring to supporting the aims of theconference. • it is interesting to establish that a virtual world cansupport a professionally-oriented synchronous gathering ofhundredsof people. • Keynoteswereleastsuccessful, with many believing that streaming video wouldhavebeenmoreeffective.

17. Discussion(Cont’d) • Proposeda model of large scalesocial interaction that begins with Goffman’sobservations. • three things must happen to supportproductive large scale social interaction: • Coalescence: interactantsin a large gatheringneed to coalesce into small groups to enable coherentconversation • Focused Interaction: a group needs to initiateand manage its focused interaction. • Remixing: as time goes on, groups must changetheir make up over time.