Nicolas Nova on location awareness and collaborative problem-solving
Nicolas Nova visited the Institute today, and gave a talk on location awareness in mobile gaming. (In an odd coincidence, Doug Engelbart was also at the Institute today.) These are my notes from Nicolas' talk. As with all talks, the speaker shouldn't be held responsible for my misinterpretations, errors, etc..
124 University Avenue, part of Nicolas' Flickr photostream
Nicolas is a grad student at EPFL, in the CRAFT Lab (center for research on training in technologies), which does work on CSCW and distance learning, particularly interactive and mobile technologies. Interested in socio-cognitive processes in SCSW and games; also works part-time as a consultant for video game companies.
Background. For his MA thesis, Nicolas worked on importance of location-awareness (of others) in collaborative problem-solving in VR.
Knowing the location of others does lots of good things: it improves task performance and coordination, supports division of labor, communication and "grounding" (shared understandings of a situation), and "mutual modeling" (inferences we make of collaborators' backgrounds, intentions, etc.).
Location awareness is also a factor in some forms of group work, esp. joint activities performed in different locations: for example, firefighters, soldiers, emergency workers, medical personnel, and many others need to know where their colleagues are. In these and other cases, the location of actors has meaning for practitioners.
Location awareness is also an ingredient in building mutuality of knowledge. What we infer from others' position (now) what they're doing, what they're looking; traces (past) lets us infer what they know or have experienced; direction (future) lets us infer destination, intention, and strategy.
Navigation (e.g., where's the nearest Starbucks) is single-user; what's more interesting is that place annotation (tagging places with information) and location tracking are multi-user.
The experiment. Wanted to study how people make use of location awareness information when collaborating for problem-solving and group coordination. Constructed an experiment midway between lab and ethnography: takes place in the real world, but with certain factors controlled: a mobile game called CatchBob! (a 3-player collaborative treasure hunt), which was played on the EPFL campus with tablet PCs. There were two versions of the UI, which showed a map of the playing field, and let players add annotations and send messages to each other. They were identical in every respect but one: one version showed the locations of other players on the screen, while the other did not.
The experiment yielded both quantitative information regarding performance and process measures (path, messages, workload, errors), and qualitative data (questionnaire and group debriefing).
Analysis. There weren't significant differences in overall performance between players who could see each other on-screen, and those who couldn't. But players with displays also made more mistakes when recalling their partners' movements, while players without displays relied more on messaging (mainly about positioning, and directions, but esp. about strategy). People who wrote more messages also recalled collaborators' paths better, and seem to have reshaped their strategy through map annotations.
Current conclusion. Automatically knowing where the others are does not facilitate task performance, and it does not facilitate the representation of the others' path in space. Players without the tool took better advantage of the annotation feature; they picked up facts relevant to the task, and discussed what they needed to do more effectively than those with the display. No differences in cognitive workload.
Some Big Ideas
Not giving the tool is better...
- To support collaboration processes; it fosters elaborated explanations, self-regulation, strategic explications.
- For coordination activities. Users with the automatic position of others assumed that it was relevant and did not discuss other issues.
- It's better to let users express what they think is relevant for the task.
- It's less intrusive, gives people control over how/whether to announce their presence, lets us get rid of privacy issues.
Location is not enough. Automatic location-awareness is just information, but self-declared positioning is both information and a communciation of an intention: you want others to know where you are, and the communication itself has meaning.
Location awareness must be matched to task needs. The treasure hunt itself was not a single task; it really was several tasks. Sometimes automatic location awareness got in the way; some times it was useful.
Your notes are pretty accurate Alex! and thanks for the visit (again)
More information here (papers): http://craftsrv1.epfl.ch/research/catchbob/
Posted by: Nicolas | August 12, 2005 at 11:43 AM