Encouraging uses of location-aware systems

Reading (again and again) articles about location-awareness for a journal paper I am writing, I ran across "The Carrot Approach: Encouraging use of location systems" by Kieran Mansley, Alastair R. Beresford and David Scott which I found quite interesting. The paper addresses the lack of understanding about why location-aware applications can be useful and what factors can motivate people to use them, through a case study of AT&T's Bat system. The use of Bat they're interested in here is the one of the person-locator application or of context-aware paging. As they describe, the system is quite efficient as an indoor positioning provider. Accuracy and coverage are excellent but they noticed a "lack of genuinely useful applications and a strategy for their deployment". they developed a classification of the intended users of the location system (in their case, the staff and students within the lab) with the aim of targeting applications at the needs of specific social groups. Using the prisoner's dilemma approach, they show which ones are relevant.

What they found is that:

"We model the utility to an individual of an application by the formula Utility = AU2 + B where U is the number of participating users and A and B are constants. AU2 is the Metcalfe-effect and B the single-user payoff. Applications fall into one of three categories: Type I : those useful to isolated individuals (high B); Type II: those useful to small subgroups (high A, small set of users U ); and Type III: those only useful when the whole lab participates (high A, whole lab U ). Many traditional applications (e.g. the “person-locator”) are Type III applications; most of the applications we present here are either Type I or Type II. (...) We analysed the recent decline in Active Bat usage from a game-theoretic standpoint and argued that many existing location-aware Type III applications have fallen into disuse as a consequence of the well-known Prisoners’ Dilemma. We described how this trap could be avoided if Type I or Type II applications are provided which are of immediate use to individuals and small social groups. Furthermore, increased overall participation has an overwhelmingly positive effect as users of the location system receive a community benefit from increased take-up, both from being able to locate colleagues more reliably and from increased privacy. We claim this principle justifies the existence of applications that have no intrinsically useful purpose (such as games)."

Why do I blog this? quite relevant for current writings and talks about location-awareness in mobile social computing. The game-theory approach is original and brings interesting arguments to the table.

What's interesting IMO wrt to person-locator is this notion of

"from a game-theoretic standpoint, this application may be modelled by a multi-player prisoners’ dilemma. In real-life each person chooses whether to wear their bat or not whereas in the prisoners’ dilemma each prisoner chooses whether to co-operate with the authorities or not. Both wearing a Bat and co-operating have an associated (small) cost. If everyone co-operates (i.e. everyone wears their Bats) then the whole group receives a benefit. However, from the point of view of an individual it is always better not to co-operate (i.e. not wear their Bat) while secretly hoping that everyone else does; this is said to be the dominant strategy. It does not matter how great the benefit (i.e. the size of the coverage area) is; if all the players are rational then no-one co-operates and no-one wears their Bats. Therefore coverage area and applications like the “person-locator” cannot explain the difference in uptake between the LCE and AT&T. "

Mansley, K., Beresford, A.R. & Scott, D. (2004). The Carrot Approach: Encouraging use of location systems. In Proceedings of Ubiquitous Computing: 6th International Conference, Nottingham, UK, September 7-10, 2004, pp. 366-383.