Ubicomp, Limitations, Failures and Malfunctions

Posted: December 20th, 2005 | No Comments »

Starting to set a framewok:

Ubicomp, limitations, failures and malfunctions

The increasing ubiquity of computer and their diffusion into our environment requires reconsidering the complex interplay between technology and the human. While technology providers suggest that there are not limits to connectivity and mobility, service coverage and stability is anything but seamless in the real world. Moreover, since the infrastructure structure is meant to be invisible it will be necessary to develop an understanding of what failure means and how malfunctioning is communicated to the user.

For the moment there has been papers on the impact of uncertainties on individual experience, not on individual and group performance at a task. The closest works are on rather ethno-oriented-interaction design Defining Uncertainties in Can You See Me Now? (on the user experience of uncertainty) might be the psycho oriented Evaluating the Effects of Displaying Uncertainty in Context-Aware Applications (that shows that memory task is increased by explicitly displaying uncertainty information, and we have to be careful with the workload generated by showing uncertainty information.

Phenomenon of uncertainty
Sources of uncertainty in ubicomp environments are due to technical limitations (communication latency, deviations in positioning, data synchronization, fluctuating signal strength, power consumption, autonomy, processing and overall system complexity) and sometimes economical limitations (patchy network coverage and cost). I have to understand more the socio-psycho impacts or uncertainty.
High-level questions
[Group coordination in unstable ubicomp environments]
How do groups establish a shared context and coordinates their actions in real-world ubiquitous systems prone to failures and malfunctions?

[Detecting uncertainty]
How do users detect uncertainty and what are the clues?

[Impact on the users]
Understand how people use and react to discrepancies and sources of uncertainties in ubicomp infrastructures. More precisely, How do users know and learn how to avoid and rectify the system’s mistakes (e.g. learning to detect the seams and cold spot.

[Impact on the gro
up]
How do groups address the difficulty of ubiquitous systems (e.g. when lost of connectivity, positioning accuracy)?

Specific questions in CatchBob!
Variables at disposal for each player:

  • Elapsed time (in seconds)
  • Times (in seconds) with no, bad and good positioning accuracy
  • Connected and disconnected times (in seconds)
  • Ubiquitous index with positioning accuracy and connectivity
  • Attempted synchronization
  • Failed synchronization
  • Words sent
  • Nasa Load Index
  • Mental load
  • Physical load
  • Areas covered (the campus is divided in 20m2 areas)
  • Back path (player returning on his footsteps)
  • Covered path (number of areas covered by another player)
  • Players representation of each other’s path
  • Meanings of annotations (direction, strategy, acknoledgement, question, order)

Conditions where with awareness tool (aka automatic positioning) and without awareness tool (manual positioning). 10 groups for each condition.

Strategy and uncertainty

  • How did users react to bad connectivity and positioning accuracy and how did it affect their strategy? Did they return on their footsteps? Did they exchange more messages? What are the meanings of the messages?

Uncertainty and task performance

  • What are the impacts of bad connectivity and positioning accuracy to the accumulated paths and the numbers of areas covered

Perception of uncertainty

  • Are users sensible to loss of connectivity (because our connectivity numbers are rather good, but many players where verbose about the experience of positioning and connectivity)
  • Did bad connectivity and bad positioning increase the work load (Nasa Index)?
  • What is the representation of the other players path in a situation of bad connectivity and connectivity?

Approach
Mix of qualitative and quantitative data. Investigating the processes rather than the outcomes. By processes meant the interaction with the artefacts/application/service but also the group processes when they collaborate using it. The outcome or the performance is often less intersting than the processes that occured.