Task and Interruption Management in Activity-Centric Computing

Throughout history, the design of interactive computing systems has
always been inhibited by technological limitations of the time. For
a truly groundbreaking paradigm shift to occur (reshaping the very
nature of human-computer interaction), a tremendous amount of research
and engineering is required. Therefore, most computing systems
instead build on top of an existing stack of contemporaneous
technologies, inescapably adhering to their underlying interaction
paradigms. When left unquestioned, such an incremental approach
inadvertently shoehorns system design into preexisting notions of
how computing systems work. Thus, it is likely that design decisions
imposed by technological constraints of the past have needlessly
been carried over to modern-day systems. With information technology
now forming a major part of our daily lives and giving rise to
new emerging design challenges, it is prudent to address these not in
isolation, but by fundamentally reevaluating the current computing
paradigm.
To this end, activity-centric computing has been brought forward
as an alternative computing paradigm, addressing the increasing
strain put on modern-day computing systems. Activity-centric computing
follows a top-down approach to design using the full context
of human activity as the starting point of analysis. The focus no
longer lies on individual technologies, but on how computing systems
are used as mediators within the broader context of human
intentionality, thus also taking into account the encompassing community,
environment, and dependencies on other technologies. Users
can aggregate resources, work, and collaborate on them within goaloriented
workspaces that are meaningful to the user, as opposed to
having to adhere to data structures imposed by specific technologies.
Such systems have been deployed successfully in a variety of different
domains, including healthcare, experimental biology, and software
engineering.
However, several recurring open issues have been identified based
on the deployment and evaluation of different activity-centric computing
systems. Broadly speaking these impact the scalability and
intelligibility of current research prototypes. In this dissertation, I
postulate that such issues arise due to a lack of support for the full
set of practices which make up activity management. Most notably, although
task and interruption management are an integral part of personal
information management, they have thus far been neglected
in prior activity-centric computing systems. Advancing the research
agenda of activity-centric computing, I (1) implement and evaluate an activity-centric desktop computing system, incorporating support for
interruptions and long-term task management; (2) provide empirical
data on the overhead of switching between activities when using contemporary
desktop computing systems; and (3) implement a software
architecture facilitating developers to aggregate resources handled by
independent applications into one central activity manager.