2.1 The Presentation of the Data Overload Challenge

The challenge presented to the team was to develop information visualizations
that would help intelligence analysts deal with the avalanche of electronic,
textual data that is available for them to use in generating a coherent response to
a question. Specifically, there was interest in helping analysts find appropriate
information to formulate answers to questions outside their immediate expertise
under a tight deadline such as 24 hours, often referred to in the intelligence
community as Quick Reaction Tasks (QRTs).

In one sense, the problem seems paradoxical because all intelligence analysts
agree that access to more data ought to be a benefit. However, the benefit in
principle has not been matched by the benefit in practice. The sheer volume of
the data creates a situation where it is difficult to determine where to look in the
data field, it becomes easy to miss critical information, and determining the
significance of data in relation to the ongoing context is challenging.

Although data overload, or perhaps more descriptively data avalanche, in
intelligence analysis is a compelling problem for real practitioners who really
have to do important work and need immediate help, we felt that, in order to
make more than incremental progress, it was important to step back and better
characterize the problem before committing to a particular design direction.
Therefore, rather than immediately constructing and evaluating information
visualizations for text summarizer algorithms as originally requested, we delved
into Cognitive Systems Engineering (CSE) research bases and our team members'
past experiences with researching and designing parameter data displays in
space shuttle mission control (Malin et al., 1991; Thronesbery, Christoffersen,
Malin, 1999), display and mode proliferation on aviation flight decks (Sarter and
Woods, 1992; Billings, 1996), agent explanations and alarms in anesthesiology
(Cook and Woods, 1996; Johannesen, Cook, and Woods, 1994), distributed
military command and control (Shattuck and Woods, 1997), navigation in
computer displays and spreadsheets (Woods and Watts, 1997), and alarm
overload in nuclear power generation (Woods, 1995a) with the following
questions:

• What is the definition of data overload?
• Why is data overload so difficult to address?
• Why have new waves of technology exacerbated, rather than resolved, data
overload?
• How are people able to cope with data overload?



TABLE OF CONTENTS