Hilbert Problems in Visual Interface Development Evaluation

1
Hilbert Problems in Visual Interface
Development/ Evaluation
Terry SullivanDepartment of Technology
CognitionUniversity of North Texas
2
Overview

• What's a "Hilbert Problem"?
• Background theory-vs-empirical evaluation
• The Problems and a few "answers"
• Discussion

3
What's a "Hilbert Problem?"

• Named after David Hilbert
• Unsolved but tractable problems
• Answers contribute to sustained
  development/evolution
• Do we have Hilbert Problems?

4
Theory-vs-Empiricism

• Strong theoretical advocacy
• Visual bandwidth
• "Set-at-a-time" display
• Empirical results much less clear/consistent
• Barriers to systematic/meaningful empirical
  evaluation

5
Why Are (Aren't?) VIs Special?

• VIs are still software
• VIs are still interfaces
• User tasks underlying VIs are cognitively complex
• Answers require interdisciplinary approach

6
The Problems (Three Themes)

• Theory/Practice A Necessary Unity
• Current Status
• Pragmatics of Evaluation/Experimentation

7
Theory/PracticeA Necessary Unity

• State of current understanding of
• Perception
• Cognition
• Information Behavior

8
Current Issues in Development/Evaluation

• Task Specification?
• Sufficiently robust measures?
• Even across different tasks/interfaces?

9
Pragmatics of Evaluation/Experimentation

• Automated testing?
• Special tools and test suites?
• Test collections?
• Formative evaluation?

10
Insights

• Tasks don't change just because interface is
  visual
• Signal/noise ratio
• Throughput, not bandwidth

11
Discussion/Brainstorming

• Explore ideas for further development
• Research agenda/opportunities