Transcript Probing the Limits: Can an Engineering Science be built

An Engineering Science
within HCI?
William Newman
UCLIC
15 June 2007
Overview
•
•
•
•
•
•
•
Technology’s human-enhancement role
Engineering science
Past attempts within HCI
What it means to do engineering science
Human-human communication as a domain
Opportunities within this domain
How this could affect HCI research
Technology’s role in society
• Enhancing us as human beings
• Increasing...
– our bodily efficiency
– our sensory efficiency
– our intellect
(Rogers 1983)
• Can we disregard the
enhancement imperative?
The Science of Enhancement
Design’s dependency chain:
• Measurement
• Prediction
• Models
• Theories
• Data
Provided by engineering science.
Engineering Science in HCI:
Past successes
• GOMS, Cognitive Walkthrough
• Providing models of interaction
• Supporting analyses and predictions of
how technologies support human activities
Application of Engineering Science:
Project Ernestine*
• Enhancement of telephone operators’
efficiency in handling calls for assistance
• What to measure? Call handling time
• How to predict? CPM-GOMS models
• The prediction: new design 0.65s slower
• Field-trial finding: 0.63s slower
• How to enhance? Personal response system
• New prediction: 0.9s faster
*Gray, John and Atwood (1993)
The Challenge for HCI
• Design’s diminishing emphasis on
supporting repetitive work
• Can we develop an engineering science for
today’s design?
• What kinds of research would this involve?
• What does it mean to do engineering
science?
• Would it divide the research community?
What it means* to do
Engineering Science
A.
B.
C.
D.
E.
F.
Select a domain of design
Collect lots of data
Try lots of analytical approaches
Identify patterns, build and test models
Identify criteria for measuring behaviours
Transform models into design tools
And iterate!
* according to W.G. Vincenti (1990) and others
Selecting the Design Domain
• From within Rogers’s areas of human
enhancement: with some additions:
enhancement,
– bodily efficiency
– sensory efficiency
– intellect
– enjoyment
– human-human communication
An Engineering Science of
Human-Human Communication
• Why?
– a context for much of human-computer
interaction – even in Project Ernestine!
– offers a purchase for modelling (Sacks et al.
1974, Goodwin 1980, Clark 1996, ...)
– room for improvements to technology
• Where are some opportunities?
– meetings, including face-to-face
– e-mail
– writing
Enhancing Communication in
Face-to-Face Meetings
• Today’s technology designs show little
consideration for the needs of conversants
percentage of pauses
3.0
2.5
2.0
1.5
1.0
0.5
0.0
2
4
6
8
10 12
pause duration, seconds
14
Finally:
How this could affect HCI research?
• First, must there be a methodological
divide?
• Or can existing methods contribute?
level of science
models
metrics
guidelines
findings
implications
effort
In summary:
How this could affect HCI research?
• Lengthy start-up research involved
but...
• Need not create a methodological divide
• Intellectually challenging
• Different from research that developers do
• Provides what designers often need
• Remains relevant while technology advances
• Offers a place for truly inter-disciplinary
collaborations