Doing usercentered design

1
Doing user-centered design
2
Outline

• I. The failure of intuition
• II. Some first principles of user-centered design
  (UCD)
• III. Applying the principles in practice The
  UARC/SPARC story

3
The failure of intuition

• Traditional software designers have a fatal flaw
• While they excel at formal logic, spatial
  reasoning, and math
• These same factors produce poor intuitions about
  what will make a system useful and usable for
  just plain folks (JPFs)

4
It worked fine for me...

• Designers can almost always run their own
  programs
• CS experiment my office mate could understand
  it
• It is very difficult to imagine how JPFs will
  respond to a system

5
Substitution for intuition

• Which will do better?
• Common sense knowledgeOR
• Psychology of system design

6
Expected gains

• Median increased user efficiency of 50
• Small gains (20 - 80) -- often a reflection of
  cleaning up flaws in the UI
• where there are typically 40 flaws!
• Big gains (gt 80) -- complete re-conceptualization
  of the users task

7
The user-centered design process
Conceptualize Study the intellectual difficulties
Build Get reactions tomock-ups
Trials Collectsystematic data
Modify extend design, evolution
8
First principles of UCD

• Get feedback from actual users
• For example
• Bailey (1993)
• controlled experiment recipe file system
• 2 conditions video VS no video
• designs by designers with video feedback were 30
  more efficient

9
First principles of UCD

• Compare alternatives
• For example
• Nielsen and Levy (1995)
• build two versions
• select the preferred option
• systems designed according to test-two and pick
  may be 25 more efficient

10
First principles of UCD

• Dont rest on your laurels
• For example
• Mac/Lisa interface (e.g., Tesler et al.)
• weekly usability testing
• long-term commitment to user feedback
• Mac-style interfaces are 48 to 78 more efficient
  than text-based interfaces

11
Applying the principles

• Study the intellectual difficulties from the
  users perspective
• Get user reactions to mock-ups
• Analyze systematic data on user behavior

12
Can we generalize to services?

• How to analyze the problem?
• What is the analog in the service domain to the
  mock-up in the systems domain?
• What data would you collect?

13
The nature of space physics
14
Remote observatories
Kangerlussauq, Greenland
15
UARC scope, 1998
16
The UARC testbed
Polar UVI
Session manager
TING model
Radar displays
Chat window
17
Applying the principles

• Study the intellectual difficulties from the
  users perspective

18
Applying the principles

• Study the intellectual difficulties from the
  users perspective
• use scenarios

19
Applying the principles

• Study the intellectual difficulties from the
  users perspective
• use scenarios
• Get user reactions to mock-ups

20
Applying the principles

• Study the intellectual difficulties from the
  users perspective
• use scenarios
• Get user reactions to mock-ups
• e.g., MacroMedia Director

21
Applying the principles

• Study the intellectual difficulties from the
  users perspective
• use scenarios
• Get user reactions to mock-ups
• e.g., MacroMedia Director
• Analyze systematic data on user behavior

22
Applying the principles

• Study the intellectual difficulties from the
  users perspective
• use scenarios
• Get user reactions to mock-ups
• e.g., MacroMedia Director
• Analyze systematic data on user behavior
• self-reports
• automated logs

23
Use scenarios

• Interviews with users
• Describe tasks and activities
• Understand contingencies
• Map behavior to system features

24
Mock-ups

• Hypothesis data views should be aligned along a
  time line
• Simulate this interface in Director
• Discoveries
• worked for continuous data
• image data worked better in a movie viewer

25
UARC original UI

• Problem
• bewildering array of choices
• each selection required a separate pass through
  the menus
• Result
• lots of wasted time
• frustration!!!

26
(No Transcript)
27
UARC UI re-design

• Goals
• allow scientists to reduce choice options
• allow selection of options in a single pass
• Approach
• dialog box
• drop down filters

28
(No Transcript)
29
(No Transcript)
30
(No Transcript)
31
Systematic data

• Content analysis of chat logs
• Logging user activity
• Surveying users

32
Content analysis Science talk
Composition of UARC communication by content and
participants, 1993-95
50
40
Scientists
30
Proportion of communication
Students
20
Developers
10
0
Science
Session
Technology
Display
Socializing
Content
33
Content analysis FTF vs CMC
FTF
CMC
70
60
50
40
30
20
10
0
Science
Technology
Display
Session
Social
34
Logging activity Usage
1000
Min
Median
750
Max
500
Hours logged on
250
0
F '93
F '94
W '94
W '95
SP '93
SP '94
SP '95
SU '93
SU '94
SU '95
Year and season
35
Content analysis Participation
35
30
25
20
15
Number of conversational turns
10
5
0
Participants Most active to least active
36
Costs of UCD

• Significant usage is important for field studies
• Reliability robustness are important to users
  doing real work
• but hard to achieve

37
Costs of UARC UCD

• Gathering bug reports data on user behavior
• Regular meetings to review bugs problems
• One week per month of debugging by programmers
• .5 FTE support person
• Maintenance of operational environment
• Support of campaigns

38
(No Transcript)