Development and Evaluation of Emerging Design Patterns for Ubiquitous Computing

1
Development and Evaluation of Emerging Design
Patterns for Ubiquitous Computing
Eric Chung Carnegie Mellon
Jason Hong Carnegie Mellon
Madhu Prabaker University of California, Berkeley
James Landay University of Washington
Alan Liu University of Washington
2
What Are Design Patterns?

• Design patterns communicate common design
  problems and good solutions in a compact form
• Started in architecture, recently for user
  interfaces
• Ex. Navigation Bar

3
Design Patterns for Ubicomp?

• Ubicomp pushes computing into physical world
• Wireless networking, sensors, devices
• Still in early phases of ubicomp, so why create a
  pattern language now?
• Speed up diffusion of interaction techniques and
  evaluation results
• Help us see links between ideas, see whats
  missing
• Like first periodic table
• Help designers avoid bad standards
• Avoid blue links and poor privacy

4
Our Work on Ubicomp Design Patterns

• Developed 45 patterns for ubicomp
• Evaluation with sixteen pairs of designers (32
  total)
• 9 pairs in first round of eval, 7 pairs in second
  round
• Compared the design of a location-enhanced app
  with and without patterns
• Better communication? Novices and experts?
  Privacy?

5
Talk Outline

• Overview
• Method for Creating the Patterns
• Evaluating the Patterns
• Future Work

6
Method for Creating the Patterns

• Iterative process over three months
• Literature review to extract ideas
• Tried to do top-bottom, too hard
• Bottom-up much easier, card sorting to organize
  into groups
• 80 pattern candidates, focusing on interaction
  design
• 2 pages each
• Critiqued by four other researchers
• Cut to 45 patterns for the first evaluation

7
Example PatternA12 Enabling Mobile Commerce
8
Example PatternA12 Enabling Mobile Commerce
9
Some More Example Patterns
D Fluid Interactions
C Techniques for Privacy
B Physical / Virtual Spaces
A Application Genres
10
Bus Stops for Relating Patterns
11
Talk Outline

• Overview
• Method for Creating the Patterns
• Evaluating the Patterns
• Future Work

12
First Round of Evaluation

• Nine pairs of designers
• Prototype a location-enhanced guide for shopping
  mall
• Gave each pair a set of general goals to support
• Could add any reasonable features, use any
  reasonable technologies
• 80 minutes to prototype, 10 minute presentation
  to client
• Will focus on qualitative results
• Had judges rate designs quantitatively,
  statistics hard though

High Exp (6 yrs) Low Exp
Patterns 2 pairs 2 pairs
No Patterns 3 pairs 2 pairs
13
Observations from First Round Eval

• Patterns helped novice designers
• Novices without patterns struggled with tech,
  features
• Novices with patterns fared better, patterns
  useful for getting ideas and explaining concepts
  to one another
• Patterns helped experts with an unfamiliar domain
• Skim thru patterns to get ideas, see range of
  possibilities
• Patterns helped designers communicate ideas
• Expected designers to adopt names (unrealistic in
  retrospect)
• Common to see designers point at pictures
• Many design pairs leveraged a web pattern
  language
• Navigation Bar, pages, cookies, bookmarks

14
Observations from First Round Eval

• Patterns helped designers avoid some design
  problems
• Most teams came up with similar solutions in both
  conditions
• But teams w/o patterns had to re-visit solutions
  more often
• Had to re-invent wheel and re-learn mistakes
• Patterns did not help with privacy
• Most design teams identified privacy as a problem
• But the teams didnt use our patterns
• Designers generally liked the idea of patterns
• Good idea to identify design patterns for
  ubicomp
• But Too many patterns to digest
• If we had more time, Im sure that we would be
  able to use these patterns to tailor them to our
  own ideas.

15
Second Round of Evaluation

• Reduced to 30 patterns
• Edited some content, added more links
• Seven pairs of designers
• Six pairs had patterns, one did not
• Already knew what non-pattern condition results
  were
• Same task
• Same amount of time

16
Observations from Second Round Eval

• 9 of 12 thought patterns helped with design task
• 11 of 12 thought patterns would help with future
  designs
• These patterns are almost like a checklist. You
  can cover all of your bases.
• Patterns used more often to communicate ideas
• Some patterns used to inspire designs
• D5 Serendipity in Exploration, app should not
  be a pushy salesperson but allow for free
  roaming.
• One pair used patterns to annotate ideas
• B1 Active Map next to the sketched UI
• But only one group used the privacy patterns

17
Future Work

• Continued evolution and evaluation of the
  patterns
• Why didnt privacy patterns work as we expected?
• Unclear format? Too abstract? Too specific?
• Not enough links? Too many patterns?
• Important b/c we want to avoid expected privacy
  problems
• Landay and Prabaker working on ubicomp patterns
  for the home at Intel Research Seattle
• 20 new patterns for the home
• 22 pairs of designers, half with patterns, half
  without
• Data analysis in progress

18
Summary

• Design patterns for ubicomp
• 30 patterns in current set
• Evaluation with 16 pairs of designers
• Generally useful in design task for generating
  and communicating design ideas
• Still didnt use privacy patterns
• Our patterns can be downloaded at
• http//guir.berkeley.edu/patterns
• Any feedback appreciated
• Help us evolve them!