Scott Klemmer

1
Thoughts on teachingproblem-finding incomputer
science education

• Scott Klemmer
• CS Faculty Lunch
• 07 June 2005

2
CS at a Crossroads

• Early innovation is done
• The big commercializations of that early work
  have occurred
• personal computers, the internet,
• The growing pains of moving from adolescence to
  maturity.

3
Challenge

• Give a person a fish, they eat for a day teach a
  person to fish, they eat for a lifetime
• until the world becomes over-fished
• Teach someone to think about food sources, and
  they eat for all time

4
Challenge

• Stanford computer science undergraduates spend
  thousands of hours solving problems over their
  four years, but far, far less finding them
• Craftsmanship is important, but so is
  problem-finding

5
Why is it this way?

• My hypothesis teaching and evaluating
  solution-finding is clean while teaching and
  evaluating problem-finding is much messier
• Problem-finding is implicitly taught to graduate
  students and those entering industry through an
  apprentice model

6
Current algorithm

• The big innovations both in research and industry
  have been the result of insight and vision
• Our current algorithm for fostering this insight
  and vision is
• 1) value it a lot
• 2) make silicon valley an attractive place for
  creative, innovative, insightful people

7
Hypothesis

• If we spent more time teaching problem-finding,
  Stanford would produce even higher-quality
  leaders in industry and academia

8
A possible counter-argument

• Thomas Edison, Innovation is 1 inspiration and
  99 perspiration
• The potential value of an idea is determined by
  the quality of the inspiration, and the extent to
  which it realizes that value is determined by the
  quality of the perspiration

9
A different counter-argument

• Teaching problem-finding is important, but it's
  not what CS is responsible for teaching. It's the
  domain of the business school or somewhere else
• Music and the visual arts more commonly recognize
  that craftsmanship is important but not
  sufficient we can take a cue from this

10
Okay, I agree what do we do?

• When something is important in the curriculum, we
  should do three things
• Put it in the mission statement
• Teach and assign techniques
• Evaluate how well students do

11
Some initial steps that HCI is taking

• Teaching need-finding
• Observe errors in the world
• Structured observation of a user population
• Be the user
• Using the studio critique

12
The primary challenge

• As currently employed, these teaching methods are
  much more time-consuming per student
• Increased TAs are one way to reduce faculty load

13
The opportunities

• Train future academics and industry professionals
  that can more creatively find problems of
  important value to the economy
• Potentially create a more diverse student
  population