Policy Evolution within an Organization

1
Policy Evolution within an Organization

• James H. Hines
• Sloan School of Business, Massachusetts Institute
  of Technology
• Jody Lee House
• Department of Electrical and Computer
  Engineering, Oregon Graduate Institute

Funded in part by NSF IOC AwardSES-9975942
2
The Problem

• System-wide company improvement is difficult
  because companies are too complex to solve.
• How can we improve organizations in the face of
  ignorance?

3
A solution?

• Biological evolution has produced excellent
  organizations.
• Can we identify analogs of natural evolution that
  will help human organizations to likewise excel?

4
GeneOrganismPolicyOrganization

• Policy
• Implicit or explicit
• Examples
• Pricing
• Hiring
• Capacity Expansion
• Flywheel sales
• Synonyms
• Decision rule
• Rule of thumb

5

• A policy produces
• A stream of decisions
• Activity in the firm
• Changing the policies, changes the organization

• A gene produces
• A stream of proteins
• Activity in the cell
• Changing the genes changes the organism

6
Where are evolutionary packets stored?

• Genes are stored on chromosomes in cells
• Policies are stored
• In written manuals?
• In committees?
• On computers?
• In brains of people

7
Processesvs
Genes
Policies

• Mutation
• Recombination
• Natural selection and the sex drive
• survival of the fittest

• Innovation
• Inter-personal learning
• Pointing and pushing mechanisms
• learning from the fittest

8
Pointing And Pushing Mechanisms

• Point to successful people
• Push others to learn from them
• Examples
• Promotion and hierarchy
• Pay scales
• The best and latest computers
• In house training?

9
A brief look at sex
10
Recombination is key

• Combine parts of fit organisms to create fitter
  organism
• Example 4-digit number, A gt B fitter

8,765
7,999
8,999
11
Learning is Similar to Biological Recombination
Fred
Phyllis
brain
Time 1
policy
Phyllis teaching
Fred learning
Time 2
12
Why learning is difficult to call to mind

• The donors idea is well integrated
• The rest of the donors idea is difficult to
  recognize as an idea

13
Overview
Step 4 If using
Step 3 Promote
teams Mix
managers and
Managers
reform teams
Step 2 Evaluate
performance of the
Step 5 Managers
system dynamics
learn
models
Step 1 Run system
Step 6 Managers
dynamics simulation
innovate
models, using policies of
the managers
14
Step 1 Run SD models
3
2
1
15
Step 1 The Project Model Detail
DesiredPeople
People
HireFire
Rate
CorrectlyDoing
Remaining
Productivity
Time
Correctly
Done
NormalQuality
ltTimegt
Doing
Quality
WorkToDo
DueDate
UndiscoveredBugs
Anticipated
TimeTo
IncorrectlyDoing
Complete
Anticipated
TimeTo
BugDetecting
DueDate
Anticipated
Change
TimeToDetectBugs
Production Rate
Schedule
ltPeoplegt
ltTimegt
ltProductivitygt
16
Step 2 Evaluating Performance

• Fitness function can be based on any variables in
  the model
• Variables can be combined using any functional
  form
• In the following we use two simple fitness
  functions
• Time to ship (LastPossible Actual)
• Number of bugs (LinesOfCode BuggyLines)

17
Step 3 Promoting managers

• Rank individuals based on relative performance
• Promote according to rank.

• The promotion algorithm requires specifying the
  promotion base. A promotion base of 2 means
• The highest performing managers new position is
  2 theOld
• The lowest performing managers new position is
  (1/2) theOld
• Everyone elses promotion is evenly spread out
  between 2 and 1/2

18
Step 3 Promotion Algorithm Detail
Team-based promotion
19
Step 4 If using teams mix them up

• Randomly?
• Spread out the best?
• Concentrate the best?

Step 5 Learn

• Select a teacher by roulette
• Learn from the teacher by recombination

20
Step 5 Learn ? p(learn)
Teachers Policy32 or 1000 00
Learners Policy10 or 0010 10
OLD
Randomly choose a crossover Point, say 2
Randomly choose which part the learner will
obtain and which he will retain
0010__ ____00 OR____10
1000__
0010 00 8
?
1000 10 34
Learners Policy34 or 100 10
Teachers Policy32 or 1000 00
NEW
21
Step 6 Innovate ? p(innovate)
111111
110111
Flip !
After
Before
22
Learning, no pushing/pointing Learning Drift
Optimal value 8
23
Learning, no pushing/pointing Random Consensus
24
Learning with Pointing/PushingIndividuals
25
Next steps

• Measurement through knowledge elicitation with
  partner companies
• Who learns from who and why?
• How are implicit policies a function of
  organizational structure?
• Integrated simulation