Organization Design Optimization Using Genetic Programming

1
Organization Design Optimization UsingGenetic
Programming

• Bijan Khosraviani
• Raymond E. Levitt
• John R. Koza
• Stanford University

2
Criterion E

• (E) The result is equal to or better than the
  most recent human-created solution to a
  long-standing problem for which there has been a
  succession of increasingly better human-created
  solutions.

3
Criterion H

• (H) The result holds its own or wins a regulated
  competition involving human contestants (in the
  form of either live human players or
  human-written computer programs).

4
(No Transcript)
5
Organization Design

• A complex multi-dimensional problem involving
  properties of individuals and sub-teams, activity
  assignments, percentage allocation for each
  activity, and decision-making policy

6
Constraints

• Organization designs are subject to constraints
  such as size of staff, skills of team members,
  adherence to minimum process quality

7
Fitness

• Compliant organizational designs are rated by the
  duration of the project

8
Human Performance

• Over 50 teams of 4-5 people have worked on a
  benchmark problem involving the construction of a
  biotech plant
• Every year since 1996, graduate students in CEE
  242 course at Stanford have broken into teams of
  4-5 to work on the problem
• Also, every year since 1996, practicing
  professional project managers from industry have
  broken into teams of 4-5 to work on the problem
  as part of a summer program at Stanford in the
  Dept. of Civil Engineering

9
Human Results

• Between 1996 and 2001, teams regularly created
  designs that were superior to previous designs
• Each new best design won a bottle of champagne
• Since 2001, no team has created an improved
  solution

10
GP-Created Design

• The organizational design created by genetic
  programming (and described in a GECCO-2004
  late-breaking paper) is 2 better than the best
  human-created design from over 50 teams of 4-5
  people between 1996 and 2001.