Examining Leadership Dynamics in Agent Based Modeling

1
Examining Leadership Dynamics in Agent Based
Modeling

• By Alex McGuigan

2
Purpose/Goal

• Explore leadership dynamics in Sugarscape
• Analyzing leadership traits in combat
• Ultimate goal to discover what leadership trait
  is most attractive to followers
• What leadership trait forms the most successful
  groups

3
Scope

• Use Sugarscape as a base
• Add a more complete system of combat and traits
• Same simplistic, interconnected style as
  Sugarscape
• Intelligence Vision
• Strength Metabolism
• Wealth Sugar

4
Traits

• Vision
• Sight, obviously
• Intelligence
• Evaluation of others
• Lower vision underestimates strength of enemy

• Metabolism
• Sugar consumed per turn
• Strength
• Higher metabolism increases effectiveness in
  combat
• Strong agents have a harder time gathering

5
Traits

• Sugar
• Food
• Groups are food sharing
• Advantage for group gathering
• High metabolism agents fed by low metabolism
  group members, protects them in return

• Morale
• Has been scrapped

6
Combat

• The equation Metabolism 4 Sugar
• Agents drop 2 sugar when they die in combat
• Agents will move to the highest value sugar spot
  they can see, sugar at site 2 if there is
  another agent there
• Agents don't seek combat specifically, only seek
  highest concentration sugar
• Agents don't attack agents that they know are
  stronger

7
Combat

• If the highest value sugar spot is occupied,
  check to see if the occupant is stronger
• Your combat vs. theirs, tie to defenders
• Vision influences this, low vision underestimates
  their strength
• Your combat vs. their combat Vision/4
• Vision maxes at four, 4 vision has perfect
  perception, 1 vision attacks nearly
  indiscriminately
• If your's is perceived as higher, attack
• Run combat formula, tie to defenders, loser dies,
  winner gains 2 sugar

8
Group Dynamics

• Decentralized groups, emerge spontaneously
• What motivates them to form groups?
• Groups are beneficial, food sharing, shared
  vision
• Group advantages to gathering?
• Group combat?

9
Group Dynamics

• Who will be the leader?
• Highest Sugar person?
• Highest total stats?
• First member?
• Leader advantages
• Select new members vision based
• Limited command over group?

10
Design

• MASON version of Sugarscape
• Test by placing agents of random traits in
  Sugarscape, see what happens
• No randomness inside actual mechanics
• Simplicity is the goal

11
Development

• Combat dynamic coded
• Large series of difficult bugs remain in system
  of sugar and occupant detection
• Group dynamic largely planned
• Traits implemented, group functionality partly
  coded

12
Results

• Robert Axtell, one of the two creators of
  Sugarscape approves of the project
• Actual testing is held up by aforementioned bugs