Cooperation in a Competitive Environment

1
Cooperation in a Competitive Environment

• Supervisee John Richter
• Supervisor Philip Sterne

2
Pre-Contents Explanation

• The manner in which this presentation shall be
  given shall adhere to the well-known Scientific
  Method
• As such, it may differ from the presentation norms

3
Contents

• Section -1 Explanation of Scientific Method
  use
• Section 0 Contents
• Section 1 Characterization
• Section 2 Hypothesis
• Section 3 Prediction
• Section 4 Experimentation
• Section 5 Conclusions

4
Section 1 Characterization

• This, according to Wikipedia.org, can be defined
  as a careful characterization of the subject of
  the investigation
• Characterization The act of describing
  distinctive characteristics or essential features

5
Section 1 Characterization

• To characterize
• The agents
• Making use of a hybridisation of ANNs, GAs, and
  RL
• The game
• Greediness pays, but cooperation is sustainable

6
Section 1 Characterization

• Derived Characterization
• Game Rules
• Two kinds of food
• Specialization
• Agent Rules
• Donations, generosity
• Deception and trust

7
Section 2 Hypothesis

• This researcher hypothesizes that despite
  independence and greed being possible, a society
  of sufficiently intelligent agents programmed
  with simple rules will display cooperation as a
  form of emergent behaviour.

8
Section 3 Prediction

• This researcher predicts
• Emergent cooperation
• See Reynolds, C. Boids. http//www.red3d.com/cwr/b
  oids/index.html
• Sub-cultures
• Early high-death rates, followed by long-term
  stability
• Eventual negligent amounts of deception
• Hunters and Gatherers forming regular trade
  partnerships

9
Section 4 Experimentation

• The present progress extends mostly into this
  phase
• Progress into components
• ANN - 99.9
• GA - 100
• RL - 0
• Other (Hybrid, etc.) - 0

10
Section 5 Conclusions

• Watch this space, and final write-up for more!

11
Final Notes

• This research falls into the category of MAS
  (Multi-Agent Simulation), and more specifically,
  MASS (Multi-Agent Social Simulation).
• One of the chief authors in this field, which
  this researcher highly recommends reading, is
  Cristiano Castelfranchi

12
Questions

• For all those who would like an alternative to
  questions, the following haiku has been selected
  as one of particular quality. This researcher
  hopes you enjoy it
• In the falling snow
• A laughing boy holds out his palms
• Until they are white