Trust, Kinship and Locality in the Iterated Prisoner's Dilemma
Trust, Kinship and Locality in the Iterated Prisoner's Dilemma
The Prisoner's Dilemma is maybe the best-known paradox in Game Theory. In this game, a player meets another player, and must choose to cooperate or defect: the best outcome for player i would be achieved if i defected while the opponent j cooperated, which would be the worst outcome for player j; both players would prefer mutual cooperation over mutual defection; and finally, the scenario is "symmetric". From the perspective of game theory, the rational choice in the prisoners dilemma is for both players to defect. It is called a "dilemma" because in fact both players would prefer mutual cooperation { but this outcome is impossible, because if one player cooperates, the other would rather defect. [Axelrod 1984] ran a tournament in which players played the prisoners dilemma against a number of opponents in a series of rounds (the "iterated prisoners dilemma"). He found that "cooperative" game playing strategies could flourish in this tournament if they were given the opportunity to encounter other cooperative strategies. This study is aimed at investigating the following issues around the iterated prisoners dilemma: (i) Trust: Suppose every agent is equipped with a value tw indicating how trustworthy it is; what happens if we take account of such a value in making decisions; how does this affect the dynamics of cooperation/defection? (ii) Kinship: Suppose we have a model of "family distance", so that agents are classified into families, being less likely to cooperate with those that are more distant in family terms. How do such concerns affect the dynamics of cooperation? (iii) Locality: Suppose agents are arranged on a graph, and retrieve trust information by querying and using trust of their neighbours. How does graph topology affect the dynamics of cooperation? E.g. is it the case that "gregarious" agents (with lots of neighbours) perform much better than "lonely" agents (with only 1 neighbour)? We will investigate through experiments the notions of Trust and Reliability applied to the Prisoner's Dilemma context, under multiple aspects. We will review works related to these topics in order to introduce some efficient solutions for dealing with the above issues.
Venanzi, Matteo
ba24a77f-31a6-4c05-a647-babf8f660440
Venanzi, Matteo
ba24a77f-31a6-4c05-a647-babf8f660440
Venanzi, Matteo
(2009)
Trust, Kinship and Locality in the Iterated Prisoner's Dilemma.
University of Rome "La Sapienza", Department of Computer Engineering (DIS), Masters Thesis.
Record type:
Thesis
(Masters)
Abstract
The Prisoner's Dilemma is maybe the best-known paradox in Game Theory. In this game, a player meets another player, and must choose to cooperate or defect: the best outcome for player i would be achieved if i defected while the opponent j cooperated, which would be the worst outcome for player j; both players would prefer mutual cooperation over mutual defection; and finally, the scenario is "symmetric". From the perspective of game theory, the rational choice in the prisoners dilemma is for both players to defect. It is called a "dilemma" because in fact both players would prefer mutual cooperation { but this outcome is impossible, because if one player cooperates, the other would rather defect. [Axelrod 1984] ran a tournament in which players played the prisoners dilemma against a number of opponents in a series of rounds (the "iterated prisoners dilemma"). He found that "cooperative" game playing strategies could flourish in this tournament if they were given the opportunity to encounter other cooperative strategies. This study is aimed at investigating the following issues around the iterated prisoners dilemma: (i) Trust: Suppose every agent is equipped with a value tw indicating how trustworthy it is; what happens if we take account of such a value in making decisions; how does this affect the dynamics of cooperation/defection? (ii) Kinship: Suppose we have a model of "family distance", so that agents are classified into families, being less likely to cooperate with those that are more distant in family terms. How do such concerns affect the dynamics of cooperation? (iii) Locality: Suppose agents are arranged on a graph, and retrieve trust information by querying and using trust of their neighbours. How does graph topology affect the dynamics of cooperation? E.g. is it the case that "gregarious" agents (with lots of neighbours) perform much better than "lonely" agents (with only 1 neighbour)? We will investigate through experiments the notions of Trust and Reliability applied to the Prisoner's Dilemma context, under multiple aspects. We will review works related to these topics in order to introduce some efficient solutions for dealing with the above issues.
This record has no associated files available for download.
More information
Accepted/In Press date: July 2009
Organisations:
Agents, Interactions & Complexity
Identifiers
Local EPrints ID: 272205
URI: http://eprints.soton.ac.uk/id/eprint/272205
PURE UUID: 8a32f38e-5de3-4403-b14d-3be4ea704f4c
Catalogue record
Date deposited: 16 Apr 2011 12:25
Last modified: 10 Dec 2021 23:32
Export record
Contributors
Author:
Matteo Venanzi
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics