Essa é uma revisão anterior do documento!
Tabela de conteúdos
Books
The Complexity of Cooperation
Evolving New Strategies
Axelrod tournament. Genetic algorithm (Holland 92). Gens implementing a “decision tree” that indicates the strategy given the last three payoffs. The strategies always evolve to something near TFT. Sexual reproduction helps the search process.
Coping with Noise
Axelrod tournament with 1% of noise (invert one's action). Ecological simulation: the fraction in the next generation will be proportional to that's rule score in the previous generation. Generosity: there is a chance of not escalating if the opponent escalates in the previous round. Contriction: cooperate after the others defects in response to one's defection (that can arise because of the noise), contrite. CTFT is better than GTFT is better than GPavlov is better than Pavlov.
An Evolutionary Approach to Norms
n-player game. A norm exists in a given social setting to the extent that individuals usually act in a certain way and are often punished when seen not to be acting in this way.
When somebody escalates, the other players may perceive and then punish. It leads to cooperation some times, but not always. But, if the observer that does not punish can be punished by another player (a metanorm), the model always converge to a coperative and almost always punishing behaviour.
Choosing sides
Theory of landscapes. Two groups where each agent belongs to one of them. Each turn, one agent may exchange its group in order to enhance its satisfability. Each pair of agents has a symmetric satisfability value.
The main purpose of ABM is not prediction but a deeper understanding of how fundamental social process operate.
The Geometry of Ecological Interactions: Simplifying Spatial Complexity
Games on Grids
Axelrod views neighbors as role models whose behaviour can be imitated. Or differently: each cell being taken over by an offspring of the previous owner or of one of the neighbours, depending on who did best in the previous generation - a kind of colonization.
Wave Patterns in Spatial Games and the Evolution of Cooperation
Dynamics in human and primate societies
The evolution of cooperation in an echological context: and agent-based model
J. W. Pepper and B. B. Smuts, p.45
Anti-chaos, common property, and the emergence of cooperation
Wave Patterns in Spatial Games and the Evolution of Cooperation
Trajectories to Complexity in Artificial Societies: Rationality, Belief, and Emotions
J. E. Doran
Agent-based artificial societies at least enable us to examine the coherence of the various theories of the origins of sotial complexity that have been proposed, while including details of cognitive processing. It may be even possible to gain original insights. The complexity of the system as a whole has made it very difficult to achieve a comprehennsive understanding of its behaviour by systematic experimentation.
Artificial societies are tipically heavily parameterized, with explicit parameters merely the “tip of the iceberg” of implicit assumptions. Among the relevant cultural beliefs that habitually and, perhaps wrongly, are taken for granted are […] that rationality is the ideal and the norm.
Meeting the Challenge of Complexity
Proceedings of a Special Workshop on Land-Use/Land-Cover Change, 2001 Eds. D. C. Parker and T. Berger and S. M. Manson
Why I no longer work with agents: a challenge for abms of human-environment interactions
H. Couclelis
Evolution and the theory of Games
J. M. Smith
``But if there are three or more pure strategies, there may be no ESS. (p. 19)
Problem with his model: ``Individuals do not move far from far from where they were born. (p. 20)
Bourgeois: if owner play Hawk, otherwise play dove (p. 22)
Stable Strategies: ESS (Evolutionary), DSS (Developmentally), CSS (Culturally)
Games of Life
K. Sigmund
Must the offsprinf inherit the same strategy? Not necessarily. It need not be innate in the first place […] other ways of spreading are conceivable - spreading by imitation […] or by infection (p. 169)
a population may seem to exhibit highly variable behaviour, although all its members adopt the same strategy: if you are weak, retreat; if you are strong, escalate. This strategy is conditional, but not mixed. (p. 173)
Reinhard Selten states that in asymmetric games, there can be no evolutionary stable strategies which are mixed. Essentialy, this is because a strategy can never encounter copies of itself.
M. Smith: only surrender signals allow reliable prediction.
“Escalate if you are owner; if not, not”. An intruder should make a stand for it, since this means no real risk. “Bourgeois”: there will be no escalated conflict at all; disputes are settled as if by consent.
Recent experiences count more than past.
Sickening as it literally is, this experiment dispels the widely held belief that any stimulus can be used to teach any response.
The Art of War
The Complexity of Cooperation
Axelrod
Agent-based modelling as a third way of doing science.
The large-scale effects of locally interacting agents are calles “emergent properties” of the system.
