Tabela de conteúdos
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. Constriction: 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 cooperative and almost always punishing behaviour. He also studies norms between different groups (white and black, for instance).
Choosing sides (and) Setting Standards
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 satisfiability. Each pair of agents has a symmetric satisfiability value. The main purpose of ABM is not prediction but a deeper understanding of how fundamental social process operate. There is a Nash Equilibrium because if someone individually exchange its location, it will be worse to him.
The author uses real data of the European countries in the Second World War.
Building new Political Actors
It takes as given the existence of lower-level actors, and generates higher-level actors from the interactions among them. […] The heart of the model is a tribute system in which an actor can extract resources to extract still more resources.
One agent may ask another for demand or fight. If it would cost less than paying then fight is the best option. A combat between two nations implies in each one loosing 25% of the other's wealth. Agents can join or leave groups according to their decisions (subservience, protection, friendship, hostility). 8 agents in a one-dimensional circular grid, each on having two neighbours. The model has different results in each run, and the author analyses 3 of them.
Disseminating Culture
Replication of ABM
Resources for ABM
deprecated