Information Technology Reference
In-Depth Information
Reproduce: well-fed human agents can produce one child ( p =0 . 01). The
energy value and culture are shared with the child. Other variables are set to
default values.
Update job: Periodically, humans change their job by updating the job vari-
able. Harvest can be selected by anyone without restrictions. Institutional jobs
(protect and harmonize culture), can only be selected if offered by the social pol-
icy SP . The job selection is simple: the agent tries first to get an organizational
job. If none is available, it selects harvest. To avoid erratic behaviors, an agent is
committed to a job for multiple rounds (randomly between 20 and 70 rounds).
Get food at village: The agent moves towards its village. If the agent is at its
village and the food value of the village is high enough (bigger than max energy ),
the human agent performs an eat action.
Do job: The human agent performs its job .
Meet other humans: each human agent engages in some social interactions
with another randomly selected human agent from the same village. If these
two humans are from the same family or under control of a loyal policeman,
they discuss with each other. They can mix their culture vector if their cultural
distance is close enough (according to the Axelrod cultural model [1]). This
mechanism is important for our purpose: more agents lead to a more important
cultural divergence. If the food is sparse, the two agents come into conflict. One
of them is hurt ( hp is reduced by 1) and complains about the lack of physical
protection ( social dissatisfaction is increased by 0 . 2).
Two randomly-drawn agents belong to the same family with a probability
s/
is the village size. Two agents
are under police control with a probability P.L v ( police ) /
|
v
|
where s is the family size (set to 30) and
|
v
|
where P is the num-
ber inhabitants a policeman can control (set to 10) and L v ( police )isthesum
of policemen loyalty . These mechanisms can of course be implemented in a dif-
ferent way. For instance, in a former version we represented explicitly family
networks or policemen can patrol and defend their nearby space. The core prop-
erty (whatever the complexity of its implementation) is that a population size
increase triggers observable organizational failures (observed by o 1 and o 2 ).
Each round, each agent acts only once in the environment: during the “get
food at village” or the “do job” phases. Available actions are: move :movesa
distance of 1. harvest : sets the local full patch to harvested and fills the agent's
bag with harvest tech level units of food. harmonize : merges 20 items of the
culture of an agent to culture v (for sake of simplicity we did not forced agents to
be nearby to make this exchange occur). eat : the agent removes max energy
|
v
|
energy units of food from its village and sets energy to max energy .
Village Rules. Village rules (Figure 3) represent orders given by chiefs (if any).
They are split in 2 phases: organization creation and the social policy manage-
ment ( SP ). If no police organization is present, agents evaluate their desire for
physical protection social dissatisfaction . Each round, a vote is performed to de-
termine if the police is required: agents approve if social dissatisfaction> 0 . 7. If a
majority is reached, then a police institution is created and a leader (the human
 
Search WWH ::




Custom Search