Arid
social-ecological systems:
· Ecologically, low levels of primary productivity
and highly variable rainfall
· Socially, communities tend to be sparse, mobile,
and remote from economic and political opportunities.
· These factors interact to drive further
variability in the social an economic domains.
They discuss weak and strong ties.
“In arid systems the relative importance of weak ties is likely
to be driven by local heterogeneity of resources , leading to periods of localised
resource deficiencies: the more common and unpredictable local deficiencies
are, the greater the number of ties required to insure against catastrophically
low levels of a resource. From this discussion, it can be seen that the term
‘weak tie’, whilst technically correct, may be misleading in arid networks – these
are ties which need to be persistent and which are also relatively cheap to
maintain. Accordingly, henceforth, we use the term ‘wiry tie’ – lean, but efficient in the face of resource
limitations.”
The model:
Description:
Individual nodes in a network mathematically optimize who
they share resources with. This models behavior in human societies where
sharing resources may promote survival in bad times, but also involves costs.
Each run of the model assumes a fixed number of nodes, each
having a fixed effort-budget which they can allocate to create persistent,
non-decaying ties to other nodes. The model explores optimal static network
structures.
The network is formed by nodes making individual (non-cooperative) choices about which ties to invest in, given that the effectiveness of their investment is affected by the investments of all other nodes. Resources are injected into nodes probabilistically, providing them with direct benefits, and there are indirect benefits to other nodes as resources flow via the network’s ties. The degree to which resources flow through the network ties is based on the strength of relationships between nodes. Once a node has obtained resources, those resources depreciate over time. We also assume that the resources shared do not lose value (e.g. sharing knowledge or equipment, or offering support) and hence sharing the resource does not reduce the resources of any individual. The problem for each node is therefore how to allocate its networking efforts to maximize its median expected resources held over time.
The model is quite
simplistic.
At random times a random node is chosen and resources enter
that node. Resources then diffuse non-reductively (i.e., each node does not
lose resources by diffusion, but longer paths with weaker ties lead to less
resources going any particular direction).
N = 10 nodes are
initialized in a fully connected network. Each node has 5 “tokens” it can
allocate to neighbor nodes – more tokens on a node strengthens ties. For some
resource state of the model an optimization algorithm is implemented:
A node is selected at random, and the allocation of a token
is changed to increase the strength of tie to a random node (and decrease it
from another random node). The change is kept if and only if it increases the
payoffs of both the focal and target nodes.
Only 20 different runs of the model were run for each
condition.
This model is sort of neat, but not well analyzed or
explored.
A finding that
makes sense is:
Greater resource variability (i.e., resources injected into
the network less often) led to denser networks with more weak ties. This makes
sense in terms of hedging ones bets – when resources are scarce, cast a wide
net. When resources were seen often, it made sense to have fewer connections
with stronger ties, since your neighbors would likely get hit, and the amount
of resources you get decreases exponentially as ties weaken.
No comments:
Post a Comment