*a) *Two possibilities for the mechanisms of interactions

- Inter-individual tagged systematic interactions;
- Inter-population tagged random interactions.

- Given that the firms try to
*manage*their social networks, the first approach seems more relevant. In this case, each individual firm communicates with each of its neighbors. - The second approach (used by the evolutionary game theory) can nevertheless be relevant in some particular situations: market interactions between firms using different standards or from different countries.
- I will focus on the first approach.

*b)* Two kinds of models of neighborhood

- Spatial models,
- Graphs.

\

- In the spatial models, the structure of neighborhoods are inscribed
in a particular topology:
- line,

- circle (closed line),
- lattice,

- torus (closed lattice),
- honeycomb (Loeschian model)...
- This topology conditions of course the dynamics of the model. This is the weakness of this kind of models! But we can control quite easily their influence and workings, this is their force. They are also very easy to program (mainly by arrays of firm objects).
- The graphs do not depend on a particular topology and possess some general properties (given by the graph theory).
- The density of their connections can be modified playing on one parameter (the initial probability of connection). But these connections are stochastic and their structural influence on technology dynamics cannot be analyzed..
- There is clearly a
*trade-off*here between the richness of intuitions and the generality of results.

\

Sam 18 Oct 09:10:57 DFT 1998