This got me thinking about how Computer Simulation is arising as a new type of experimental approach. This is especially powerful when a simple set of principles can interact in complex and unpredictable ways. There can be severe limitations to reasoning about such things symbolically or semantically. Simulation, gives a direct way to test if our understanding of a system matches what it actually does.
This is, it seems, one area where the conversation on population genetics is impoverished. There is a host of people reasoning semantically about population genetics. However, as a rule, population genetics is not-intuitive. A simple set of rules produces deeply non-intuitive behavior. The field, for this reason, has come to rely on simulation to demonstrate that our reasoning about genetics matches the reality of how it actually works.
This article on Computer Simulation in Science is fascinating, and very close to the work I actually do for a living:
In Engineering we do simulation via finite element analysis. One important part of using simulations in engineering is to validate the simulation with real world testing. The correlation is usually very carefully established.
Of course. Sometimes we will model things by simulation. Then we will model them in an experiment, looking to demonstrate a correspondence between the two. This is a very common approach.