And then you simulate a zombie outbreak in your virtual society.
A REALISTIC simulation like this on a global scale would be ridiculously useful in multiple fields. And while several groups have been trying to do something similar, there just isn't enough data for it to work properly at the individual agent level.
(At higher levels you have for example http://www.gleamviz.org/2009/09/us-winter-projections-mitiga... )
Step 3 in the simulation is to make it a game.
http://en.wikipedia.org/wiki/Agent-Based_Computational_Econo...
and it's fun, but somewhat useless. Without analytical results, simulation results tend not to be insightful. You may see that something happens, but you don't understand the causal relations.