simulating complex converging processes
This weekend I had a bit of time to pick up my hobby horse of worrying about ways to simulate virtual societies at a relatively fine grained level. One of the big problems that I have grappled with in the past is how to make the population pyramid come out right in the absence of a very detailed model of nutrition, health, accidents and warfare. On a more general level, the problem is one of making two models interact correctly that have complex and incompletely understood feedback aspects to them.
So, on the one hand we want the populations to be able to take actions that improve their survival, directly or indirectly in their off-spring; on the other hand, all the data we have is the overall population pyramid. How do you design a system of rules of actions and consequences that will appropriate converge to these averages?
In the past, I simply felt that the population pyramid was the most important aspect of the problem; when the virtual population members were generated, I sampled their life expectancy from the pyramid, and that was their fate, so to speak, a form of algorithmic predestination. The problem with the approach is that it makes it difficult to punish or reward behaviors immediately; how do you simulate a county-wide famine or a battle rout in this scenario? You cannot play evolution of successful strategies; the winners will be those that have been chosen to live the longest.
But then I recalled an article I had read about a strategy for making NPCs in online games more interesting vis-a-vis human player behavior. The writer proposed to support message exchange between meeting NPCs with respect to specific actions that the players had taken when encountering the NPCs. Effectively, the NPCs were gossiping to each other, when meeting on the map, about how the players had treated them--including treatments that the NPCs only had heard gossip about. As a result, whole communities would quickly turn for or against the players, depending on communication rates etc.
This communication of experiences inspired me to exchange good experiences with bad experiences in terms of life expectancy. So, in the case of agriculture, those that have the best harvest and therefore the best nutrition, get to take some age expectancy from those that have bad harvests. They effectively swap positions in the population pyramid. The hand of fate has been tempered with some "free will".
So, on the one hand we want the populations to be able to take actions that improve their survival, directly or indirectly in their off-spring; on the other hand, all the data we have is the overall population pyramid. How do you design a system of rules of actions and consequences that will appropriate converge to these averages?
In the past, I simply felt that the population pyramid was the most important aspect of the problem; when the virtual population members were generated, I sampled their life expectancy from the pyramid, and that was their fate, so to speak, a form of algorithmic predestination. The problem with the approach is that it makes it difficult to punish or reward behaviors immediately; how do you simulate a county-wide famine or a battle rout in this scenario? You cannot play evolution of successful strategies; the winners will be those that have been chosen to live the longest.
But then I recalled an article I had read about a strategy for making NPCs in online games more interesting vis-a-vis human player behavior. The writer proposed to support message exchange between meeting NPCs with respect to specific actions that the players had taken when encountering the NPCs. Effectively, the NPCs were gossiping to each other, when meeting on the map, about how the players had treated them--including treatments that the NPCs only had heard gossip about. As a result, whole communities would quickly turn for or against the players, depending on communication rates etc.
This communication of experiences inspired me to exchange good experiences with bad experiences in terms of life expectancy. So, in the case of agriculture, those that have the best harvest and therefore the best nutrition, get to take some age expectancy from those that have bad harvests. They effectively swap positions in the population pyramid. The hand of fate has been tempered with some "free will".