Conflict simulation, peacebuilding, and development

simulation, and simulation

halPaxSims is devoted, of course, to simulations involving human participants. However, as Drew Conway from Zero Intelligence Agents reminds us in a recent comment here, researchers also make use of agent-based modeling to try to better understand conflict processes such as insurgency and counterinsurgency. In these sorts of simulations, there are no human “players,” only a set of computational-models/actors that interact, and react to the changes that these interactions create. As computational power has grown (and become much cheaper and more accessible), moreover, such modeling can become increasingly sophisticated.

Such models can be very useful in understanding how, for example, changes in hypothesized relationships between variables at the agent level might change aggregate outcomes. I must admit, however, to a certain degree of cynicism about how robust their findings are, given the inherent difficulties of quantifying (and specifying quantitative relationships between) elements of human behavior. I would be especially concerned, in a training and educational setting, about the extent to which such simulations hide potentially dubious assumptions about that behavior “under the hood” so to speak, beyond the view of more casual observers.

The same potentially applies to “artificial intelligence” in human-computer simulations. While having some simulation responses driven by AI (rather than human) players has many advantages, it is important that participants understand the limits of AI behaviour, and avoid the temptation to “game the game” and exploit AI vulnerabilities (something that anyone who regularly plays first-person-shooter computer games will readily understand).

Nonetheless, I strongly recommend Drew’s website for those interest in quantitative approaches to conflict issues (and not simply ABM simulation of them). There’s much of interest there.

2 responses to “simulation, and simulation

  1. Gary Milante 12/06/2009 at 6:38 pm

    Also, from my perspective, I think one of the under-appreciated benefits of using a live roleplay simulation to teach is teaching people how important the non-rational, emotional, social aspects of decision-making are to what solution is reached.

    In Carana we give the same information to three teams of players at the beginning of a four day simulation and then part of the learning comes from demonstrating how different the solution space ends up being for the teams, though everyone has nearly identical preferences and they are just a bunch of technocrats trying to reach an optimal solution. Teams can often trace back their own solution space to relationships between the government/international actors, to how they divided up their time, to how they ordered their decision-making or maybe just to having a really pushy President or Minister of Defense.

    It might be difficult to model this kind of interaction – do some artificial agents have louder voices? Are some better looking? Do some have training in mediation?

    I know it is heresy for an economist to admit it, but much of the outcomes of these complex political systems is really as dependent on who is playing the game as much as how rational they are.

  2. Drew Conway 14/05/2009 at 4:49 pm

    Rex, thanks for the plug. I have been following your blog with great enjoyment since we were on the CTLab robotics symposium together.

    Keep up the great work!

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: