Conflict simulation, peacebuilding, and development

Daily Archives: 02/02/2009


NATO Wants Sim Afghanistan to Test War Plans

Noah Shachtman, Wired: Danger Room, 2 February 2009

NATO commanders in Afghanistan want a virtual version of the country, to test out battle plans and forecast future unrest.

Afghanistan’s often-explosive mix of tribal, ethnic and religious power politics has been catching outsiders off-guard for the last couple-thousand years. This time around, America and her western allies are trying two controversial, competing approaches, to prepare for the surprises. One embeds in combat units social scientists, trained in making foreign cultures more understandable. The other dumps everything that’s known about the country into a software model — and then watches what develops in this Sim Afghanistan.

Last last week, NATO began its search for for the newest “simulation capability.” This one should “be able to model the Afghanistan engagement space in the Political, Military, Economic, Social, Infrastructure and Information (PMESII)

 domains,” a call for white papers notes. With all that information in hand, war planners can then “assess and validate how specific future events or actions could impact on the current situation through the creation and simulation of a hypothetical/simulated environment.”

According to the RFI that NATO has issued, the simulation is intended to provide ISAF with “a simulation capability to assess the potential projected outcomes of possible courses of action, resulting from the planning process”:

This will allow ISAF operational planners to assess and validate how specific future events or actions could impact on the current situation through the creation and simulation of a hypothetical/simulated enironment. Because of the complexity of the Afghan environment and the interaction with people and organizations, the capability shall include aspects of Political, Military, Social, Infrastructure, and Education (PMESII) domains as well as other aspects of the environment including terrorism and insurgency.


"Minesweeper Afghanistan"—I assume they're looking for something a little more sophisticated than this...

As with Noah Shachtman at Wired and several military experts quoted in his piece, I’m rather doubtful of the ability to computer-model the “non-kinetic” parts of COIN operations with any meaningful degree of accuracy. This could be problematic, given that the most difficult part of COIN and stabilization operations are precisely how military operations affect local attitudes, local structures of power, the impact of aid, the nature of local tribal structures and coalition-building, and so forth. If one simplifies this (or fails to allow for such unintended consequences as collateral damage to civilians), it seems to me that one risks operations planning built on very dubious foundations.

hat-tip: Mike Innes at Complex Terrain Lab

using “unrealistic” simulations as a learning tool

tropico_coverartA few years back, for my introductory political development course, I set as a possible book review assignment the computer game Tropico. In the simulation, a player assumed the role of the ruler (democratic or otherwise) of a stereotypical “banana republic,” and attempted to survive and prosper. To do so, they had to make decisions about government taxation and spending, build political coalitions, compete (or commit fraud) in periodic elections, forge external alliances, avoid coups and revolution, and even imprison or liquidate opponents. Depending on the persona the player had established, success might be measured in terms of development, citizen satisfaction, or the amount of money diverted into a Swiss bank account. Many of the choices were fun (Should the old fort be used as a tourist magnet, or a prison for political dissidents? Should I pay the palace guards more to prevent a coup?), and if you failed badly you might have insurgents attacking from the jungle or see hundreds of little tiny angry citizens rioting outside the presidential palace. It ran well on both Macs and PCs, and I was able to pick up loads of cheap copies for the library from eBay.

Tropico was far from a realistic simulation. Most notably, it modeled (as many of the SimCity-type simulations do) a very statist economy. El Presidente made all the major economic choices: what kinds of crops to plant, what kind of housing to build for workers, what factories and other investments should be made (and where), and so forth. In short, it looked more like Cuba than anywhere else. 

The “unrealism” of Tropico as a simulation, however, was the very point of the assignment: students were asked to critique the game, to identify its analytical and ideological assumptions about politics and development, and generally utilize their own knowledge of these issues in reviewing the software. 

All-in-all, the whole assignment worked rather well, with some top-notch reviews among the 150 or so students who chose this option. I haven’t repeated the exercise in subsequent years, but certainly might do so if I ever run across another game/simulation that would serve as well.

when is a simulation not a simulation?

oilI came across this today: Oil Shockwave: An Executive Crisis Simulation. It is the report of a 2005 “simulation” designed to explore the policy implications of terrorist attacks and other events that cause a rapid increase in world oil prices. I had several immediate thoughts:

  1. I should get corporate funding like that for my simulations. Sheesh, talk about slick… ceiling-height back-projected crisis graphics! Expensive furniture! Glossy simulation reports! This tops even Gary’s fancy plasticized team badges for Carana…
  2. Oil prices might spike above $100 a barrel! OK, so we’ve learned that since 2005.
  3. This isn’t really a simulation. 

Instead, it seems to have been a preplanned scenario, in which pauses in the various stages were then used to foster a policy discussion among the high-level participants. It is not at all clear, however, that there was any interactivity, or that anything the participants said or did actually affected how the scenario unfolded.

This isn’t necessarily a bad thing, of course: such a  format can be used to encourage participants to think about alternative futures and identify the associated implications and policy options. It has the advantage that one doesn’t need complicated game mechanics, and there are no rules or umpires for participants to dispute.

A full-fledged simulation, however, involves both an iterative dimension (in that decisions taken at T1 affect the context for future decisions at T2), and an interactive one (in which the impact of decisions is contingent on decisions made by others). Both, of course, much more accurately model the real world. However, if the underlying simulation mechanisms are weak, the result can be an unrealistic and unsatisfying disaster.

simulation links at SWJ

From time to time, the issue of using simulations to train for irregular warfare, COIN, and peace operations comes up at Small Wars Journal. I thought I would flag a few of the relevant threads for those who might be interested:

%d bloggers like this: