2c. Models: Sand and Chaos

A model made of sand
"[T]here is a fundamental tension between building a model or set of models that embody all one's knowledge in a particular subject ... and building a model to help inform decisions on a particular question..."

Paul K. Davis & Donald Blumenthal, The Base of Sand Problem: A White Paper on the State of Military Modeling 22-23

Just as sandcastles are built from sand, wargames are built from models.  Even physical wargames employ models, in that casualties and the conditions of combat are invariably modeled rather than real (the Coliseum to the contrary).  Board games are rife with models--beyond the obvious positive models of movement, combat, and damage are also the multitudinous realities that are excluded in the form of negative models: e.g., the weather never changes, weapons never misfire, soldiers never get sick, etc.

When wargame designers evaluate games, they invariably critique the accuracy (realism) and usefulness (playability) of models.  A common error in such critiques, noted by Davis and Blumenthal, is that these two criteria are directly proportional.  Another error, noted elsewhere by Steve Jackson, is that these two criteria are inversely proportional.  

The Base of Sand
As Davis and Blumenthal have noted, if the goal of a wargame is to be a predictive tool or an instrument for strategic training, building a wargame on a set of faulty models is like building on a base of sand.  Any achievements made through the simulation are compromised by the flawed assumptions on which the models are based.  Almost all the goals of simulation (prediction, model exploration, strategic training, and policy formulation) are harmed by faulty models.  Even the sheer enjoyment of a wargame will often be marred when models are not accurate.  For many serious wargamers, model accuracy is an important consideration. This can be true even in fantasy or science-fiction wargames, where the adherence to certain authorities (e.g. novels, movies, or conventions) seems to be required.[1]A picture from AWE (the advanced warfighting experiment)

Achieving model accuracy, however, is a complex problem.  Accuracy can exist in several dimensions.  Model accuracy is most often seen as predictive or factual accuracy: if a piece of artillery has a range of 800 yards in reality, it should have the same range in a wargame.  If soldiers can march 50 miles a day in reality on good roads, they should march at the same rate in a wargame.  However, quality models are not perfect replications of reality; instead, they aggregate complex phenomena into simpler systems. More useful models are often less detailed.  For instance, a squad-level wargame might have a combat model where a single gunshot wound will either kill or incapacitate a soldier.  While this might not be realistic in a strict sense of the word (flesh wounds exist), it would be foolish to suggest that a more realistic model would require players to calculate hit location, blood loss, individual stamina, history of high blood pressure, etc.  The second model would be more detailed, but this detail would force devote considerable time to subjects unrelated to the purpose of the wargame.

The decision of where and how to aggregate complex phenomena into simpler systems should reflect the goals of the wargame.  For instance, many popular commercial wargames emphasize combat and avoid the less exciting issues of supply lines and support personnel (field hospitals, mechanics, engineers, etc.). While this may be a legitimate model for entertainment, if the goal of a game is strategic or educational, this narrow focus may be inappropriate. 

In addition to the more mundane ways in which models can be faulty (lack of data points, bad measurements), three modeling problems peculiar to war games deserve mention.

Even where good models exist, chaos theory casts doubts on the ability of any model toFractal Imagery meaningfully predict the behavior of complex systems.  Chaos theory reveals how the shape of complex systems can be dramatically and unpredictably altered by adjusting events that may seem utterly insignificant at the time of their performance.  Hence, it is not illogical to suggest that an improperly manufactured nail might turn the tide of a war.[2]  Still, this does not imply that military organizations should devote their energies to the perfection of horseshoe nails.  Chaos theory does not offer a new model of prediction, instead it casts skepticism on whether some phenomena can be predicted at all.  The ultimate contribute of chaos theory to wargames might be an underscoring of the "grain of salt" with which wargame prediction and wargame-derived strategies are usually taken. 

The Edge of the World
With the exception of role-playing and Pentagon-style political-military games, most wargames focus on particular events with both spatial and temporal limits.  These limits restrict the movement and strategy of players.  In board games, this means that at some point players may find themselves trapped by running up against what is literally the edge of the (game) world.  For instance, in a game played on a map of North America, troops can not be moved to Venezuela.  Though less visible, other aspects of war games create similar barriers by preventing innovative tactical approaches outside the scope of the game. For instance, cease-fire talks are seldom an option in wargames.  These limitations have two implications.  First, in so far as wargames are used for strategic training, players are trained with rule-constraints that are not present in reality. Second, from the point of view of policy formation, world edges often make separate things which are in reality inseparable.  From a philosophical perspective, the limitations on space, time, and action frequently found in wargames may reinforce the notion that war is a discrete and separable enterprise, when in fact it is often inseparable from the non-military events that create and sustain it.

Moebius Effects
Finally, a significant vulnerability of wargame models is that they run the risk of simply "proving" that which they assume.  A wargame can only prove its own models wrong when it produces counter-intuitive results (e.g. Finland conquering Europe).  Yet this is precisely the kind of result, that, when encountered, adds some new knowledge.  If a simulation's results are rejected or accepted simply on their ability to agree with intuition, the simulation itself arguably has no value.  A Moebius Strip

The Moebius problem is made even more difficult in historical wargaming, where historical battles are replayed.  Often, these games are evaluated on their ability to produces the "correct" historical result.  Yet if this desire to produce a given outcome influences the formation of game models, it undermines the validating power of the simulation itself.  An example of a Moebius effect is demonstrated by the board game Origins of World War II by Avalon Hill, designed by Jim Dunnigan.  In attempting to explain the political situation that gave rise to World War II, Dunnigan created a game where all players operate with conflicting objectives and only one player can win.  Such initial conditions seem to guarantee that history will be repeated.[3]

Previous Page | Site Map | About the Site

1(back). For example, to offer hard-core Trekkies a Star Trek game where traveling over light-speed was impossible would be dangerous -- as would offering a game where traveling at Warp Factor 15 was commonplace.

2(back).  "For the want of a nail, the shoe was lost; for the want of a shoe the horse was lost; and for the want of a horse the rider was lost, being overtaken and slain by the enemy, all for the want of care about a horseshoe nail." -- attributed to Benjamin Franklin.

3(back).  According to Thomas Allen, Dunnigan found the game to have predictive historical value: "the way the game had been played had given [Dunnigan] an insight into war and human nature." Allen, Wargames at 94. Actually, the game merely provides insight into the game.