Well designed simple models of war have another quality, the ability to recreate a number of variable outcomes taking into account the variables that created such outcomes without devoting too much energy to demonstrating the mechanics of said variables. As such, simple models can be just as accurate as more complex models.

The increased complexity does not always produce a more accurate result, and in fact tends to create a less accurate result by introducing an increased range of error induced by the increased number of variables and their relationships.

Simply translated: More detail = more opportunity for errors. The more detail you have, the higher the probability of a game failure somewhere during the course of play.

Accurate games, simple, or complex, can be enhanced any number of ways, through redundancy, through design simplification, through degradation analysis, through worst-case design, and through transient analysis.

In general, the reliability gain for additional redundant elements decreases rapidly for additions beyond two or three redundant elements. Care must be taken that reliability or accuracy gains caused by increased redundant detail are not offset by the additional interruptions created by that detail. i.e. conflicting rules lookup, additional pieces with a very limited function that could be included as something else, etc.

Design Simplification seeks to test whether a rule is redundant, and one does this by dropping the rule in question and seeing how it effects the game as it is played. The trick here is to simplify only so long as performance is not altered making the game unreliable or inaccurate, or perceived as inaccurate. There is a limiting point to simplification. "Make the rule as simple as possible, but no simpler than that, hey?"

Degradation Analysis looks at a game through the span of its' designed lifetime. First one needs a burn-in period, and in the wargaming world this is done through extensive playtesting prior to game release. Many "broken" elements will make themselves apparent with playtest groups. It is important to use both control groups and "wild" playtest groups for this. One also puts some thought into designing the game in advance to be flexible to change so that it doesn't "break" if there is a shift in focus on what is important in a game over the years. One way is to put in some "change" compensation rules in advance, and another is to make it possible to add new rules to compensate for unforeseen changes in the environment. Finally, one should design the game for a fixed, finite lifetime in advance. Designing for infinity means you'll be infinitely designing...

Worst Case Analysis takes one part of the game and presumes that it will break or is "broken". Can the game still be played without the rule or piece in place? If the failure occurs, can the game still be perceived as an "accurate" model of what happened, or what could happen?

Transient Analysis is also known as stressing the game, in wargame design this is best conducted during playtest by encouraging extreme interpretations of the rules, and deliberately overstepping the bounds of the rules. (Anyone seen this at a convention lately?) One can also vary the tempo and pace of the game requiring the turns to be played at faster or slower intervals to see if the game will hold up under the changing play conditions.

Aside: All I did was apply basic reliability engineering principles to wargame design. A good book on this is Electronics Engineers Handbook 3rd edition 1979 McGraw-Hill ISBN 0-07-020982-0 More information on good design practices can be found in chapter 28 Reliability Design & Engineering.

This article appears in MagWeb (Magazine Web) on the Internet World Wide Web.
Other articles from military history and related magazines are available at http://www.magweb.com