Anyone who has played a real-time strategy game probably ponders this question when preparing for a fight: Is it better to have a huge number of relatively weak units or a smaller number of extremely powerful ones? But the question predates computer games and virtual war machines. Theoreticians started tackling the problem in response to the real-world carnage of World War I, where it was posed as a question of army size versus fighting strength.

In the years since, the ideas developed for human warfare have been adapted to apply to non-human combatants, most notably social insects, which can mobilize large forces when engaging in combat. In the early 1990s, researchers who studied ants argued that finding the right balance between force size and capability depended partly on the environment. Complex environments, they theorized, favored smaller numbers of capable units that could occupy key terrain. Simpler environments, by contrast, would allow massive waves of weak units to Zerg rush an outnumbered opponent.

That idea has been difficult to test empirically. But three researchers from the University of Western Australia (Samuel Lymbery, Bruce Webber, and Raphael Didham) have now put it to the test, using a combination of Age of Empires and live ants.

Theory meets combat

The first sentence in the results section of the Lymbery, Webber, and Didham paper is striking: "To illustrate the potential effects of environmental complexity on group combat in an idealized setting, we first arranged simulated battles using the real-time strategy video game Age of Empires II." A simulation is somewhere between theoretical models and real-world behavior, so it's not a huge step forward. But, as we'll see, the combatants in Age of Empires are easier to rate in terms of combat strength—and much more cooperative when it comes to fighting.

To model a high combat strength, the researchers chose Elite Teutonic Knights, which has the highest individual combat strength in the game. Opposing them were Two-handed Swordsmen, a relatively generic combat unit. Balancing them out is the fact that you can load up as many of each unit as you want. The researchers also could pick terrain with varying degrees of complexity.

As expected from their battle ratings, every one-on-one Knight vs. Swordsman battle ended with the Knight winning. And that held true for up to four Swordsmen; once five or more Swordsmen were loaded in, the Knights invariably lost.

The researchers then ran a series of scenarios with a team of nine Knights facing off against a variable number of Swordsmen. And by various measures, the elite Knights did better in complex environments. While the nine Knights started losing against 50 Swordsmen in simple environments, they could still beat nearly 70 Swordsmen in complex environments. It took fewer Swordsmen to kill their first knight in simpler environments, and the survival of the weaker unit was better in simpler terrain.

So all of this was in keeping with theoretical ideas. But it still doesn't constitute real-world data.