The always interesting Vic Davis has recently posted about appeasement as an historical tactic and game mechanic.

Appeasement as a strategy is often a bad choice. It can be a catastrophic choice under certain conditions. Perhaps the most famous from the last century is the sad spectacle of the Munich Conference in 1938 and Neville Chamberlain’s worthless paper waved in the air with the declaration of “Peace in Our Time.”

It’s funny how a single historical event can completely alter how people think of certain diplomatic strategies. Appeasement was a standard diplomatic tool for most of the modern era, and the bad name it got in 1938-39 has convinced many leaders since that you should never give in to demands, that all concessions are followed by further demands, and that anyone who does anything is just another Hitler on the rise. (See the Suez Crisis for an example of when the Munich analogy blinded Britain to what could have been a sensible negotiation.)

But as a game mechanic, appeasement has issues, as Davis notes. Appeasement can make your opponent stronger, and in a zero sum situation like a game, strengthening your opponent is the last thing you want to do. Good games also enforce fog of war, so imperfect information makes every demand a potential threat.

One problem with appeasement that Davis does not get into is when the shoe is on the other foot. If a player never has a good reason to concede to a demand, why should the AI opponents concede? If appeasement is a bad strategy for the player, isn’t it also a bad strategy for the computer?

Davis writes:

[R]efusing to appease the Dane carries consequences and costs as well. The analytical framework helps identify those costs. Often one cost is that the sword must be bloodied. If the system within which you operate includes the use of force as an arbitrator of disputes then to ignore it as an option is to handicap yourself tremendously.

My bold.

It’s not just “often one cost”, it’s “almost always the cost”. Sure sometimes there are financial penalties like upkeep and the whole guns/butter thing. But for the most part, the calculation between what is a reasonable cost for not meeting an unreasonable demand is broken. And, since war is often the only viable consequence of failure to compel, you will never see anyone make demands of an equal opponent. It’s all about bullying the weak nations, the Melian Dialog in game form. The strong do what they can, the weak do what they must. There are, conceivably, no sticks to beat the strong with.

It’s sort of unfortunate that so many of the implied threats in games are military. The only way to really penalize an actor is through war and conquest. In the Total War games, you can make demands of your neighbors, but the only way they’ll cave in is to show the flag with a neighboring army and add the “Accept or We Will Attack” codicil. The range of other compellent actions is too narrow. No threats of an embargo, or funding an enemy or fomenting revolt.

The other big part of compellent threats missing in most games is reputation. Some games will track whether or not you are a warmonger or landgrabber. But they don’t track whether or not you back up your threats. The rational actor theory of compellence (I wrote my dissertation on compellent threats, by the way) argues that past behavior of an actor is a predictor of future behavior. So if you make a threat as part of a demand, and don’t follow through when called on it, you will have trouble being believed in the future. (In the real world of high politics, this doesn’t work so nicely since the compellent threat is too rare to form solid predictive models of individual behavior. So leaders rely on other stuff like “national character”, psychological profiles or seeing into Putin’s soul.)

In a game that easily models whether a nation is capable of taking Madrid, it shouldn’t be so hard to model whether or not they are likely to. Civ IV’s personality feature is a small step forward here, except that the personalities are generally fairly rigid – some leaders are terrible neighbors, no matter what you do. There is very little “learning” about what an opponent will do, like there is in a multiplayer game; if you end up on a small continent with Shaka, Napoleon and Montezuma, you either restart or hope you are playing the Romans near iron, or Sitting Bull.

Hmmm. That was more rambling than it sounded in my head. Fill the comments with comments.