Discover more from David Friedman’s Substack
Games as Training
At a time so long ago that long distance calls were expensive, I was in Europe, my parents were in America, and we had to coordinate our plans. At some point while doing so it occurred to me that this was a problem that I had spent many hours learning to deal with.
By playing bridge.
Subscribe for free to receive new posts.
The score in bridge goes to the partnership not the player, so bridge is, from the standpoint of game theory, a two player game in which each player is two people who are not allowed to talk to each other. Communication between partners is permitted only in bids, moves that double as messages. Each partnership has a system, an elaborate code, defining what information about the player’s hand each bid communicates. I had spent many hours playing bridge, my family being a convenient foursome, hence many hours being trained in coordination with limited communication.1
Having realized that bridge could be viewed as a training exercise, it occurred to me that the same is true of other games. Chess is training in tactics and strategy, useful skills in business, war, life. In poker as in bridge the moves in the game communicate information — but between opponents not partners. The problem for each player not how to communicate but how not to, how to avoid revealing true information about his cards, better yet signal false information, while extracting true information about the cards of opponents from their moves. That makes poker training for bargaining, whether in business or diplomacy.
It is also, I am told, good practice in probability theory.
The game Diplomacy, played multiple times by the same players, teaches the value of commitment, of keeping your promises. Also of not trusting other people unless you have reason to believe they keep theirs.
Back when our children were young we got them Game Boys with Pokemon cartridges. They logged something like eighty hours a month, perhaps more, on those cartridges for many months thereafter, more time and more attention than I, at a similar age, put into all of my schoolwork combined. The skill they were learning, how to find their way around a world and accomplish goals therein, was in one sense useless, since the world was a fictional one. But being able to find one’s way around a new environment and accomplish things within it is a useful real world skill.
Watching a conversation on DSL, currently my favorite online forum, I recognized the influence of another game. The participants were classifying people, real or fictional, by the nine-fold alignment grid from Dungeons and Dragons. D&D is enough a part of the culture I am immersed in that it makes sense to me as well to classify people that way. A friend whose instinct on seeing a button labeled “do not push” is to push it is chaotic good, perhaps chaotic neutral, certainly chaotic.
Playing D&D teaches a classification scheme, a collection of mental tools, a way of thinking. A language.
From time to time I find myself in an online exchange with someone who is not competent to keep up his side of the argument. I know that there is little point to spending time and effort arguing with him but do it anyway, seduced by the temptation of an easy win. It is a base pleasure, one I should not indulge myself in, the same base pleasure as beating up on a non-player character in World of Warcraft or continuing a game of Strategic Conquest2 past the point where I have obviously won. Thinking of it that way helps me see what I am doing and why I shouldn’t do it.
Labyrintspiel is a game where you guide a metal ball through a maze punctured by holes. The essential skill is moderation — tilt by too much to get past one hole and you overshoot and go into another. It is good practice for driving on icy roads, which requires the same skill. Or running the Federal Reserve.
Doing It On Purpose
All of these are games invented and played because they are fun. A different category consists of game designed to teach a skill.
My favorite example is Robowar,3 a game in which players program on-screen robot tanks, using a simple computer language. You program your tank, your friend programs his, you load them into the computer and watch them fight, each following its program. The first time you do it your tank runs into a wall and batters itself to death. Repeated revisions, interrupted by debugging of the program, eventually produce a tank that may, with luck, spot the enemy, shoot at it, notice when it is being hit and take evasive action.
Robowar makes computer programming, a high level skill that can get you a job, into a children’s game.
There are two such games that I invented, neither of which, unfortunately, exists. One is a version of World of Warcraft modified for language instruction. When you are at level 1, non-player characters sometimes speak to you in French, always where the context makes it easy to guess the meaning. As you go up levels, more and more of what the game says to you is in French. When you get high enough, you have to use French yourself, single words and eventually simple sentences, in order to get the game to respond to you. If you get good enough, you switch to a French server.
Language by immersion, with no need for a passport or airline ticket.
There are two ways in which large social structures are built. One is conquest. The other is mutual advantage. Many computer games embody the first. I wanted a game that would embody the second.
In Hansa, you are building a trade league.
The following is from the instructions I wrote more than twenty years ago, back when a friend was trying to program the game:
A league is made up of cities located on a map. Each individual city has a population, a production function, and a utility function. The production function uses one input, labor, to produce any of several goods. Producing one unit of a good requires a particular amount of labor, and using N times that amount of labor will produce N units. The cost per unit will be different for different goods within one city and for the same good in different cities.
A city's utility function represents the total happiness of the inhabitants; it depends on the goods they consume. The additional utility from consuming one more unit of a good decreases as the amount of that good consumed increases and increases as the amount of any other good increases. The utility function is identical across cities. A city’s “autarchy level” is defined as the highest level of utility it can achieve without trade.
Cities within a league can trade with each other. Transportation costs (in labor) depend on the particular city pair; typically transportation cost is larger the farther apart the cities are and is much smaller by water than by land.
Trade, production, and consumption are organized by the player; he instructs each city in his league how to allocate its annual labor among different goods and which of the goods it produces should be shipped to other cities. Trade occurs only within a league. A league may at any time send trade missions to try to persuade non-member cities to join it.
A city bases its decision to join a league on its observation of the benefits that the league confers on its member cities. The probability that the invitation to join will be accepted is an increasing function of the average utility gain (utility above autarchy) of the member cities of the league, weighted by how close they are.
Cities can not only join a league, they can also leave one. One reason for leaving is that a city concludes it would be better off on its own. This may happen in any year that the city's utility is lower than its autarchy level—the utility it would have if it engaged in no trade at all. The larger the gap, the higher the probability that the city will secede from the league.
A second reason for leaving one league is to join another. If a member city of a league receives a mission from a different league, the probability that it will accept is an increasing function of the difference between the weighted average utility gain of the cities in the inviting league and the utility of that city. Thus a player who wishes to defend cities on the “frontier” between his league and a neighboring league may do so by keeping them happy—arranging trade so that their utility gain from membership in his league is large.
One reason I wanted Hansa was to promote the idea of mutual advantage over conquest, market over hierarchy. The other was to let players discover the principle of comparative advantage. In deciding what cities to try to pull into your league you not only want cities that are good at producing things your cities are bad at, you also want cities that are bad at producing things your cities are good at.
Hansa was part of the Living Paper project, which started when I wrote three computer programs to go with my Price Theory, was intended to expand to more such programs plus Hansa, died more than twenty years ago. If anyone wants to revive it as Open Source, the materials are on my web page.
The first law of which is, as all good bridge players know, “Trust your partner.” It’s a pretty good rule in other contexts as well.
Strategic Conquest may be the first computer game that I spent a lot of time playing, more than thirty years ago. Playing it is how I discovered the proof of a really addictive game — when you stop playing to go to the bathroom, it’s because you really have to go. It is now available online via emulation and turns out to be still addictive even if, perhaps because, it is much less elaborate than more modern games such as Victoria 3 or Hearts of Iron.
An earlier version of the idea was called RobotWar and ran on the Apple II.