by Andrew Feather
Have you ever noticed after playing in a few games of Diplomacy that certain players are remarkably consistent in their behaviour? What this article attempts to show is that it is possible to predict at the start of a game, if you have enough detail on the past games of your opponents, the likely trend the game is going to follow, and to adapt your own strategy accordingly.
The first thing that needs to be done when you look at a past game is to make a table showing the losses and gains made by each power, and to calculate the arithmetic mean (AVG) and standard deviation (SD) for each year that they were in the game as a participant. here’s a table I’ve done earlier for Ipsos, which was a game in Spring Offensive that has recently finished.
|IPSOS||01||02||03||04||05||06||07||08||AVG||SD||Game Rank||AVG Score Rank|
What the AVG indicates is the average number of builds per year, the higher the number, the more expansive that player is at gaining units. the prize for elimination, or a net loss of units, is a negative score. This is because that player should have diplomed better to improve his starting position in the game. Also, the longer the game goes on, the lower the eventual AVG will be for those players with positive scores. This indicates a slower game, either because of lack of opportunities or because a stalemate position developed which froze the game down. For those players with negative scores the longer their game goes on, the lower the net negative AVG, making it more of an incentive to stay with the game for as long as possible.
To summarise what this means in terms of strategy, if you are going to win the game, do it as quickly as possible splitting the opposition so that stalemate lines don’t develop. If you are losing the game, try to make it last as long as possible so that you reduce your net negatives, and maybe get yourself enough time to get yourself back in the game.
The standard deviation (SD) score tends to indicate the players “aggressiveness” factor and their willingness to take risks. It measures the spread over the biggest gains and the biggest losses compared with the AVG. It is useful to compare the AVG and the SD together to get an overall idea of playing style of your opponents. Low SD players expand and decline in a regular flow, E.g. 0,1, 1, 0, -1; whereas high SD players go in “start-stop” bursts, E.g. 0, 3, 0, -2, 1. Putting these two factors together suggests a diagram along the lines of:
|This indicates a player who grows in size steadily, taking opportunities as they arise, but making sure his position is consolidated.||This indicates a player who aggressively goes for growth gambling on the hope that the opposition cannot organise themselves against him.|
|This indicates an overly-cautious player who lets events carry on around him while he makes sure non of it affects him.||This indicates a player who adopted a risky strategy for growth, but was unable to stop the opposition from uniting against him and so suffered the consequences.|
I have to admit that this model is far from complete. There are still some questions that remain; for example, do certain countries suit a certain playing style? Would the fact that Austria tends to get eliminated early on further discourage people from playing that country? Do certain countries require different weightings? Hopefully this will become a matter of discussion, and someone may try to go through the statistics of outright winners for each country to see if there is a certain style of player that suits a particular country, or enable a player to alter his style of play to suit that country’s requirements.
To give an example of the interpretations possible from this exercise let us consider the game Ipsos, which had its end-game statement publish in Spring Offensive No.24.
The first thing to note is that only Germany and Italy improved on their starting positions, everyone else ended worse than they started with or were eliminated. Germany in particular expanded rapidly, averaging two builds per year, in a race with Italy, who was averaging one build per year. Both these players had very aggressive strategies which blew away the cautious players, like Russia with a SD of 0.707, and forced the rest of the players into adopting risky strategies, SD’s greater than one, which didn’t quite work out. Germany and Italy dictated events efficiently between them, and prevented the other players from co-operating effectively to stop them from forging ahead in the game. by splitting the opposition Germany and Italy were in a two-horse race. The risk attitude of the German player is well illustrated in 1906 when he gained 6 builds, one from Italy, which although it allowed Italy to gain 3 builds in 1907, gave Germany the overwhelming advantage he needed to go for a win. Nevertheless it could so easily have gone wrong for Germany in 1906. It is debatable as to what the best strategy would have been to tackle the dominance of Germany and Italy in this game. Fight fire with fire in an aggressive style? or fight fire with water in a cautious style to slow the game down and cut down the opportunities for German/Italian expansion? As they say, Diplomacy is a game of psychologies and personalities!