At World DipCon VIII, I played a type of game that I (and most of the British and European players) have never played: variable length tournament Diplomacy.

Diplomacy is a long game. Played properly, it can last twenty game-years or more, and playing face-to-face, a game-year usually takes around forty minutes. This makes for a game that is rather difficult to organise for a tournament. Tournament rounds usually only have six or seven hours of playing time available, and so fitting in a full game is not possible. Speeding up the game is not an option, as even forty minutes per game year can feel rather rushed at times. And so the only other alternative is to limit the number of game years that can be played.

However, this causes problems. Because the end of the game is false, players play to maximise their position at the end of the available time, rather than playing as they would play in a full game. Last-turn stabs are common, with players breaking alliances that they would have needed to continue were they to survive to the end of the full normal game, simply to gain an extra centre or two at the false-end to which they are playing. Larger powers can even be allowed to reach a position from where, given more time, they could force a win.

There have been attempts to solve this problem before. One idea was to score the game based on the centre count a year or two before the year in which the players were told the game would end. However, once players caught onto what was happening, this stopped working. Also, there is the idea of selecting a random game-year in which to end the game from a list of possibilities, and announcing it to the players only when they arrive at that game-year. However, this also causes problems, especially if a year towards the end of the available range is chosen, as players can still play the last couple of turns strictly for centres, with the panic that comes with knowing that the end of the game is near.

And so, I decided that what is needed is a system in which the end of the game is more like the end of a real game. In order to do this, I determined there are two requirements that are not in any of the previous methods. Firstly, the end of the game should not rely on any hidden information, such as the unannounced scoring method or the randomly selected end-point. Anything along these lines is subject to attempts by the players to outguess the system. Second, the point at which the game ends should be decided by events on the board, and not by a single fixed decision for all games in the tournament.

So I came up with the following plan. Like a full game, the game is played until either somebody wins, or until a draw is agreed by all survivors. The change is in what is required for someone to win. In order to win, you must simultaneously fulfill the following three conditions:

1. Have more centres than anybody else on the board
2. Have more centres than you had the previous game-year
3. Have at least the number of centres specified in the table below:

   Game-Year:	1904	1905	1906	1907	1908	1909	1910	1911	1912	1913+
   Centres to Win:	18	17	15	14	12	11	9	8	6	5

The first of these three conditions is obvious. You cannot win unless you are the single biggest power. The second condition is to force you to grow to win -- you cannot just sit on a number of centres, and wait for the winning condition to come down to your level.

The third condition is what makes it all work. As the game years go by, the ease with which you can win becomes greater, and so the game should finish in reasonable time. Some games will finish by about 1907, and the vast majority should be over by 1911. As players approach the winning conditions, then, like in a full game, it should become the objective of the other players to stop worrying about their own success, and instead play to stop the leader from getting any bigger. If they do not manage this successfully, then they will lose, and if they do manage to stop the leader, then getting a draw vote accepted should not be too much of a problem. Thus we see that the perceived disadvantages of a "fixed-end" (be it clock-time or game-year) are avoided.

Of course, some tournament directors may want their game-rounds to run slightly shorter or longer than that recommended above. In order to achieve this, the required centres table can be adjusted by adding or removing 18's at the beginning. So for instance, to make the game three years shorter, the following table would be used:

Once all the games are over, the next job is to put all the players from all the games into a rank order, and decide on the tournament winner. For this, all game winners beat all draw participants, and all draw participants beat all losers and eliminated players.

Winners are subdivided by winning year, earlier beating later, and then by number of centres. Should there be any ties on both these criteria, they are then split by centre count in the previous years, working backwards until a difference is found. The player with the higher centre count at this point is the winner.

Participants in a draw are ranked by the number of people in the draw, and then subdivided by number of centres held.

Losses and eliminations are ordered by number of years played and the player's number of centres at the end (if any).

Finally, there is one other problem issue in tournament scoring that I do not address here: how to rank players in tournaments having more than one round. Simply ranking results is sufficient for one round, but when there are multiple rounds, then there must be some way to combine results to give each player an overall score. However, this requires that some decisions be made and agreed to as far as what should beat what. For instance, should a solo win and an elimination beat two three-way draws? This is a separate problem in itself, and so no attempt is made to solve it here.

Comments on the above system are very welcome. Will it work better than the current systems? Will it solve the problems it sets out to solve? Will it make for a better game? I would appreciate some opinions on this.

David Norman (David@ellought.demon.co.uk)

If you wish to e-mail feedback on this article to the author, and clicking on the envelope above does not work for you, feel free to use the "Dear DP..." mail interface.