The Lack of Diplomacy Opening Theory

By Jamie Dreier

This is an essay I wrote for and distributed to players in my Newbie game, Younguns. I had given the players some standard (Gamers' Guide Second Edition) comments on their opening moves, but said that I would not comment much after that. I decided to explain why not: why there is nothing general and standard and useful to say about openings beyond the relatively simple remarks from the Gamers' Guide. 

I believe there are two reasons. It is instructive to compare Diplomacy with chess, which has voluminous opening theory, and with backgammon, which has much less, but still a substantial amount of opening theory. 

There is an extremely obvious difference between Diplomacy, on the one hand, and chess and backgammon, on the other, and that is the number of players. Two player games are not susceptible to coalitions. Everything that's good for one player, in chess or backgammon, is bad for "every other" player in the same game. And this means that tactics dominate those games completely. Once you find the tactically best move, you know what your best move is. It is a mistake in Diplomacy to assume that the tactically best move is the overall best move. Let me explain by an example. 

Tactically speaking, it is almost always better to have more centers than fewer. That is, comparing two possible future positions, you should almost always prefer the one in which you have more centers, from the tactical point of view. But it is a notorious fact that for some powers, rapid early expansion is to be avoided like the plague. This is a consideration for every power, but for some more than others. Russia is especially vulnerable. If Russia gains four centers in the first two years, she looks so big and threatening that she is likely to attract a coalition against her. Even an eight-center Russia cannot survive a concerted attack by England, Austria, and Turkey (say). In the endgame this matter is even more touchy than in the opening. 

Besides rate of growth, there is commitment to an ally. It is pretty easy to see why the Churchill opening is superior, tactically speaking, to A(Lpl)-Yor. A(Lpl)-Edi preserves the possibility of convoying across NWG, and has no tactical disadvantages compared to A(Lpl)-Yor. But, England might wish to signal an intent to ally with Germany. She might hope to persuade Germany to move against Russia, or at least to remain neutral in a coming war against Russia. A(Lpl)-Yor almost proves that England will be convoying across NTH, likely leaving NWG free to move to the Barents Sea, and ruling out F(NTH)-Den, F(NTH)-Hol, or F(NTH) moving into or supporting a French move into Belgium. If diplomatic considerations favour alliance with Germany, A(Lpl)-Yor may be preferable to the Churchill, despite its tactical inferiority. 

So this is one main reason that there is little in the way of opening theory in Diplomacy. Whatever tactical considerations might be adduced, they can always be swamped by diplomatic ones. And diplomatic considerations are much less susceptible to analysis, depending as they do on psychology, and on extremely complex factors. 

Now for the second reason, a less obvious one. Allan Calhamer, Diplomacy's creator, said that there is no luck in Diplomacy after the initial random assignment of powers. He was mistaken. Diplomacy does have lots of luck in it. Let's see why. 

To fix ideas, take this fairly common scenario. Turkey has made peace in the opening with her neighbours, leaving her free to sail out into the Mediterranean. Getting the jump on Italy, and probably with an assist from an Austrian fleet in Greece, Turkey dislodges the Italian fleet in Ionian, which retreats to TYS. Italy, let's suppose, has no army available to defend Naples. Turkey has two main options. She can order F(ION)-Tun, or F(ION)-Nap. F(ION)-TYS might be good, too, but let's suppose Turkey feels she needs the immediate build if possible.) Italy also has two options: she can defend Tun with F(TYS)-Tun, or she can defend Nap with F(TYS)-Nap. Now, which is the better move for Turkey? Well, there is no better move. Suppose F(ION)-Nap were better. Then Italy would be able to tell that it was. So Italy would order F(TYS)-Nap. That means F(ION)-Nap is not as good as F(ION)-Tun, for obvious reasons! Likewise, F(ION)-Tun could not be the 'better' move. There is no "better move," tout court, here; there are only the better move given that Italy orders F(TYS)-Nap, and the better move given that she orders F(TYS)-Tun. And Turkey doesn't know which Italy will order. 

Technically speaking, Turkey must adopt a "mixed strategy." This is a term from game theory. It means that Turkey should be introducing a randomising factor, say, the toss of a coin. (It could be a weighted coin -- maybe Turkey believes that taking Nap will be more damaging to Italy than would taking Tun, and so will weight her coin toward the Nap side. But her ideal tactical mixed strategy must give some chance to F(ION)-Tun, and some to F(ION)-Nap. Note that Italy is also forced into the same strategy. The luck factor is now obvious. To take an Italian center, Turkey must be lucky. To hold all her centers, Italy must be lucky. Luck does play a role in Diplomacy. 

Technical Note: I have just explained the fact that Diplomacy is a "game of imperfect information." This is another term from game theory. The idea is this: when you submit your orders, you don't know what orders others will submit. From the perspective of the game theoretician, you could pretend that all other players' orders have already been submitted and plotted on the board, but you can't see the board. You have to make your move in partial ignorance of the current position. You have imperfect information. Since chess has sequential moves, it is a game of perfect information. Backgammon should be thought of as an imperfect information game, because each player is ignorant of the dice rolls. Imagine that a thousand rolls of the dice for each player are made in advance, but the players can't see what they are. They are revealed one at a time. Think of these as part of the position, and you see a parallel with Diplomacy. 

The luck factor in Diplomacy makes the combinations of opening strategies explode. There are infinitely many possibilities! France could order A(Mar)-Bur, or A(Mar)-Spa, or A(Mar) S A(Par)-Bur (among the plausible openings). That's three. But she could also toss a coin to decide between A(Mar)-Bur and A(Mar)-Spa. And she could give the coin any one of an infinite number of weightings.  

This explosion of possibilities threatens to make opening theory unmanageable. The explosion can be tamed to some extent, temporarily. Compare backgammon, where first moves for each possible opening dice roll are well-analysed, but there is virtually no opening theory after that. Diplomacy is similar. Both are games of imperfect information, where luck is a factor, and the variety of probabilities leaves analysis in the dust. 

I believe that these two factors -- the combinatorial explosion engendered by the information imperfection and the infection of tactics by considerations of allegiance, coalition, and diplomacy -- are what make the game so interesting; that these are responsible for the remarkable absence of useful opening theory in Diplomacy. 

This article is reprinted from The Diplomatic Pouch No.1

 Stephen Agar Comments 

While I see the thrust of this article I disagree with it on several points of detail. For example, the Northern Opening, Edinburgh Variation (which the Americans call the Churchill Opening) is by no means clearly superior to the Yorkshire Variation, for the simple reason that the former may end up denying England any build at all if Russia opens with a Northern Opening and France orders F(Bre)-ENG. The move to Yorkshire means that England can still put two units on Norway if necessary and cover London. Indeed, unless England has a particular reason to leave his options open for convoying with NWG or NTH, I would say that the Yorkshire Variation is undoubtedly the wiser move. You will note from the Richard Sharp article elsewhere he called the Yorkshire Variation "obviously superior". Which goes to show there may be merit in knowing the basic of opening theory after all! 

As to the out and out guesses, I believe these happen far less often than James makes out. The Tun or Nap dilemma which James refers to probably wouldn't be such a dilemma in real life - other factors would determine which centre Italy would rather lose if he had to and Italy would be sensible to order accordingly (to attempt to out-guess Turkey and lose the more valuable centre would be less than optimum play, unless diplomatic leads suggest that the chance of success is good). There are so many other factors to take into account around the Mediterranean (Where are the French? Will Italy get a build that season anyway? Where is the next Turkish fleet? Is Turkey in a position to convoy into the mainland? etc. etc.) that the optimum strategic play will be apparent to Italy and the consequences of attempting a double-bluff and failing will be obvious. 

I believe that Opening Theory in Diplomacy is useful for the first year, not the first move. Unlike Backgammon, the position of the pieces after the first two moves is crucial and the choice of opening does directly influence how many new centres you can take and how many of your own home centres can be taken by others. Some openings are "safer" than others, but can lead to fewer gains (e.g. a Southern Hedgehog compared to a Balkan Gambit, Budapest Variation).