As we all know, there really is no better way to improve at Diplomacy than to play. There are not many independent lines of study to pursue, as there are in Chess, that will noticeably improve your play. But there are a few.
I myself found that my understanding of the topology of the board, and consequently of some large-scale structural features of games, increased markedly when I tried some of Danny Loeb's posers. What space borders the most other spaces? The answer helps you understand a very important "bottleneck." How many spaces border no coastal land space? The answer drives home the importance of sea power. Which pair(s) of spaces are connected by the largest number of distinct sequences of different spaces? Draw the lines to see how a locally victorious Northern power can invade the South (or vice versa).
Documentation of various openings and their purposes is helpful, and those given in the Gamer's Guide are almost indispensable, but treatment of issues that arise in the middle and endgames are fewer and far between.
There is one subject of study, initiated in the Gamer's Guide, that really does help. It is the study of stalemate lines. (I won't give the definition of stalemate lines or list the ones given in the Guide, but even the reader unfamiliar with the concept will be able to piece the idea together from what I say below.) I found it the most useful section of the Guide, and I recommend it (I believe that it was reprinted in an issue of EP). But with one very notable exception, the lines published in the Guide are not very "practical." By "practical" I mean roughly, "exerting influence on actual games." I should quickly point out that in order to exert an influence on a game, a stalemate line does not actually have to be formed, or even nearly formed. It is enough that at least one player recognizes that the line could be formed, even in the distant future, and takes some action based on that recognition. For example, Germany may be tempted to try a not-very-tactically-sound invasion of Austria fairly early in the game on the grounds that if a certain stalemate line is not crossed early, a budding Austro-Turkish alliance may establish it later, and the beautifully laid Franco-German strategy will fizzle into a 4-way draw.
The exceptional line is Line 2 in the Guide. It exploits the Gibraltar bottleneck with F Mid H, F Por S Mid, F NAt S Mid; the Denmark bottleneck with U Den H; and the StP bottleneck with F StP H, F Nwy S StP. I have seen this position or the threat of it exert influence on several games. Why? I suspect it's because the position, or chunks of it, arise naturally from England's course of expansion (or that of an alliance that includes England). The stalemating powers do not have to plot their whole game around establishing the line; rather, they can try for a win and fall back into the line if things turn sour. Another reason may be that it is very simple to visualize. Gibraltar, Denmark, and StP just look like bottlenecks. So it is easy to keep the line in the back of one's mind.
The other lines are certainly theoretically possible, and there's no really obvious reason why they shouldn't exert practical influence (except for Line 3, which is quite freakish). But in fact I've never seen them come anywhere near fruition, and I doubt that they've exerted any influence at all in the games I've played.
So I'd like to add a new line to the documented ones, or rather a few related lines. They are all not-too-distant relatives of Gamer's Guide Line 1, as it happens, but they have a very important property not possessed by any listed lines: They hold 17 centers. By the Guide's definition, a stalemate line can't hold more than 17 centers. But if a line holds fewer than 17, then a determined power on the other side could win even if the line is set up. The Guide notices this, of course, and the rationale (and a sensible one) is that stalemate lines can be used to stop a coalition of powers from wiping out defenders on the other side. That's true. But at least some of the time what you want is a sure-fire way to stop a single power from using its momentum to win alone. If you could set up a 17 center line, you'd be free to try your luck at knocking off the front-runner at no risk.
Just such a situation arose in Horse, a game run by Ken Lowe's Adjudicator. The key powers were Russia (Dave Cebula), Austria (Nick Waterson), and Turkey (me). An R/A/T coalition rolled over Italy and Germany, then expanded out into the West via France in the South and England in the North. In the course of the expansion, Russia grew a bit faster than the other two allies. When we proposed a three way draw, Russia vetoed. Dave was going to try for the win.
France, the only other power left, agreed to puppet for Russia, and a quick center-count showed that if France would be able to turn over all his centers to the Bear, Russia would have 16.
Here was the practical stalemate line that Nick and I worked out.
First, notice that we could be fairly secure in the following centers: Austria (3), Balkans (4), Turkey (3), Italy (3), Tunis (1). To prevent a Russian win, we needed 3 more. The natural candidates were Munich (which Austria held at the time), Marseilles and Iberia (held by me, and in any case likely prey for Turkish fleets), and Sevastopol (if Turkey could establish a stronger Black Sea presence than Russia could). Now stalemate cognoscenti will quickly recognize that Munich is out of the question. It simply can't be held from below, because it has too many Northern neighbors. So we needed three out of the other four.
It turns out that there is a good line that holds all four, and one that holds Sev but not Por, as well as one that holds Por but not Sev. In the actual game, we reached a position from which we could guarantee a draw holding Por, and we might get Sev too, but Russia could hold on to it if he made just the right moves. I'll list both categories of lines. The other elements of variation turn on whether the T/A defenders can hold Gal, and/or Mid and Gal (in the line which holds Por).
I'll give the Eastern and Western halves of the lines separately.
Mar must be held, and Spain. Either Wes or NAf, or, in 1 and 3, both. Here are the variations. (As should be obvious, "U" means a unit where it doesn't matter whether it's a fleet or an army.)
F NAf H
F Wes S Spa
U Spa H
F Lyo S Spa
U Mar H
U Pie S Mar
F Mid H
F Por S Mid
F Wes S Mid
U Gas H
A Mar S Gas
F Spa(nc) S Gas (could be an army, too)
F NAf S Mid
F NAf H
F Wes H
F Por S Spa
U Spa H
U Mar H
U Pie S Mar (or a fleet in Lyo)
Tyrolia must be held, and Bohemia, and the former always supports the latter. Budapest and Vienna are occupied, but their action depends on whether Galicia is also occupied. [Completely irrelevant note: has anyone else noticed that on the Conference Maps the Ukr area is labeled "Ukraina"??? Do I have a rare edition, or what?] Rumania is occupied. The rest varies. (I'll drop the specification of unit type from here on.)
Trl S Boh
Bud S Gal
Vie S Gal
Rum S Sev
Trl S Boh
Vie S Gal
Bud S Gal
F Bla S Rum
Trl S Boh
Vie S Boh
Bud S Rum
Arm (or Bla) S Sev
Trl S Boh
Vie S Boh
Bud S Rum
Bla S Rum (not Bul or Ser!)
Here are some remarks about the variations.
I didn't list separate variations for holding Gas without Mid or vice versa, because once one goes the other can't be held without luck.
Of course, if you want to hold at exactly 17 centers, then you have to use the East and West variations in the proper combos, holding either Sev or Por but not both.
It is interesting, I think, that there is no interdependence between the two halves of the board. Changing variations in the West make no difference to the orders in the East, and vice versa. The reason for this is pretty obvious, see the next comment.
The importance of Switzerland is highlighted in these variations. For one thing, it accounts for the independence of the flanks. No unit that can influence the Western flank can also influence the Eastern flank. Notice also that Switzerland's presence is exploited in each flank. Marseilles needs one or no supports because of its impassable neighbor. Tyrolia not only needs no support, but it can offer useful support to Bohemia because the only hostile neighbor is also one of Bohemia's (so that Munich can cut Trl support or attack Boh, but not both). Units sitting right on the stalemate line often cannot order useful support, but they can when they exploit bottlenecks. (The Rum support of Sev in variation East 1A is another example, but there the bottleneck is created in a less obvious way.)
The line requires between 13 and 15 units. The Western lines take 6, or in West 2, 7 units. The Eastern line takes 7, or in East 1B, 8 units. So you'll have between 2 and 5 units that aren't doing duty in your stalemate line (but 5 only if you hold both Por and Sev).
Why bother listing the positions that hold Galicia? After all, these reduce easily and without risk to positions that don't hold it, so that when you've achieved the one you automatically achieve the other. Similarly for the position that holds Mid and Gas; it can fall back into a position that still holds Portugal (though if Gas is lost, you'll need something to back up Mar).
The reason is that having reached your stalemate position, you don't necessarily have to accept a draw. You may want to fight on for a 2- way draw (or even better...). After all, your side holds half the board, why shouldn't you have a shot at wiping out the opposition? (In Horse it was never really an option, because the two allies knew that they probably couldn't pull off the precarious 2-way draw. I guess I should mention that the actual game was a three way draw: T/A/R). You can set up the advanced position, use it as plateau, and there's even the chance that the opposition won't notice that you have a fallback line and so will expect you to protect the advanced one at all costs. That could give you an edge.
I hope we'll see some more articles along these lines. A study of bottlenecks? How to invade England? I think some thoughts on the interface between theory and practice are long overdue.