A Study of Italian Wins on the Judges

Derek McLachlin

Of the countries on the standard Diplomacy board, Italy is generally considered the weakest. Indeed, data posted to the rec.games.diplomacy newsgroup by Douglas Massey indicated that in 3799 standard games played on the judges as of July 10, 2000, Italy fared the worst of all countries in terms of final results (Table 1).

Reasons that have been suggested for Italy’s poor showing include the lack of nearby neutral Supply Centres from which to build early momentum, the remoteness of Tunis from other parts of the board, poor avenues of expansion from its home Centres, tension with Austria over the Venice/Trieste border, and having to cross the major stalemate line to achieve the most efficient victory (Windsor, Hucknall, Ray.)

Numerous strategy articles have been written giving advice on how to play Italy (see the Italy section of the Diplomatic Archive). Naturally, each article reflects the biases of the individual authors, and the authors are often sharply divided on what the best Italian strategy is. For example, while an early attack on France has been advocated as the best strategy (Kauzlarich), other authors (Sandy, Turnbull, Agar) feel Turkey is the best first target, and still others (Bergman, Hopkins, Windsor, Sharp) advocate attacking Austria first. Often proponents of one strategy will dismiss approaches adopted by others as risky, foolish or just plain silly.

But is there such a thing as a strategy that is foolish in all cases? Perhaps, but probably not among those mentioned in these articles. In fact most people would likely agree that in a game as complex as Diplomacy, one should not be tied to any particular strategy (for example, see points raised by Joshua Randall). In the interests of exploring all of the options available to Italy and what makes them work, I thought it would be valuable to examine the examples we have of Italian success, and ask how that success was achieved.

The purpose of this article is to look at Italian victories in standard Diplomacy, and try to determine how those victories came about. In doing so I hope to shed some light on the various pathways Italy might take to victory, so that Italian players might be aware of what options have worked in the past.

This type of study is facilitated by the availability of summaries of games played. For this study I looked solely at games played on the judge system, primarily because game summaries are readily available for these games. I’d like to thank Doug Massey for collecting these summaries and placing them in an easily accessible format.

To collect examples of Italian victories, I searched the summary files from 1988 onward for games meeting the following criteria:

-         standard Diplomacy map and rules (including the allowance of white and partial press)

-         Italy obtained 18 Supply Centres

-         Not a real-time game

-         Full game summary available, including yearly Supply Centre ownership

Thirty-nine games, finishing between January 1996 and May 2000, met these criteria. Games previous to 1996 are not represented because full summaries are not available; games after May 2000 are not represented because summaries beyond that month had not been incorporated into the files at the time I finished my compilation of the data. Greg Greenman called my attention to a 40^th game mastered by him meeting these criteria that finished in July 2000. Greg has posted the entire history of the game, 99ratss, including the adjudication for all phases and all of the partial press on his web site. The complete information on this game proved valuable as discussed below, and I thank Greg for making this game available.

Supply Centre Ownership

One set of data that is easily obtained from the summaries is a simple tally of the number of times each Supply Centre is owned by Italy at the conclusion of the game (Table 2).

Table 2. Supply Centre ownership by Italy at the end of the games. Distance indicates the number of moves it would take a unit to move from an Italian home Supply Centre to the given Centre, discounting convoys.

A quick look at the table shows a group of 16 Centres that were owned in at least 70% of Italian victories. This core group of SCs will probably surprise no-one and includes the Italian, Austrian and Turkish home Centres as well as Tunis, Greece, Serbia, Bulgaria, Marseilles, Spain and Portugal. These SCs lie on the southern half of the board and are all either bordering or one space away from the Mediterranean Sea. This distribution indicates that most successful Italian rulers sought to dominate the Mediterranean region, which lies for the most part on the eastern half of the major southwest to northeast stalemate line. On the whole the SC distribution data suggest that successful Italians have based their strength on naval power.

Interestingly the German home SCs, which are relatively close to Italy, were taken less than half of the time by victorious Italians. Although Italians often crossed the major stalemate line into France and Iberia, they did not enter Germany as often (note that Brest and Paris were taken as often as Munich even though they are twice as far away, and Kiel and Berlin were taken in only 15% of Italian wins). The difference seems to be that four of the five SCs in the French cluster (France plus Iberia) are coastal, while the German SCs are accessible to Italy primarily by land. This trend supports the idea that successful Italians dominate the seas.

As a general rule, the SCs taken in the final year tend to be those more distant from the Italian home SCs. One exception is Munich, which is of course only two spaces away from Venice but, when taken, is taken in the final year over half of the time. Budapest is also two spaces away from Venice and is taken in the final year one-quarter of the time. These high proportions may reflect the land-locked nature of these SCs, which makes them less accessible to a power emphasizing fleet strength.

One other fact worth noting is that each SC on the board is represented at least twice. Therefore Italy appears to have considerable flexibility in terms of which areas it can control on its way to victory.

Patterns of Dominance

Although the data in Table 2 provide a broad impression of Italian success, they do not tell us about patterns in individual games. In an attempt to look at different patterns of Italian wins, I divided the 40 games into three classes: games in which Italy’s gains were predominantly in the east, in the west, or split between the two (Table 3).

Table 3. Classification of Italian victories into predominantly Eastern, predominantly Western, or split. Numbers for each country indicate the mean (and, in parentheses, the median) average Supply Centre counts at the end of the game.

For the purposes of this classification I divided the board into eastern and western sections, discounting Italy and Tunis. The eastern part encompassed the 13 SCs represented by the Austrian, Turkish and Russian home SCs with the exception of Stp, and including the four Balkan neutrals. The western section included France, Germany, England, Stp, and the remaining neutrals (17 SCs in all). If Italy in a particular game took four or fewer SCs from the western part, it was called an “eastern” victory. If Italy took four or fewer SCs from the eastern section, the game was denoted a “western” victory. Games in which Italy took at least four SCs from each section were called “split” victories. I included Stp with the western section because it is more easily taken and defended from the west; placement of Stp in the eastern part of the board would not have affected the classification of any game.

In over half of the victories, Italy took 10 or more SCs in the east, while in only four victories (10%) did 10 or more SCs come from the western side of the board. In view of the numbers in Table 2, these results are not surprising.

The average SC counts for the other powers at the end of the game give us hints about the relative strengths of the other countries throughout the each class of victory. The clearest example is in the four western wins; in these games Austria averaged 13.3 SCs at the end of the game, while all other powers averaged one or below. Therefore the western wins apparently came about because of long-term Italo-Austrian alliances in which Italy either outraced Austria to 18 Centres or stabbed Austria at the very end of the game.

The numbers for the split wins suggest that Italy usually allied with Russia in the east and England or Germany in the west. The eastern games might also have featured co-operation with Russia in the east, and either France or Germany in the west. In the eastern and split games Austria fared relatively poorly, while in none of the games did Turkey finish well. In fact, the SC counts for Turkey at the end of the 40 games were two (three times), one (three times) and zero (36 times).

From Table 3 it appears that alliances with any of the other powers, with the exception of Turkey, have led to Italian victories. If we assume that a correlation exists between SC count at game end and friendly relations with the victor, then the most common partners for successful Italians have been, in order, Russia, Germany, France, England and Austria.

Paths to Victory

In Tables 2 and 3 we have looked at which Centres make up Italian wins, but we have seen little about how Italy goes about attaining those Centres. To see the paths Italy takes to victory, I separated the Diplomacy board into seven sectors corresponding to the seven powers. For each game, I then charted the order in which Italy permanently entered the various sectors. Italy was considered to have permanently entered a sector when it captured a SC in that sector, and maintained at least one SC in the sector for the remainder of the game. When Italy permanently entered two sectors in the same year, the sector closest to the bulk of Italy’s other gains was given sequential priority. The Italian sector itself was ignored since all Italian winners maintained at least one SC in the Italian sector for the entire game. The SCs in each country’s sector are given in Table 4.

Table 4. Division of the board into sectors. Rumania and Belgium were not included in any sector because of their positions between powers, making it less clear to which sector they should belong. Greece was not included in the Austrian sector even though Austria usually takes Greece in the game’s early stages because, when allied to Italy, Austrian players frequently relinquish Greece to Italy voluntarily. In such cases Italian ownership of Greece would be incorrectly interpreted as a hostile incursion into the Austrian sector. Nwy and Swe were included in the English and Russian sectors respectively, because in 1901 these SCs are traditionally considered “natural” neutrals for those countries to take.

Such analysis reveals the order in which Italy made permanent inroads into the various regions of the board (Table 5).

Table 5. Sectors permanently entered by Italy, ordered sequentially. Numbers in parenthesis indicate the number of occurrences of the given pattern. For example, reading along the top, Austria was entered first 21 times; of those 21 times, Turkey was entered second 14 times; and of those 14 times, France was entered third eight times, giving a total of eight occurrences of the sequence ATF.

To make classification of games easier, only the first three sectors permanently entered are shown. In all cases the first three sectors of expansion represent ³50% of the SCs taken outside Italy’s own sector and thus these sectors are most useful for analysis of the paths to victory. In most games, four sectors were actually entered, and sometimes Italy made it into five (although never all six).

One should avoid assuming from this analysis that Italy’s path to victory was always straight, proceeding from one sector to another in a direct ascension to dominance. In many cases, the games did proceed in such a way, but it was not uncommon for Italy to suffer one or several serious setbacks before finding a way to win. The stipulation that Italy had to hold a SC a sector for the rest of the game for that sector to be counted was an attempt to smooth over the bumps along the path to the win.

From Table 5 it is clear that the first successful attack (as judged by permanency of retaining the taken SCs) in about half of Italian victories was against Austria. First attacks on Turkey and France account for about one-quarter of Italian wins. According to the system described, Germany was not attacked first in any of the Italian wins. (However, the game listed as AGT, hundido.USIN, might be more appropriately listed as GTF. Italy took Trieste early in this game and kept it throughout, but never took another Austrian SC. Austria was strong for the entire game, and analysis of the SC summary suggests a long-term Italo-Austrian alliance. In terms of final SC ownership, hundido resembles much more closely the other three games classified in Table 3 as “western wins,” which all appear as FGE in this table.)

The most striking thing about Table 5, in my opinion, is the diversity of avenues taken to victory. True, Austria and Turkey were the first two targets in almost half of the games (18), but there are many other options represented. Most paths to victory (30) involve attacks on countries in both the eastern and western triangles, but attacks solely against the east (7) or the west (3) were also successful on occasion. Because of the diversity displayed in Table 5, it seems almost certain that not all possible paths leading to Italian victory are represented among these 40 games.

Patterns Among the Paths

The diversity of strategies for Italian wins brings us to a crucial question. How is the Italian player to know which strategy will work best in a particular game, given the characteristics of the individual players in that game? To answer this important question I examined the SC ownership tables for all 40 games in an attempt to chart the general alliance structures that existed in each game. I then sought patterns in the alliance structures that might correlate to the paths to victory shown in Table 5.

A point to consider is whether one can accurately determine the alliance structures that existed in a game from the SC ownership table. Predicting the alliance structures in such a way surely misses to some extent the rapid allegiance shifts that occur in virtually every game. But can the essence of the major conflicts be outlined in a very “broad brush” picture of the game?

The availability of the entire move and press history for 99ratss [www.spencersoft.com/diplomacy] allowed me to test my ability to discern alliance structures from the SC table. I examined the summary for 99ratss without looking at any other information from the game. After I had made predictions about the alliances within the game, I looked at the move and press history to see how accurate my analysis had been.

Here is my analysis of 99ratss based solely on the SC ownership table:

The left and right columns indicate action in the west and east, respectively, according to when in the game the alliances were in evidence. Parentheses indicate alliances or conflicts that were either not clear or progress slowly. In any conflict the power(s) to the left of the ‘vs.’ seemed to have the upper hand. The word ‘stable’ indicates that no dominant alliance arose in the west at the beginning of the game that resulted in the early elimination of one power. The blank space indicates no change in the alliance pattern from the mid- to late game. As you can see I only attempted to record the very basic conflicts that occurred; trying to capture all the nuances of alliance shifts and power fluctuations that occur during a game results in an unwieldy amount of detail.

Upon examination of the complete history of the game, I found that this basic analysis captured the essence of the game’s progress. A more detailed history of the game, based on actual orders and press, is given below.

The general trends were more or less correctly identified. The west was slow to develop but settled into an EF vs. GR conflict that was relatively balanced. (As is apparent from the press, misunderstandings and mistrust between E, F, G and R resulted in little real progress for any of these powers in the north and west during the entire game). Italy and France were at odds in the Mediterranean in the mid-game. In the east, Italy pounced on Austria while, independently at first, T and R fought. The east settled into an I vs. R conflict, and in the end Italy won by gaining ground primarily in the east.

Some events were mis-assigned or simply missed altogether. In the west, England’s stab of France in 1905 was construed as active IE co-operation. The SC summary does not show that France was the initial aggressor in the Mediterranean. France’s eventual passivity toward Italy was undetected. Also, in the latter stages of the game Germany was actively working against Russia. In the east, I missed the short-lived IR co-operation against T. I also mistakenly assumed that the two-Centre Turkey remained anti-Russian in the mid-game when in fact RT joined together against Italy. It is also generally difficult to tell from the SC table which powers band together to try to stop the leader in a game’s final years, and this game was no exception. As it turned out the original Russian made pro-Italian orders in 1909 but in 1910 a new Russian joined EG in trying to stop Italy. Turkey, disgusted by the lack of co-operation between England, Germany in Russia in 1909, decided to attack Russia in 1910 and leave his SCs undefended to help Italy win. On the whole, these mistakes or oversights in the analysis seem relatively minor.

Therefore, I feel that the basic alliances and conflicts of 99ratss were depicted with reasonable accuracy in the initial analysis using the SC ownership table. I constructed similar alliance charts for the other 39 games, organized the games into the routes of expansion shown in Table 5, and looked for patterns (Table 6). In this table I included all routes of expansion that occurred at least twice. For convenience I lumped games according to their first two areas of expansion, except for ATR and ATF, for which there were enough cases to justify a separate category for each.

Table 6. Alliance and conflict trends in games divided according to Italy’s expansion route. Trends do not necessarily occur in every game, but in most games of the group.

The table shows diverse trends across routes of expansion, but a few things stand out as common to many games. First, in many games in which Italy expands eastward first, the western triangle was slow to sort itself out. Such a situation would allow Italy to concentrate forces in the east without having to worry about a victorious western power trying to enter the Mediterranean in force.

Second, Russia frequently formed lasting alliances with Italy. A clear trend in games in which Italy has an early eastern focus is for Russia and Italy to attack one of A or T and then take out the other one once the first one is eliminated. Perhaps as a consequence of success in the east, Russia also was often involved in the western half of the board.

Third, as predicted in the discussion of Table 3, lasting IT alliances have not formed in Italian wins. Note that every other country is represented in the last column of Table 6. The dismal results for Turkey in Italian wins indicate that Turkish failure might be key for Italian success, giving some credence to Richard Sharp’s statement, “(Turkey’s) got to go.” This finding may help explain the popularity and attractiveness of the Lepanto opening for Italy. However, in most cases Turkey was not taken care of immediately and directly by Italy. More often in Italian wins, either Turkey was the secondary target for Italy, or Austria fought Turkey with Russian help or minor assistance from Italy.

Other trends in the data are not as clear. I would advise Italian players to use Table 6 in game-specific settings by considering the likely route(s) of expansion available in a particular game, observing what alliance and conflict conditions have led to Italian victories in the past, and then trying to bring about such conditions. But be wary: each game has individual and possibly unique characteristics, so players should evaluate strategies carefully. Always be on the lookout for a new path to victory; just because it has not been done before doesn’t mean that it cannot be done!

A Question of Friendship

In general, which countries generally make the best allies for Italy? Many authors (Windsor, Sandy, Hucknall, Hopkins, Agar, Drews, Morris) have pegged Russia as a preferred ally. The data presented in this study are clearly in accordance with these authors’ opinions. Successful Italians frequently co-operated with Russia for significant portions of the game (Tables 3 and 6). Such an alliance makes sense, since Russia and Italy have two common neighbours in the east but no common border. This arrangement makes reasons for co-operation easy to find and stabs more difficult to carry out.

Apart from Russia, Austria has also been mentioned as a good ally for Italy (Agar, Turnbull). This study agrees to some extent, although Italian success based on a strong Austrian alliance has been relatively rare (4 of these 40 games; Table 3). It is probably safe to say that the AI alliance can be powerful, but is difficult to establish in the long-term because of the adjacency of Venice and Trieste.

Paul Windsor has suggested that England and Italy make good allies based on his calculation of “fear factors.” The fear factor shared by two countries takes into account the distances between their home SCs, and whether or not the home SCs are required for the swiftest possible victory by those powers. While it is true that the homelands of England and Italy are distant, and that wins by Italy have not typically involved conquest of the English homeland, my interpretation of the data in this study is that Germany is the more common and probably better ally for Italy. Scott Kauzlarich also favours Germany over England, “but not by much.”

It is true that Munich was taken more often in these Italian wins (17 times) than any English home SC (the highest is Lpl with 11 times; see Table 2). This fact probably does reflect the proximity of Munich to the Italian homeland. In 10 of 17 cases, however, Munich was taken as one of the last SCs in the push for 18, after Italy had shed its major alliances. In these cases the basic fear factor indicated by the layout of the board is irrelevant; all players should experience maximal fear from an Italian who is about to win!

I believe that Paul has overestimated the threat Germany and Italy pose to each other by emphasizing that most of each country’s home SCs are part of the other’s swiftest route to victory. While this may be strictly true based on distance, Table 2 shows that Kie and Ber have been taken in only 15% of Italian wins, and even Munich was taken less than half of the time. Therefore Italy’s “swiftest” route to victory was not usually the route actually taken by victorious Italians. Furthermore, Edi and Lon were taken as often as Kie and Ber, while Lpl was taken in more than one quarter of Italian wins, suggesting that, despite its distance from the boot, England is as viable a target for Italy as Germany.

Considering the review of trends in Italian wins, there were more instances where Italy co-operated with Germany than with England (Table 6). I believe Germany was preferred because Italy most often pursued a primarily naval strategy (Table 2), and since England is also mainly a naval power, the two countries can easily run into conflict about ownership of the Mid-Atlantic Ocean and the SCs in that region. On the other hand, Germany often pursues a land strategy, and if fleets are built they are used in the far north away from Italy’s usual areas of expansion. In addition, Italy and Germany may co-operate effectively against France and in Austria, while other than France, England and Italy arguably have no region where they may co-operate as equal partners.

Also consider that while Germany is close to Italy, an attack of one by the other must be accomplished by land and across the major stalemate line. Switzerland greatly impedes such attacks, and unless the aggressor can devote extensive resources to the campaign, the defender is likely to be able to repel or slow down advances with relatively little difficulty. Aid from other nations in an attack between Germany and Italy is likely to result in the third party gaining most of the spoils. In the end it’s just not worth it for either country. Therefore I would argue that, in general, Italy and England make a less stable alliance partnership than Italy and Germany.

Alliance with France has been shown to lead to Italian success, although this partnership might be better termed long-term neutrality. Italy and France can co-exist, but possibilities for active co-operation are limited.

Summary

This article has tried to explore past Italian wins on the judges to see how they were achieved. Trends observed in the games studied may be summarized as follows:

1) In most cases Italy pursued a naval Mediterranean strategy.

2) Russia was the most frequent alliance partner in the east.

3) Germany or England (in that order) were preferred allies in the west.

4) Turkey fared strikingly badly.

5) Russian involvement in the north helped distract the western powers from Italy.

6) In Italian wins where Italy headed east initially, the west was frequently slow to develop.

It is my hope that by pointing out the variety of options open to Italy, and recording how Italy has successfully pursued these options, the country may be played more effectively in the future to the benefit of the hobby in general. As a final word I again urge Italian players to keep an open mind when weighing different strategies; there are many paths to Italian victory, and the best one for your game might be over uncharted territory.

Derek McLachlin (derek@chait-sgi.rockefeller.edu)

If you wish to e-mail feedback on this article to the author, click on the letter above. If that does not work, feel free to use the "Dear DP..." mail interface.