Winning at webDiplomacy

I’ve spent some time recently revisiting a teenage enthusiasm for the game of Diplomacy, in which (ideally) seven players with differently matched forces try to dominate the map of Europe circa the start of the twentieth century. The webDiplomacy site allows you to test your mettle not only against human players, but also against AIs. This can go quite fast, at a move every minute or so.

I am horrified and fascinated that the AIs continually defeat me, a mere human player. The couple of times I’ve tried “Gunboat Diplomacy” (where there is also no communication) with real humans, I found the going much easier, as opponents with brains made out of meat are more likely to make mistakes. Out of a couple of dozen battles with AIs, I have won only two; one where as Turkey I managed to break out of my corner before the rest of the countries had properly got started (which I think is the least difficult way to win against the software), and another as Italy where I combined good luck with a couple of good tactical shouts and a coherent overall strategy.

I realise that reading about other people’s solo games is about as interesting as hearing about other people’s dreams, but heck, this is my blog and so I’m going to tell you about how I beat the robots as Italy, which is normally seen as one of the weaker of the seven Great Powers – the others being Austria (sic), England (sic), France, Germany, Russia and Turkey. In summary, I did it by waging aggressive war successively on Austria and then Turkey, and benefited from a prolonged stalemate between England, France and Germany and a curiously passive Russia.

Taking it turn by turn::

1901

Spring 1901: As Italy you know you’re either going to ally with Austria or (more likely) attack them. When Austria is played by a robot, no alliance is possible. Then the choice is, go for the subtle attack on Tyrolia? Or the straight stab into Trieste. What the heck, I thought. Go for broke and the straight stab on Trieste. And it worked!
But a crucial factor was the failure of the Russian army in Warsaw to move. If it had bounced with the Austrian army heading for Galicia, the whole game would have been different from the start.
Autumn 1901: Again a risky choice, hoping that Austria would not try and retake Trieste and that I would guess correctly as to which of Vienna and Budapest would be defended from Galicia. And again, it worked. Once again, the Russian army in Warsaw was curiously passive.
Three builds as Italy in 1901! That never happens. At this stage I seriously began to consider the possibility that I might be able to win the game. That French fleet in Marseille was a bit worrying, though with Brest blocked I guess there was no alternative.

1902

Spring 1902: I was fully committed to an eastern strategy now, and the first thing was to complete the crushing of Austria with a hard move on Budapest and retaking the Ionian Sea. My army in Rome moved to Apulia for future action.
Autumn 1902: This time my forces were strong enough to take Budapest, and with the fleet in the Ionian I convoyed the new army from Apulia into Albania, the first of a couple of crucial convoys in the game. And I guessed correctly that the French fleet was headed west rather than east. Up north, Sweden, the last neutral supply centre, fell to Germany.
Autumn 1902 retreats: the Austrian army in Budapest was forced to disband. (I think the arrows in Brest are to show that you could not retreat there.)
Winter 1902 builds: Only me and Germany. And the northwestern half of the board all looked rather stuck.

1903

Spring 1903: Continuing my eastern strategy; I already controlled all three Austrian home centres, so my new army moved from Venice into Trieste in preparation to attack Serbia; but also I shuffled my fleets eastward to be ready for Turkey, hoping (correctly) that the Turkish fleet in the Aegean would be used to attack Greece. A bizarre French set of orders in Brest and the Mid Atlantic Ocean meant that not a single one of France’s units actually moved.
Autumn 1903: And now I had three units to concentrate force on Serbia, while also moving naval power into the Aegean. Up north, Germany had a breakthrough by capturing both Belgium and the North Sea from England.
Autumn 1903 (retreats): I crushed the Austrian army in Serbia, while Germany also eliminated the English fleet in Belgium – the only supply centres to change hands this year.
Winter 1903: again, only Germany and I had builds. This is the point where, in a human game, Russia and Turkey would have agreed that I was the common threat and combined to resist me. But the AIs didn’t realise that I was now quite far ahead.

1904

Spring 1904: I still had half an eye on France and shifted the newly built army from Venice to the border just to monitor the situation. But the main business was the eradication of the last Austrian unit, sheltering in Greece, combined with an opportunistic strike on Rumania [sic] – which to my surprise succeeded; the AI should have prevented it with support from Sevastopol. On the other side of the board, Germany convoyed an army to Yorkshire, with the German/Russian border rather quiet.
Spring 1904 retreats: my forces eradicate not one but two opposing fleets, the last Austrian in Greece and the Russian in Romania.
Autumn 1904: A gamble that Russia would not try and retake Rumania or stab at Vienna paid off, allowing me to consolidate forces on their border; but my attacks on Bulgaria and Smyrna both failed. Even so, I think this was the point of no return, with my six units in the southern theatre facing four Turks and two Russians who were unable to coordinate.
Winter 1904 builds: With stalemate in the north and resistant crumbling in the south, I was the only country with new units – two of them, in return for Greece and Rumania. An army in Venice of course just to keep an eye on France; but rather than build a fleet in Naples, I opted for another army, with an eye on swiftly overrunning Turkey.

1905

Spring 1905: Again, the anticipated Russian counter-attack on Rumania didn’t happen; I made a half-hearted push on Bulgaria, but the main action was convoying an army to Syria, plus an opportunistic stab at Constantinople – which unexpectedly worked. At the other end of the board, Germany captured Edinburgh, and France convoyed an army to Wales, England being mashed up between the two.
Spring 1905 retreats: the English army retreated from Edinburgh to Clyde. (I use England here as that is the name of the country in the game, even though most of the remaining action is geographically in Scotland.)
Autumn 1905: Thanks to the capture of Constantinople, my attack on Turkey was running ahead of schedule. However, I felt that the best way to consolidate would be a tactical move from Constantinople to Smyrna, combined with the attack on Bulgaria which was now unstoppable – in fact the Turkish army in Bulgaria moved to Constantinople, with support from the Black Sea, when it saw me coming. Meanwhile France captured both London and Liverpool.
Autumn 1905 retreats: massive annihilations, with the German army in Edinburgh, displaced by England, and the English fleet in London, displaced by France, both attempting to retreat to Yorkshire and both therefore destroyed; meanwhile further south, Germany kicked France out of Burgundy. And my attack on Smyrna eliminated the Turkish fleet there.
Winter 1905 builds: I got another two, for Bulgaria and Smyrna! Germany got a build to replace the lost army which had briefly occupied Edinburgh, and France had space for only one of the two builds that London and Liverpool would have given it. England was now down to two fleets clinging to the Scottish coast, and an undefended Norway.

1906

Spring 1906: Endgame now (at least so I thought) for my Turkish campaign, as I had enough forces to capture Constantinople again, for keeps this time; and also start shifting forces around to prepare for the attack on Russia by attacking Galicia. And in an eerie reiteration of real world history, the German incursion on France had got bogged down.
Spring 1906 retreats: The Turkish army in Constantinople was crushed, and the Russian army in Galicia went back to Warsaw.
Autumn 1906: the main action for me was to finish Turkey off by capturing Ankara while securing Constantinople, with support from Smyrna and Armenia; and I now had enough wiggle room to try opportunistic moves to Ukraine and to the Gulf of Lyons. But I misjudged the Constantinople attack, and should have done it the other way round, with Smyrna supporting a move from the Aegean rather than vice versa.
Autumn 1906 retreats: My misjudgement allowed the last Turkish army to escape from Ankara to Smyrna.
Winter 1906 builds: despite losing Smyrna, I had gained Constantinople and Ankara, so I was still up one, and could now start looking around for how I might get to 18 centres. A fleet in Naples seemed a better option than yet another army in Venice.

1907

Spring 1907: With four of my units surrounding Smyrna, time was up for Turkey; but I also had enough resources to make moves on the inadequately defended Sevastopol and Marseille. Marseille was particularly important because it is on the other side of the stalemate line that runs from St Petersburg to Tunis. Over on the other side, Germany captured Norway while France compensated for Marseille with Belgium.
Spring 1907 retreats: The Russian army in Sevastopol was eliminated and the French and Germans retreated from Marseilles and Belgium, with the last Turk off to the Syrian desert.
Autumn 1907: With the imminent elimination of Turkey, my troops were now ready to break north to Russia and west to Spain. Again, if humans had been playing Russia, Germany and France, they could have co-ordinated against me. In fact I see that the German army in Silesia did support the Austrian army in Warsaw, which I could otherwise have captured.
Winter 1907 builds: I couldn’t quite believe it, but with my three spring gains of Marseille, Sevastopol and Smyrna, I got three builds and a total of 16 centres, only two short of victory; and Spain, Portugal and Moscow were all hanging open.

1908

At this stage there is no point in being nice to people (especially if they are robots). I made sure of Spain and tried for the Mid Atlantic Ocean; the AI playing France was still fighting with England over Edinburgh and with Germany over Burgundy, allowing me to slip in on its western flank. To the East, I had enough firepower to capture a vacant Moscow. Down along the bottom, I slipped an army into Tunis for later use.
This was the end. The army which had started the year in Smyrna finished it in Spain, though France could have prevented it; my fleet strolled along the coast into Portugal; I grabbed Warsaw off the disarrayed Russians; and what the heck, I had a couple of spare armies and had an unsuccessful go at Munich too. Up north, France recaptured Burgundy and finally took Edinburgh, for what that was worth.
Three retreating units, but it made no difference.
So I ended the game with a massive 20 centres, which I don’t think I have ever managed before. None of the others had had more than eight at any stage.

So, I beat the AIs at their own game, which was rather pleasing; though I was helped by the French robot’s determination to keep attacking England rather than defend against me, and by the curious inaction of the Russians. Let’s have a nice little gloat chart:

I’ll finish with one more thing that happened after I had first written this post: I learned from Ansible that John Boardman, who started the entire postal Diplomacy hobby in the early 1960s, died a few days ago at the age of 92. I never had any direct dealings with him, but he was the source of much enjoyment for a lot of people.