Diplomacy in MTW:VI

[gunslinger]

I have wondered how the AI decided which one of these to use. At times it seemed random and may well be to an extent. It is possible that it makes a check on how long it has been since any battle occurred? With alliance rejections it probably has something to do with your empires world ranking.

I have definitely noticed that if I repeatedly pester a faction leader for an alliance (usually Da Pope) his rejections of my offer will go steadily from benign to threatening. He may start off with "These matters must not be taken too lightly," but by the third or fourth try my emmissary will feel lucky to escape with his life.

[Ironside]

Only thing I've noticed with alliances is that attacking an ally will cost 1 influence and that the factions are extremly unlikely to ally with you if you're at war with one of thier allies. Making peace or wiping out that faction you're war with makes you able to ally yourself with that faction's former allies.

[EatYerGreens]

I don't particularly envy anyone who ever had to write AI logic (let alone translate that into program code!) and it feels hypocritical to have a go at the way it behaves in the TW series... but I won't let that stop me from making one or two further observations.

Quite often, I'll see Novgorod, beaten back to two or three territories and offer it an alliance. At times when it has no shared border with me, no agents in my territories, no ships to survey the state of world affairs, it still turns my offer down! That's just an example, the same thing can happen with factions closer-to but there's no indication that the strength of your faction is taken into account.

The subject of border forts/agents came up in the Pocket Mod thread recently and it was mooted that the AI is able to see the state of the whole map, without the need for agents, so I could be wrong about their reactions being 'blind'. Debateable, perhaps.

Anyway, my best guess is that alliance decisions are boiled down to simple mathematics but it has no 'memory' and basis its decision on the 'now'. An ability to plan for the future would require a whole extra layer of coding and, in some ways, I'm still looking for evidence of the existence of any such thing. Perhaps that is one of the major reasons why the AI is at such a disadvantage against the player?


The influence thing is another conundrum. Influence goes up and up, the more territory you take and, if the text descriptions are to be believed, this increases the likelihodd of offers being taken seriously. The paradox is where you become so powerful that everyone begins to attack you whereas, in the real world, you'd expect a whole lot of appeasement to be going on.

Equally puzzling is the way the Byzantines often get used as a punch-bag by their various neighbours, squeezed down to three provinces, or less, yet the emperor still manages to maintain 8-9 influence. What's going on there? Granted, they may be winning provinces on a regular basis but they're losing others just as fast, which should be knocking influence off.

[caravel]

The subject of border forts/agents came up in the Pocket Mod thread recently and it was mooted that the AI is able to see the state of the whole map, without the need for agents, so I could be wrong about their reactions being 'blind'. Debateable, perhaps.

I believe the Ai can see the whole map all of the time, based on some observations I've made. Mainly observations of Emissaries and Assassins. The AI Emissaries I have noticed proposing alliances with other faction leaders on many occasions, where they had no prior knowledge of either where he was located, had no contact with any of his emissaries and had no other agents in his provinces (ian mode is useful for this). The same goes for Assassins. Byzantine Assassins targeting a general in Scotland, for no apparent reason and with no prior intelligence of his position.

Anyway, my best guess is that alliance decisions are boiled down to simple mathematics but it has no 'memory' and basis its decision on the 'now'. An ability to plan for the future would require a whole extra layer of coding and, in some ways, I'm still looking for evidence of the existence of any such thing. Perhaps that is one of the major reasons why the AI is at such a disadvantage against the player?

I also think that the AI reacts on a turn for turn basis. It will make a calculation based solely on the now rather than the then.

The influence thing is another conundrum. Influence goes up and up, the more territory you take and, if the text descriptions are to be believed, this increases the likelihodd of offers being taken seriously. The paradox is where you become so powerful that everyone begins to attack you whereas, in the real world, you'd expect a whole lot of appeasement to be going on.

I've never been sure that influence has an effect on his either. Though the bloat effect is a very strange phenomenon. The major effect is that the loyalty in your provinces take a drop, as though your king is isolated. Then there is the matter of being despised, having your offers of alliances/ceasefires refused and having your princesses and marriage offers rejected. Then there is the small matter of the AI factions suddenly making war upon you, in an attempt to bring you down. This points to many different factors. A hidden drop in influence would also affect your generals, but it doesn't, so i wouldn't say that had anything to do with it. I think it's a piece of code that is triggered by conquering a certain number of provinces - just over 60% of the map. It is probably also related to how MTW AI handles the diplomacy: Small factions are liked, bigger factions are not. Small factions can always marry off their princesses, marry their heirs to foreign princesses and usually get an abundance of allies. As you get bigger you get less allies. So it seems that this part of the MTW diplomacy simply works in the reverse of what you would expect. I'm not sure influence has any effect at all, and is simply a stat that controls the loyalty of your own generals and the quality of your heirs.

Equally puzzling is the way the Byzantines often get used as a punch-bag by their various neighbours, squeezed down to three provinces, or less, yet the emperor still manages to maintain 8-9 influence. What's going on there? Granted, they may be winning provinces on a regular basis but they're losing others just as fast, which should be knocking influence off.

I think that they start off with such high starting influence and superb royalty that it takes a lot to knock them down. I have seen the Byzantine Emperor with 3 influence and split into civil war on occasions after they had taken some heavy losses and lost a lot of battles, but they can take a severe hammering before that occurs. Just after the command stars loop bug kicks in they tend to be much more vulnerable. Also I've noticed reappearing Byzantines to be also much more prone to low influence and civil wars. Thids makes me think that it is mainly down to the starting stats of their royals.