On AI and Diplomacy

[Silver Rusher]

I agree with the first post, and the example with the quiz-show type question was (and i gotta hand it to you, Lofman) a very good way of expressing your point.

It isn't just RTW we are talking about. It is also every other total war game. The problem exists in MTW as well, you know. Although I must admit it is considerably worse in RTW. A more complex diplomacy system needs a more complex diplomatic AI.

[Odin]

While I agree that diplomacy is the major short coming of the total war series it seems to me that the AI acts within a set parameter of behavior that ignores diplomacy altogether.

Time and time again in STW and MTW I have been backstabbed by the AI while in alliance. Upon reflection, in most cases the AI had a numerically superior advantage to me when that backstab took place (by way of florins, provinces, armies). Seems to me that the AI acts rather simplistically and acts from some math equation that determines its action.

Why not shoot for the moon, but my fear is that in order to do a decent diplomacy model, the entire AI methodology would have to be rewritten. Can the Ai be programmed to think 5-15 turns ahead? If so I say, yeah lets go for it, but in fairness to the AI simplicity is sometimes better in the long run.

Predictable and inefficent yes, but hardly disfunctional. Part of the solution might be to have more stats for agents and give them influence as well along with titles. If you build a bishop he should be able to attain the title of "archbishop" and raise his influence and maybe be influential enough to become pope.

The rulers influence needs to be considered as well, and dread and command (taking mtw terms) should the AI attack and alliance member, who at the time might be weak militarily but has a king with superior influence?

My points are it seems to be (in MTW and STW) a simple math calculation, by province whether the Ai attacks or not. Can we do better? yes we can, but there has to be some definable formula for the procession of diplomacy. And there in lies the strength of the current system, its simplicity allows you to plan accordingly.

[Bob the Insane]

I think one of the problems with diplomacy in the TW series is that it maths is so hidden so it is hard to make sense of the AI decisions...

CIV4 as an example shows you the stat of your relationship with the other factions based on your past interactions and positions on the map... So when the AI refuses things or cancels trade agreements it is pretty obvious why.

It is also obvious in CIV4 that the AI is trying to win. If you are right next to another faction and you are both about the same size, unless you setup up a good relationship (with reglion for example) the as the CIVs grow there will be friction and a cool relationship but trade and stuff with still go on. But if you do something to leap head of your neighbour you might well find him suddenly start canceling treaties and starting to build up his troops.

The point is that the activities all make sense because the state of your relationship with the other faction (along with the reasons) was very obvious. Many times the actions of the AI in RTW might well be logical but because the player can;t see the whole picture it does not make sense.

Also the AI in TW is very agressive and when it sees even a small military advantage it will attack without any thought to past events.

[Odin]

Also the AI in TW is very agressive and when it sees even a small military advantage it will attack without any thought to past events.

Great points in your post, this one stuck out for me. There in lies the paradox of the TW series, do we want a war game or grand strategy game? I would love to have the AI be able to have multiple variables that leads to its decisions in diplomacy. I think its certainly managable as well to, but the issue for me then becomes the more I know about the AI's factoring into diplomatic decisions the more easily I can manipulate it.

Yet another paradox, the more you know the better you can prepare, where as currently, the less factors in the equation the simplicity of the AI behavior which by defauly allows you to better prepare. CIV as exampled by a few is a great example of what MTW could ideally become, with a few tweeks to the religion aspects that model would be ideal.

[econ21]

Also the AI in TW is very agressive and when it sees even a small military advantage it will attack without any thought to past events.

But also future events. I'm being attacked by the AI when I outclass them militarily and economically. I can't make much sense of it. Maybe they look at the number of enemies ranged against me and think collectively they can win (maybe they could if player skill was equal). Maybe they look at my very poorly defended border cities and think they are easy pickings. The result is that sooner or later I end up at war with everyone. That's ok, for total war, I guess but rather cramps one aspect of the game.

I second the praise for Civ4 as a model for diplomacy. Trying to maintain a peace in Civ4 is almost as tense as trying to win a war. TW does not really have that aspect.