PiterAI Kingdoms (new Artificial intelligence)

[bitterhowl]

As for my own mod I definitely tweak AI for different factions - by religious anticipate and some other lore factors. But as for your project with only one AI label - this is the simple way to make one common decision for many cases. I just added it into my code and AI became much more active - before I had tested campaign in hotseat mode up to 840 turn and there were no defeated factions at all; after adding this - at 150 turn 2 factions was defeated.

<!--
            3
            -->
                <min_entry    is_target_faction_to_outlive = "true" stance = "Neutral"/>
                <faction_attitude pts_desire="400" invade_priority = "1500" continue = "true"/>
            </decision_entry>

It uses win_conditions outlive_faction parameter. I remember that you mentioned it's bugged and decided to try to play with it myself. So with this entry and any to_outlive factions in win conditions it makes sence very much without boring coding in descr_strat.

For my mod I plan to add more pts_desire parameters for more flexible AI behavior.

[Piterx93]

As for my own mod I definitely tweak AI for different factions - by religious anticipate and some other lore factors. But as for your project with only one AI label - this is the simple way to make one common decision for many cases. I just added it into my code and AI became much more active - before I had tested campaign in hotseat mode up to 840 turn and there were no defeated factions at all; after adding this - at 150 turn 2 factions was defeated.

<!--
            3
            -->
                <min_entry    is_target_faction_to_outlive = "true" stance = "Neutral"/>
                <faction_attitude pts_desire="400" invade_priority = "1500" continue = "true"/>
            </decision_entry>

It uses win_conditions outlive_faction parameter. I remember that you mentioned it's bugged and decided to try to play with it myself. So with this entry and any to_outlive factions in win conditions it makes sence very much without boring coding in descr_strat.

For my mod I plan to add more pts_desire parameters for more flexible AI behavior.

Why set it up? I don't want the faction to mindlessly want to destroy another. And so in other files it is set so that they do not want alliances among themselves, that they have worse relations.

More is not better. The more factors, the less important everyone is.

[bitterhowl]

Why? Just because every mod (and even vanilla) has some background (historical or lore for fantasy) and according to this sometimes one faction or even culture tends to mindlessly smash another. It's not abstract chess game, it's a game about mankind.

But even if not talk about history - if you remember I asked how to force faction to find and attack weak neighbours acting as pirates, campaign_script was an answer. But have to say CS doesn't do the job even with raising invade_priority. And pts_desire do it well in combination with low military_balance.

Engine counts pts_desire anyway by internal algorythm and this factor seems to be important. Faction_to_outlive is just an example, you may avoid it. I think we can help AI to collect more strenght on one direction by checking if target is most_desirable or not. If yes - we set proactive invade tactics if no - opportunistic or no invasion at the moment.

I thought invade_priority tells faction how much does it want to atack the target. But it seems that it tells which ivade tactic to use (ofc maybe I'm wrong). I created a thread about that in Workshop.

[Piterx93]

Why? Just because every mod (and even vanilla) has some background (historical or lore for fantasy) and according to this sometimes one faction or even culture tends to mindlessly smash another. It's not abstract chess game, it's a game about mankind.

But even if not talk about history - if you remember I asked how to force faction to find and attack weak neighbours acting as pirates, campaign_script was an answer. But have to say CS doesn't do the job even with raising invade_priority. And pts_desire do it well in combination with low military_balance.

Engine counts pts_desire anyway by internal algorythm and this factor seems to be important. Faction_to_outlive is just an example, you may avoid it. I think we can help AI to collect more strenght on one direction by checking if target is most_desirable or not. If yes - we set proactive invade tactics if no - opportunistic or no invasion at the moment.

I thought invade_priority tells faction how much does it want to atack the target. But it seems that it tells which ivade tactic to use (ofc maybe I'm wrong). I created a thread about that in Workshop.

You're wrong. History is not a propaganda fairy tale but just a game of chess. You know too little about human history.

In my AI, invasion priorities work. Apparently you are doing something wrong.

You need to check how changes in AI affect your game. It doesn't make sense to watch AI fight against each other.

[bitterhowl]

Well, if I'd like to discuss about propaganda tales of medieval times I'd ask about deeds of Karl Martell or Crusades - what is propaganda about that but I got pretty enough propagana IRL and get tired with it. You mentioned propaganda first so there is a question who is much involved)

That's a bit strange for me when person rejects new information. I'd understand if you'd say - I tryed pts_desire many times and it doesn't work. But seems you didn't. I'm only on the very begining of my way with CAI and want to thank you again for your investigations and explanations. I will post my next results in workshop thread.

[Piterx93]

Well, if I'd like to discuss about propaganda tales of medieval times I'd ask about deeds of Karl Martell or Crusades - what is propaganda about that but I got pretty enough propagana IRL and get tired with it. You mentioned propaganda first so there is a question who is much involved)

That's a bit strange for me when person rejects new information. I'd understand if you'd say - I tryed pts_desire many times and it doesn't work. But seems you didn't. I'm only on the very begining of my way with CAI and want to thank you again for your investigations and explanations. I will post my next results in workshop thread.

I can't write what I think in English. The point is, it won't make AI smarter with my dish.