Why threat level not used in faction system?

[Unready]

It seems that the new faction system is missing an important parameter for vassals organizing and joining the independence faction - threat level.
Under normal circumstances, a vassal of the emperor received the duchy or kingdom will not rebel immediately against his successor
in order to break away if, after the victory, he would be surrounded by enemies far more powerful than he.
There is no point Duke of Sinai to separate from HRE because immediately after the separation, he will receive the Jihad of Egypt from Fatimids and lose all holdings long before his possible victory.
Therefore, if stronger neighbors laid eyes on your lands sound policy requires to consider that only a presence in the strong сountry allows you to keep your possessions.

Similarly, I do not see the point in that German vassals to revolt against the German emperor as on just taken from the Golden Horde in the Polish territories with Tengri religion.
Horde completely cut them all in the next 10 years and this is virtually guaranteed.
The question is why joining the independence faction they ignore these threats.

Strictly speaking this also applies to Lower Autority factions.
Its clear why this to powerful dukes, but why the weak counts are adjoined to them if their independence from neighbor dukes guaranteed only by High level of Autority.
Should not they be contrary to strive to increase this Level and not lower it.

 

[Zireael]

I agree with you completely - the AI should take future threats in account.

 

[The_Blind_One]

I agree with you completely - the AI should take future threats in account.

While they are at it. They can try and invent AI sentience...

To be fair, programming something like that is incredibely complex and prone to mistakes and bugs...

You'd be asking the computer to make calculated estimations to its future potential and threats. Computers are terrible at that. You can only program a computer for certain conditions. Threats and chances are something vague and inconcrete and don't present themselves the same way every time. You can't program for every possible condition.

Perhaps in 20 years we'll have better AI, but so far that department hasn't made much improvement these past 20 years.

FPS enemy bots are retarded and follow a simple detection and attack program.

Skyrim AI characters have like 2 lines, start talking to you out of the blue and walk around the place like they are on a schedule. That's not much different from the old school rpg's where the character walked outside their house during the day and was inside during the night...they also had like 2 lines of dialogue.

Strategy games either just give AI insane bonusses or they program their AI to maximise resources and micromangement and follow a simple program of steamroll the player to death with godlike micromanagement and preset ''best'' strategies.

None of this stuff hasn't been done before. Graphics and gameplay elements have improved tremendously these past years. But the AI is barely able to keep up with the improved gameplay.

 

[Oktobermensch]

While they are at it. They can try and invent AI sentience...

To be fair, programming something like that is incredibely complex and prone to mistakes and bugs...

You'd be asking the computer to make calculated estimations to its future potential and threats. Computers are terrible at that. You can only program a computer for certain conditions. Threats and chances are something vague and inconcrete and don't present themselves the same way every time. You can't program for every possible condition.

Perhaps in 20 years we'll have better AI, but so far that department hasn't made much improvement these past 20 years.

FPS enemy bots are retarded and follow a simple detection and attack program.

Skyrim AI characters have like 2 lines, start talking to you out of the blue and walk around the place like they are on a schedule. That's not much different from the old school rpg's where the character walked outside their house during the day and was inside during the night...they also had like 2 lines of dialogue.

Strategy games either just give AI insane bonusses or they program their AI to maximise resources and micromangement and follow a simple program of steamroll the player to death with godlike micromanagement and preset ''best'' strategies.

None of this stuff hasn't been done before. Graphics and gameplay elements have improved tremendously these past years. But the AI is barely able to keep up with the improved gameplay.

@OP, And that is why you ought to play a multiplayer game. You know, with creatures that have brains and can make such estimations and assesments.

 

[Unready]

Threat levels already used in a game: for example when AI answer to your offer of vassalisation (called power). Question is - why AI not use same criterias when think about independence if after the war he join my empire again (if of cause he will he enough time for that)?

My suggestions are not so complicate the AI​​ as you said and will not slow down the game at all.
In fact they can write a few dozen lines of code in open files in a couple of additional conditions and modifiers in decision making parts.
I suspect that the authors of major mods will do so in the near future.

However, I would like that would vanilla game saved spirit of the time.

Because single vassals uprisings across realm you explain by the panic, anger, despair, hatred etc.
But continued participation in factions with possible catastrophic consequences for his future if he win I cannot explain.

 

[riknap]

@OP, And that is why you ought to play a multiplayer game. You know, with creatures that have brains and can make such estimations and assesments.

so are you saying the OP should only play when every other vassal and co-vassal is a human player?
(yeah I know you were probably being sarcastic... )

still, the point remains that currently AI vassals who do not like their liege prioritize being independent, which while realistic historically isn't a good strategy in-game. An AI vassal who hates the liege will want to join/create an independence faction without factoring in things like geopolitics/survivability. It might be possible to make the AI "recognize" these parameters, the same way a small kingdom often has the "desires better alliance" malus with the neighboring empire yet occasionally has a "political situation[sic]" bonus in other occasions. Of course, since I'm not a software engineer, I can't say if it's plausible to write such algorithms WITHOUT hogging up too much computation cycles (remember, the reason why the devs didn't give baronies courts, ie. make them playable, is because doing so exponentially increases the number of characters and thus computation cycles per "day" which raises the minimum PC-spec bar not to mention additional development time and manpower, without necessarily adding significant gameplay content

 

[unmerged(362834)]

While they are at it. They can try and invent AI sentience...

To be fair, programming something like that is incredibely complex and prone to mistakes and bugs...

You'd be asking the computer to make calculated estimations to its future potential and threats. Computers are terrible at that. You can only program a computer for certain conditions. Threats and chances are something vague and inconcrete and don't present themselves the same way every time. You can't program for every possible condition.

Perhaps in 20 years we'll have better AI, but so far that department hasn't made much improvement these past 20 years.

FPS enemy bots are retarded and follow a simple detection and attack program.

Skyrim AI characters have like 2 lines, start talking to you out of the blue and walk around the place like they are on a schedule. That's not much different from the old school rpg's where the character walked outside their house during the day and was inside during the night...they also had like 2 lines of dialogue.

Strategy games either just give AI insane bonusses or they program their AI to maximise resources and micromangement and follow a simple program of steamroll the player to death with godlike micromanagement and preset ''best'' strategies.

None of this stuff hasn't been done before. Graphics and gameplay elements have improved tremendously these past years. But the AI is barely able to keep up with the improved gameplay.

I agree programming is the hardest part in a game... but there is always a simple way to lie down complex things...

In CK2, AI already uses few things he mentioned... like it already knows how to assess neighbors strengh (based on the potential size armies!)... then it picks weakest neighbors to conquer and avoids provoking strongest ones.

Therefore, it isn't of big deal to make weak vassals support high CA... such vassals would only need to assess their neighbors strengh compared to theirs... if they're surrounded by strong dukes or by strong foreign rulers, they won't support CA weakening and they won't seek independence unless their liege is trying to revoke them...

I noticed that kingdoms become more vulnerable to factions after their armies are wiped out... When engaged in offensive wars, maybe AI controlled kings should aim white peace if defeating enemies would induce huge losses on their side... such reasoning can be done by using again the script that AI uses to identify to assess neighbors' strengh...

I'd also like to suggest to make one-region kingdoms refuse to be joined by strong dukes, since any new vassal can create a faction to change the ruler immediately after joining the kingdom...

 

[Zireael]

I agree that since such a code is already present ingame, it should be used.