Tolerant AIs should still convert

[TheMeInTeam]

You see it often in India; nations like Malwa and Bahmanis will not convert their provinces to Sunni/Shia. While it's true that tolerance allows them to not take penalties directly, their stability is tenuous due to their unwillingness to convert provinces.

The generic sultanate religious unity requires 3 ideas; this is decades into the game, long enough that nations like Malwa, Gujurat, and so forth could convert all of their provinces before they attain it. Even after that unity NI, they still have more resistance to unrest (in case of later poor legitimacy, overextension, etc) if they convert provinces.

This argument extends to other nations with tolerance as well though.

My suggestion is that unless they are currently receiving +3 tolerance (IE 100% unity) from province that the AI should attempt to convert it using the same logic the AI uses for converting provinces in the general sense.

 

[Grand Historian]

The Bahmanis have a disparaging tendency to completely blob across India. And having tolerant nations convert provinces is counter-intuitive to the point of the bonus.

 

[tobias.mb]

I think Bahmanis take Humanist ideas. So they are eventually at +3 tolerance. But I agree. As long as the AI is not at +3 tolerance and can convert provinces in a reasonable time it should do it.

 

[TheMeInTeam]

The Bahmanis have a disparaging tendency to completely blob across India. And having tolerant nations convert provinces is counter-intuitive to the point of the bonus.

It is more counter-intuitive to have the AI play badly on purpose. Wiz said that if it's effective for the player to be doing it, he wants the AI to do it (during conversations about AI westernizing versus human). Yes, some of the sultanates take humanist eventually, and once they have it finished for +3 in heathen lands it's reasonable not to convert.

Playing 60 years with provinces that are borderline negative in terms of unrest due to religion is weaker than converting primary culture provinces you can convert easily. The AI should not be playing poorly on purpose.

 

[Grand Historian]

It is more counter-intuitive to have the AI play badly on purpose. Wiz said that if it's effective for the player to be doing it, he wants the AI to do it (during conversations about AI westernizing versus human). Yes, some of the sultanates take humanist eventually, and once they have it finished for +3 in heathen lands it's reasonable not to convert.

Playing 60 years with provinces that are borderline negative in terms of unrest due to religion is weaker than converting primary culture provinces you can convert easily. The AI should not be playing poorly on purpose.

Fair enough, but the Bahmanis are a poor example to back up your point; they're already strong enough as it is.

 

[TheMeInTeam]

Fair enough, but the Bahmanis are a poor example to back up your point; they're already strong enough as it is.

It applies to other sultanates too, including ones that less typically succeed like Sind, Bengal, Gujurat or Nagaur. Bahmanis starts with a great ruler and often finds allies that allow it to numerically overpower Vijay who is otherwise stronger, but just because they're strong doesn't mean they shouldn't play better within the context of their position. Ideally, we look at balance *after* we have the AI playing strongly in the context of its position, because that situation best mirrors the balance that would be applicable to a human player in the same scenario.

Edit:

Sind takes religious, which is tailor-made to convert provinces, then will frequently avoid converting provinces.

That's nuts.

 

[TheMeInTeam]

I have tested this mechanic a bit further since I made the suggestion.

In addition to Indian Sultanates, client states also do not convert religion. While they have humanist ideas, they are a good metric because they often have many or exclusively wrong-religion provinces. Even at 100 legitimacy, they take a substantial religious unity penalty on heathen provinces. Christian and Islamic client states can typically convert with anywhere from a 2% to 6% strength beyond province resistance thanks to decisions and/or piety. Even weaker faiths can do that to a degree; most get at least 2% from decisions.

At the very least, nations below 100% unity should probably attempt to convert provinces, as the ducat cost is minimal compared to the administrative points you save in stability and while the province unrest is 6% in most cases the global unrest from religious unity penalty is higher.

For nations that take religious, they should always attempt to convert; it's silly for the AI to take an idea group centered around converting and then refuse to convert. Perhaps it's the most effective to the AI's decision making process to attempt to convert any province that isn't +3 tolerated, but at least avoiding unity penalties is sensible.

That said, aside from this issue humanist + somewhat better coring logic has done heaps to aid the viability of client states.

 

[Grand Historian]

I have tested this mechanic a bit further since I made the suggestion.

In addition to Indian Sultanates, client states also do not convert religion. While they have humanist ideas, they are a good metric because they often have many or exclusively wrong-religion provinces. Even at 100 legitimacy, they take a substantial religious unity penalty on heathen provinces. Christian and Islamic client states can typically convert with anywhere from a 2% to 6% strength beyond province resistance thanks to decisions and/or piety. Even weaker faiths can do that to a degree; most get at least 2% from decisions.

At the very least, nations below 100% unity should probably attempt to convert provinces, as the ducat cost is minimal compared to the administrative points you save in stability and while the province unrest is 6% in most cases the global unrest from religious unity penalty is higher.

For nations that take religious, they should always attempt to convert; it's silly for the AI to take an idea group centered around converting and then refuse to convert. Perhaps it's the most effective to the AI's decision making process to attempt to convert any province that isn't +3 tolerated, but at least avoiding unity penalties is sensible.

That said, aside from this issue humanist + somewhat better coring logic has done heaps to aid the viability of client states.

Agreed, though Client States is probably vassal AI acting up again.

 

[TheMeInTeam]

Agreed, though Client States is probably vassal AI acting up again.

I don't think so. Obviously I don't have any evidence wrt non-vassal client states since that's an anomalous situation, but every instance of AIs at tolerance 0 or higher I see they'll refuse to convert, regardless of subject nation status. Nations without religious or with it doesn't matter from my observations; those without tolerance ideas or humanist will attempt to convert regardless of religious ideas as long as they have negative tolerance.

This becomes really apparent with Catholic client states. A Catholic client state has in-built -tolerance of heretics due to Catholicism, and takes a few of the decisions that add missionary strength at the expense of heretic tolerance. Thus, for this particular faith client states will have mostly 0 or higher heathen tolerance (humanism, legitimacy, no other factors) but -3 tolerance of heretics to offset humanist's +3.

The end result? Catholic client states will convert Christian heresies to Catholic, but refuse to convert heathen provinces of any kind to Catholic unless they happen to have low legitimacy. That's just...awkward.

 

[TheMeInTeam]

Here is more evidence of the poor AI in this regard:



In this screenshot, I have two client states: "Nope" and "Gooby Pls". Neither of them have converted anything up until this point. Just recently, Nope got a low legitimacy ruler. Would you look at that, the instant it happens, they start converting provinces to Sunni. Thanks to me giving them subsidies, they were one of the world's top militaries, but also had built Cathedrals en masse', allowing relatively easy conversion at all times.

Only now, when doing so gives them unrest, are they converting. However, if this AI had converted while at high legitimacy, it would have avoided any unrest issues whatsoever in the first place. If the AI simply converted the fastest province available and mirrored the same building logic it's currently using, the entirety of Nope's territory should be Sunni.

The logic given for AI westernizations is that if it's beneficial, the AI should do it too. Does this not apply to religious conversions? Permanently taking some demerit for religious unity and risking going into intolerance with legitimacy below 50 seems like subpar play compared to simply converting provinces that can be converted easily.

Rather than tolerance >= 0 refusing to convert, I suggest tolerance = 3.