Missionary Limit Discussion

[os79]

To avoid cluttering up Bug Report and Feedback thread, I'm setting up a new thread to discuss how to limit missionaries.

I have a poll above as well.

Basically, you should vote whether:

• Missionaries should be limited to default low # of 3 and no gold cost
• Missionaries should be limited by 5-7
• Missionaries should be limited by 10 with 1/turn cost
• Other...

If a vote is for Other, please explain yourself. Also, if overwhelming # of voters want Other, then you need to wait for Afforess' decision on that.
Otherwise, other options can be tweaked by me into SVN for future patch.

 

[jfuesting]

I voted for the 10/+1, but if it's possible, the AI shouldn't build more missionaries than there are cities without that respective religion. Also, if it's automated missionaries are idling, it should be deleting them down to a small handful.

 

[volus]

I voted for 10/+1 as the best choice, but my actual preference would be as-is with an AI correction.

 

[JosEPh_II]

I vote for returning to the RoM 3 limit and No gold cost.

10 with 1 gold/turn means that the AI will build 10 missionaries costing it 10 gold a turn every turn because every time it uses a missionary it will build another to replace the used one. This is a major hindrance to the AI. Tack a 10 gold/turn onto your early game and see how hard it becomes for you to grow, defend, research etc. It's another AI Handcuff.

AND already has way too many items that eat gold at a prodigious rate. It's become a "mantra" around here that having a healthy gold balance is "bad". But the AI is the one suffering/affected because of it.

JosEPh

 

[Afforess]

The removal of the limit was because it doesn't scale well for map sizes. 10 missionaries sounds great on large maps, but floods small ones. 3 Missionaries are perfect for tiny maps, but useless for larger ones. The decision wasn't to hurt player's gold supplies, as Joseph implies; but because gold simply scales better, while still creating a soft-limit. If the AI is over-producing missionaries, it's an AI issue; not a limit issue. I'll curb the AI's overabundance of missionaries next patch.

 

[Dancing Hoskuld]

My problem is that I want to be able to get them out of the build queue so I build the limit and they no longer appear and the list of units is smaller! I do this with all the national units but at least I use the missionaries and CEOs.

Edit: It is a pity that national unit limits cant scale with map size and game speed. It would simplify things.

Have you considered changing the whole way religions spread? EusebiusWorldReligion-Revival did it in Python. In that way missionaries become strategic units rather than tactical ones and the limit of 2 or 3 as in RoM was fine.

Btw I suppose I will need to make a variation to AAranda's religions to fit with what ever you choose. I did notice that a couple of the religions missionary units are not behaving like the other missionary units eg you can't build them without the required building even if your chosen civics say you should.

Edit 2: This is the basic algorithm used by Eusebius to determine if a religion spreads to a city that turn. Its a bit confused because I am in a hurry, sorry.

1. religion can spread to any city in a civ. Spread occurs so that there can be one religion per 2 pop (+1)

          if (countReligions(pCity) >= ((pCity.getPopulation() / 2) + 1)):
               return False

2. only religions already in the civ can spread to a city
3. the probability that any religion spreads is based on religious civic

Note. If the city pop 5 has one religion and the civ has 5 religions in it then try to spread each religion to the city until there are 5/2+1 = 3 religions in the city. I would suggest trying in the order based on spread rate.

 

[os79]

The removal of the limit was because it doesn't scale well for map sizes. 10 missionaries sounds great on large maps, but floods small ones. 3 Missionaries are perfect for tiny maps, but useless for larger ones. The decision wasn't to hurt player's gold supplies, as Joseph implies; but because gold simply scales better, while still creating a soft-limit. If the AI is over-producing missionaries, it's an AI issue; not a limit issue. I'll curb the AI's overabundance of missionaries next patch.

In that case, how do I close the poll earlier than the date stated? Since you have a better solution .

 

[JosEPh_II]

Afforess wrote:The decision wasn't to hurt player's gold supplies, as Joseph implies; but because gold simply scales better, while still creating a soft-limit. If the AI is over-producing missionaries, it's an AI issue; not a limit issue. I'll curb the AI's overabundance of missionaries next patch.

I Stated that it Hurt the AI.

10 with 1 gold/turn means that the AI will build 10 missionaries costing it 10 gold a turn every turn because every time it uses a missionary it will build another to replace the used one. This is a major hindrance to the AI. Tack a 10 gold/turn onto your early game and see how hard it becomes for you to grow, defend, research etc. It's another AI Handcuff.

Nor is the Statement:

AND already has way too many items that eat gold at a prodigious rate. It's become a "mantra" around here that having a healthy gold balance is "bad".

an implication. It is the rationale behind many of the recent changes discussed here in the subforum. And is My observation. Sorry you don't like it.

And finally:

But the AI is the one suffering/affected because of it.

Is also a true and observable statement that has been voiced by more than just myself.

If it's a simple AI fix then cool.

JosEPh

 

[os79]

Look, Afforess had spent hundreds, if not thousands, of hours reading how AI work in the code. What he said potentially have a strong "expert-like" punch to it, OK? I'm not saying that he is perfect but saying that we should take his word with somewhat more confidence than you seem to show in him, OK?

Continue analyzing and criticizing Afforess, we need that. However, please keep in mind that Afforess actually know what he is about as well , is my advice.

 

[Dancing Hoskuld]

Look, Afforess had spent hundreds, if not thousands, of hours reading how AI work in the code. What he said potentially have a strong "expert-like" punch to it, OK? I'm not saying that he is perfect but saying that we should take his word with somewhat more confidence than you seem to show in him, OK?

Continue analyzing and criticizing Afforess, we need that. However, please keep in mind that Afforess actually know what he is about as well , is my advice.

Except it may be the wrong solution to the problem .

Even geniuses and experts make silly mistakes sometimes. Like the one who put two dog doors in, one for the small dog and one for the larger dog. Whereas the better solution was to put in one for the large dog which was light enough for the smaller dog to use.

Actually thinking, is this change part of Better RoM or part of AND? If it is Better RoM then I may as well give up and say that AAranda's religions are not to be used with it.

 

[dudeSDCA]

The removal of the limit was because it doesn't scale well for map sizes. 10 missionaries sounds great on large maps, but floods small ones. 3 Missionaries are perfect for tiny maps, but useless for larger ones. The decision wasn't to hurt player's gold supplies, as Joseph implies; but because gold simply scales better, while still creating a soft-limit. If the AI is over-producing missionaries, it's an AI issue; not a limit issue. I'll curb the AI's overabundance of missionaries next patch.

Perfect -- Thx Afforess (and welcome back!)

 

[os79]

Except it may be the wrong solution to the problem .

Even geniuses and experts make silly mistakes sometimes. Like the one who put two dog doors in, one for the small dog and one for the larger dog. Whereas the better solution was to put in one for the large dog which was light enough for the smaller dog to use.

Actually thinking, is this change part of Better RoM or part of AND? If it is Better RoM then I may as well give up and say that AAranda's religions are not to be used with it.

Did you even read the last sentence in my post? That answer your skepticism?

 

[Fimbulvetr]

If the AI is over-producing missionaries, it's an AI issue; not a limit issue. I'll curb the AI's overabundance of missionaries next patch.

I reported that issue long long time ago, Specific Bug Reports posts #439 and #443. Since the AI has troubles even if there is a limit of 3, please post fixes to Better AI forums too.

My idea for a fix would have been: loop through all player cities, and check if there is already any missionary standing on a city plot doing nothing. If yes, don't buld any more.
That would have worked (by limiting the number of excess missionaries to 1) for the situation I saw because all unneeded missionaries will simply stand inside the city they were built in.

But the situation seems different in this mod, what exactly is the AI doing with these masses of missionaries?
In the screenshot by Cyrusfan it doesn't look like the units stay inside cities doing nothing, they are actually moving around?

 

[Afforess]

I reported that issue long long time ago, Specific Bug Reports posts #439 and #443. Since the AI has troubles even if there is a limit of 3, please post fixes to Better AI forums too.

My idea for a fix would have been: loop through all player cities, and check if there is already any missionary standing on a city plot doing nothing. If yes, don't buld any more.
That would have worked (by limiting the number of excess missionaries to 1) for the situation I saw because all unneeded missionaries will simply stand inside the city they were built in.

But the situation seems different in this mod, what exactly is the AI doing with these masses of missionaries?
In the screenshot by Cyrusfan it doesn't look like the units stay inside cities doing nothing, they are actually moving around?

It seems something like this would fix the issue:

/************************************************************************************************/
/* Afforess	                  Start		 08/18/10                                               */
/*                                                                                              */
/*  What about Civics that block religion spread and open borders?                               */
/************************************************************************************************/
/*
        iCount += ((pArea->getNumCities() * 2) - (pArea->countHasReligion(eReligion) * 3));
        iCount /= 8;

        iCount = std::max(0, iCount);

		if (AI_isPrimaryArea(pArea))
		{
			iCount++;
		}
*/
		int iLoop;
		for (int iPlayer = 0; iPlayer < MAX_PLAYERS; ++iPlayer)
		{
			CvPlayer& kLoopPlayer = GET_PLAYER((PlayerTypes)iPlayer);
			if (kLoopPlayer.isAlive() && kLoopPlayer.getID() != getID())
			{
				if (GET_TEAM(getTeam()).isHasMet(kLoopPlayer.getTeam()))
				{
					if (GET_TEAM(getTeam()).isOpenBorders(kLoopPlayer.getTeam()) || GET_TEAM(getTeam()).isLimitedBorders(kLoopPlayer.getTeam()))
					{
						if (!kLoopPlayer.isNoNonStateReligionSpread())
						{
							for (CvCity* pLoopCity = kLoopPlayer.firstCity(&iLoop); pLoopCity != NULL; pLoopCity = kLoopPlayer.nextCity(&iLoop))
							{
								if (pLoopCity->area() == pArea)
								{
									if (!pLoopCity->isHasReligion(eReligion))
									{
										if (AI_plotTargetMissionAIs(pLoopCity->plot(), MISSIONAI_SPREAD, NULL) == 0)
										{
											iCount++;
										}
									}
								}
							}
						}
					}
				}
			}
		}
/************************************************************************************************/
/* Afforess	                     END                                                            */
/************************************************************************************************/

 

[JosEPh_II]

Look, Afforess had spent hundreds, if not thousands, of hours reading how AI work in the code. What he said potentially have a strong "expert-like" punch to it, OK? I'm not saying that he is perfect but saying that we should take his word with somewhat more confidence than you seem to show in him, OK?

Continue analyzing and criticizing Afforess, we need that. However, please keep in mind that Afforess actually know what he is about as well , is my advice.

Please os79, I'm asking you nicely, stop. No More Public Defender! It has become, and is, tiresome and irksome, Please Stop it.

Did you even read My last line?

If it's a simple AI fix then cool.

Ever since I posted objections to some of killtech's changes I've been portrayed as the Evil trouble maker to the AND kingdom. And you sir have continued the vilification at every turn. Please Stop.

We all have good and bad days. And we will never completely agree to everything. I wouldn't be here if I were not trying to Help make AND and RoM better.

Have a Nice evening.

JosEPh

 

[Fimbulvetr]

It seems something like this would fix the issue:
(...)

Looks very expensive (AI_plotTargetMissionAIs loops over all player units)
and it won't catch the cases like the one I discribed, where the civ has open borders with the target but not with the different civ inbetweeen, which blocks all paths.

if (AI_plotTargetMissionAIs(pLoopCity->plot(), MISSIONAI_SPREAD, NULL) == 0) <- this will be false if any missionary goes there, you probably wanted it to only be false if an eReligion missionary is on its way.

You are definitely on the right track though; just skip that plottargetmissionais check, instead subtract the number of eReligion's missionaries you have in that area in a seperate step. (Loop over player units, if (GC.getUnitInfo(eUnit).getReligionSpreads(eReligion) > 0) iCount--;)

For the case where Open Borders is there but no path, the check for bored missionaries is still needed I think. Also long as they stay inside the city that built them, handling that case should be easy. Any coment on the missionary activities n Cyrusfan's screenshot?

Also you skipped the division by 8 at the end. On purpose? Even for internal spread the number is divided by 2 if bHoly, 4 if not, so I suggest at least dividing by 4. Or 3 if going for some religion-based victory condition.

 

[Afforess]

I'm wondering if AI_neededMissionaries should be cached and recalculated each turn. Even without looping for all the selectiongroups, its going to be expensive, and it's called by cities several times a turn AND by AI_unitValue, which is called all over the place, frequently. Unfortuantly, I don't see a good way to cache it, except maybe by area...

 

[Fimbulvetr]

I'd store it on city level, everything else looks tedious.
Area:
int** m_ppaiPlayerMissionariesNeeded = new Array [MAX_PLAYERS][GC.getNumReligionInfos()];
Looks fun too but idk ..

thx, I checked it out. debug mod and CTRL+D it is. I should test that some day.

 

[Afforess]

Yuck. I'll test it out on some late-game games, and see what my profiling turns up, without caching. Maybe I worry too much...