AIs almost never fight each other

[DrewBledsoe]

At the very least, it would have been nice to allow a decent game setup option to control how warlike the AI's would be. The Aggressive AI option by itself is not enough.

I did some investigation of the various XML properties that can be modified to control AI aggression. I read every thread discussion I could find on the subject, and attempted my own modifications of the XML. I ended up changing around 20-25 property values. It had to be done separately for each leader - most of it was in the leader head XML file.

I was able to make the AI's much more aggressive, much more likely to war with close neighbors, and almost completely eliminated positive diplo modifiers. It resulted in a good number of wars, but they were still pathetic wars where very few cities exchanged hands. Basically, the AI's inability to concentrate forces was to blame - in my test games, I posted scouts to watch all the battles.

I tried the same XML mods in tandem with the Better AI mod, thinking that Better AI would be much better at warfare. This worked better, but in the one test game I ran, I was still disappointed in how many cities actually changed hands. I believe I played the game up to about 1300 AD, and 2 cities changed hands in all the wars (other than mine) that were fought. Again, I posted scouts to watch the battles. Basically, the AI doesn't seem to concentrate their forces adequately. They attack before building up a big enough stack, and the attack gets stalled. Given the limited amount of testing I've done with Better AI, I'm not in a good position to criticize, I'm just telling you what I saw.

I'm still mystified as to why some posters say they do see a decent amount of warfare (with many cities changing hands). I know some of them crowd the map quite a bit, but in my experiments, that hasn't helped very much. I can only conclude it must have something to do with other game settings or play style. Or maybe it just has to do with the subjective nature of the question "How much AI vs. AI war is enough?".

Interesting study nice one

Most games I play, I get dozens of ai cities changing hands. I've played a hell of a lot of games, and these I now believe are the keys:- (in no particular order of merit)

1) Ages ago I modded the ai to upgrade at 100% (Ive modded dozens of other balancers but they aren't significant here). This stops "instant mass upgrade", the ai will still upgrade but much more slowly, somewhat akin to a player. Now what ever anyone says, its still much easier to defend than attack. Lets say attacking ai has its typical ragtaggle bunch (lets say early medievel) of horse archers, a mace or 2, mostly swords and axes, and a few cats. Now if the defender can inst-upgrade, then its 5 defenders will all be experienced longbows, and the attacker is going to take horrific casualties (probably attacking a hill too). But with 100% upgrades, that will probably be 1 longbow and 4 archers, resulting in few enough casualties to carry on the fight further.

N.B. By making the ai upgrade at 100%, I needed to give them lots of other bonuses elsewhere, but my "mod" is on hold till after BTS, Im hoping the new expansion will cure most of my woes.

2) Aggressive ais settings. While religion is often far, far too big a modifier (half a dozen nations together under one religious banner all being "best friends forever"), on average, it does result in more combat overall between ais....(I also took away most of the negative modifiers the player receives for playing "agg ais settings". I don't want each leader to automatically act as a jerkoff towards me, its a very artificial way of doing things (as is most of CIV IV diplomacy)...if they arrive at my borders with a huge stack and declare war, then good on them, but just not liking me because they are programmed too is tacky and annoying.

3) combined with 2).....Random Personalities. With normal personalities, its very easy to predict from very early on what will happen in every single game. It gives the player a huge advantage, knowing how each leader will react in advance. I find it also seems to give more chance, combined with agg ais, of each leader being more inclined to war, even though 8 gandhis in one game won't produce much aggresion

4) Marathon. Ok Im biased (its my only game speed), but more turns tends to produce more inter ai wars, and with marathon, its much harder to suddenly construct defence. In fact if a dozen swordsman and half a dozen cats arrive, with a similar stack close behind, against an unprepared foe,then you or the ai being attacked, is going to lose a city or 2 at very least. In normal speed, its much easier to whip/chop defences, and armies tend to be a lot smaller.

5) Huge maps...Pretty obvious here, more land = more cities = more troops = bigger stacks = better chance of taking cities.

6) A couple more civs than the default for the map size, also helps promote more wars too, again obvious.


Just my opinions, and most are oversimplifications and generalisms, but they tend to hold true as an average observation.

 

[Naismith]

1) Ages ago I modded the ai to upgrade at 100% (Ive modded dozens of other balancers but they aren't significant here). This stops "instant mass upgrade", the ai will still upgrade but much more slowly, somewhat akin to a player. Now what ever anyone says, its still much easier to defend than attack. Lets say attacking ai has its typical ragtaggle bunch (lets say early medievel) of horse archers, a mace or 2, mostly swords and axes, and a few cats. Now if the defender can inst-upgrade, then its 5 defenders will all be experienced longbows, and the attacker is going to take horrific casualties (probably attacking a hill too). But with 100% upgrades, that will probably be 1 longbow and 4 archers, resulting in few enough casualties to carry on the fight further.

Interesting - I never considered upgrading to be part of the problem before. I did consider weakening defense-type units like archers and longbows, or weakening city garrison promos, but I discarded it as too much work. (I'm kind of lazy.)

N.B. By making the ai upgrade at 100%, I needed to give them lots of other bonuses elsewhere, but my "mod" is on hold till after BTS, Im hoping the new expansion will cure most of my woes.

They tend to move on to new ground rather than solve old problems, but who knows?

2) Aggressive ais settings. While religion is often far, far too big a modifier (half a dozen nations together under one religious banner all being "best friends forever"), on average, it does result in more combat overall between ais....(I also took away most of the negative modifiers the player receives for playing "agg ais settings". I don't want each leader to automatically act as a jerkoff towards me, its a very artificial way of doing things (as is most of CIV IV diplomacy)...if they arrive at my borders with a huge stack and declare war, then good on them, but just not liking me because they are programmed too is tacky and annoying.

I completely eliminated positive diplo for same religion, and increased negative diplo for different religion. I also setup each Civ to get Mysticism as an additional free tech at the beginning of the game, to encourage the early accumulation of negative diplo from different religions. That worked out pretty well, at least as far as encouraging war is concerned.

4) Marathon. Ok Im biased (its my only game speed), but more turns tends to produce more inter ai wars, and with marathon, its much harder to suddenly construct defence. In fact if a dozen swordsman and half a dozen cats arrive, with a similar stack close behind, against an unprepared foe,then you or the ai being attacked, is going to lose a city or 2 at very least. In normal speed, its much easier to whip/chop defences, and armies tend to be a lot smaller.

I've only played one marathon game. It was just too slow for me. But I believe you are correct for two reasons. One, in comparison to other speeds, units are cheaper to build, which gives the AI a better chance to accumulate enough units to compensate for lack of strategy. Second, I've read elsewhere (Better AI forum?) that many AI declaration of wars are driven by wanting to move a unit somewhere, and not being able to do it because they don't have an Open Borders agreement. (That's why I also modded the XML so the AI had to be Pleased in order to make an Open Borders agreement.) As you know, unit movement is the same in Marathon, but everything else is slower, so I believe there is more chance for wars to occur because of this factor in Marathon. I suspect the (essentially) increased movement factor may also allow the AI to concentrate forces better. I admit a lot of this is guesswork.

5) Huge maps...Pretty obvious here, more land = more cities = more troops = bigger stacks = better chance of taking cities.

6) A couple more civs than the default for the map size, also helps promote more wars too, again obvious.

I completely agree. Again, this is probably one of the factors involved in why I don't see many effective wars in my games. I've never played a huge map. Every time I play a large map, I get slowdown in the late stages of the game, and I'm fairly impatient with that. My computer "only" has 768M of RAM. Sigh.

 

[thadian]

I almost think the AI is set to hold minimal skirmishes because the "AI Paranoia" automaticly assumes the player will suprise-ambush them the second they deploy.

Agressive AI sucks. I dont want every civ including ghandi to make a demand every turn insta warring me if i say no, even if im stronger, and building NOTHING but low level troops - Agg AI leads to nutjobs forming MASS wars against other AI's that are only relevant if Monty takes on Brennus real early, with the same low ammount of battles, and still "sending everything" to fight a player, almost 3-fold what they use against each other.

 

[thadian]

@Naismith

I completely eliminated positive diplo for same religion, and increased negative diplo for different religion. I also setup each Civ to get Mysticism as an additional free tech at the beginning of the game, to encourage the early accumulation of negative diplo from different religions. That worked out pretty well, at least as far as encouraging war is concerned.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
What of those who start with mysticism? shall i assume you rebalance by giving them a diff free tech?

 

[Naismith]

@Naismith

I completely eliminated positive diplo for same religion, and increased negative diplo for different religion. I also setup each Civ to get Mysticism as an additional free tech at the beginning of the game, to encourage the early accumulation of negative diplo from different religions. That worked out pretty well, at least as far as encouraging war is concerned.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
What of those who start with mysticism? shall i assume you rebalance by giving them a diff free tech?

That's what I did. By the way, I also modified the XML to give the human player an extra +1 or +2 base diplo (I forget which) to neutralize the extra negative diplo you (the human) get with the Aggressive AI setting. I've read Aggressive AI removes one of the constraints (some kind of "can we afford it" test) considered by the AI when deciding whether to declare war on human or other AI's.

 

[thadian]

Nice! That sounds awesome, if i may ask, how does the aggro ai play out in your mod?

I assume with your changes, not everyone is an auto-warmonger.

Its a smart way to rebalance things, what does the AI focus on building?
(of course primarily units, but aside the obvious)

I would have never known the AI had such a restraint. How in depth can you alter the computers strategy? Is there a way you can tell it... (not a coder so bear with me)

Britan has 3 cities next to frances 2 cities. France pushed a good force east, and you have a negative diplo with them, and they want to hit your friend, Celtia. Now, Brennus has a +4-8 diplo with you, and of course you know he will come begging within 4 turns. (he can put up a fight on number of units but his promotions are lacking in manpower, so he can hold himself a little)

So when you see that, you start building troops and assembling a force just enough away from their border that they cant see it. (3 spaces form their borderline is good), and using workers to make a road (if theirs not already one) to allow you to send a few fast backups just in case.

Then, when asked you send a decoy line to march around their south city toward their main force. About 2 turns later, when their army is distracted here comes the real army.

Is it possible to Mod in some of these basics? To what extent can you tell the computer to metagame?

I would think if it can consider weather or not it can maintain a war AND its kingdom that it could also be told to better judge a force to send, how to reinforce it?

Is it possible to tell it to "not know the difference" between the player and the other AI's or to tell each AI to treat each other AI as a "human player"?

Obviou

 

[DrewBledsoe]

@ Naismith and Thadian, interesting discussion guys. Naismith, I like your ideas with religion, would you say it's worked?

What got me rolling making changes, is when 2.08 patch was released, and on Monarch or above, the ai starting getting riflemen in around 1300 ish on average. Then with a space race being in full swing often in the 1700s, I realised the tech line wasn't long enough anymore, with the ai being so peaceful and teching so fast. Thats what drove me to make all tech 50% more expensive, which is the basis of my mod, but that leads to an unbelieveable amount of small and large changes elsewhere. And I was just adjusting for one overall lvl (somewhere very vaguely around Monarch, but handicaps don't mean much when some Ive changed to extremes).

It plays a much better game than I was getting with standard 2.08, but there's stuff Im still not completely happy with.

You mentioned the ai wanting to go somewhere and couldn't sometimes starting a war, but Im not sure if wars of this type are of any significance. Agg ais also lowers the relative strength comparison the ai makes, before it decides to go to war, in other words "it gets braver" (or stupider )

I know one of Blake's earlier better ai mod versions, slowed tech down by building ridiculous nos of troops, but apparently all that mostly realised was a large "cold war" with ais and human having masses of troops, but reaching some kind of permanent mexican standoff. So obviously that wasn't the way to go.....I have real hopes for BTS, if they have extended the techline a bit, and made ais more war ready in general. I don't want a version of Rome Total War, but often having ais not fighting a single war ever in 6000 yrs of history is just silly...

 

[SoxSexSax]

I find this thread a bit meh.

Last game I played (monarch), Stalin (AI) invaded Churchill (AI) at about 1800 with well over 30 riflemen/grenadiers/trebs. While the war didn't end in Churchill's utter anihilation, it totally scuppered any hopes he had of winning.

I've also played games where Mr H Capac (AI) has eliminated 4 AI civs singlehanded.

In fact, I would say that on monarch, there is at least one collossal AI war per game, unless the map setup discourages it greatly. And I've played (like I'm sure others have) dozens of games of warlords.

My wife is playing RIGHT NOW as I write this, and Mao Zedong (AI) has just attacked Cyrus (AI) with two decent sized axe/sword stacks.

I do not see the passive AI you speak of, and I suggest that it could be due to playstyles. I (and my wife, as I taught her how to play) am never in the bottom half of the world militarily, which probably makes the AI reluctant to attack me, hence they attack each other. If you have the weakest military, or near to the weakest military, why should the AI target each other over you? You are the easiest target...

 

[Naismith]

Nice! That sounds awesome, if i may ask, how does the aggro ai play out in your mod?

I assume with your changes, not everyone is an auto-warmonger.

Its a smart way to rebalance things, what does the AI focus on building?
(of course primarily units, but aside the obvious)

I would have never known the AI had such a restraint. How in depth can you alter the computers strategy? Is there a way you can tell it... (not a coder so bear with me)

Britan has 3 cities next to frances 2 cities. France pushed a good force east, and you have a negative diplo with them, and they want to hit your friend, Celtia. Now, Brennus has a +4-8 diplo with you, and of course you know he will come begging within 4 turns. (he can put up a fight on number of units but his promotions are lacking in manpower, so he can hold himself a little)

So when you see that, you start building troops and assembling a force just enough away from their border that they cant see it. (3 spaces form their borderline is good), and using workers to make a road (if theirs not already one) to allow you to send a few fast backups just in case.

Then, when asked you send a decoy line to march around their south city toward their main force. About 2 turns later, when their army is distracted here comes the real army.

Is it possible to Mod in some of these basics? To what extent can you tell the computer to metagame?

I would think if it can consider weather or not it can maintain a war AND its kingdom that it could also be told to better judge a force to send, how to reinforce it?

Is it possible to tell it to "not know the difference" between the player and the other AI's or to tell each AI to treat each other AI as a "human player"?

Obviou

Basically, with my changes there's always at least one war going on. I probably overdid it a bit. I changed all the leaders to be more aggressive, so there is less difference between them all now. But Mansa Musa is still not the warmonger that Monty is.

As far as how in-depth the changes go: I'm only modding the XML, so I can only affect certain things. The military strategy examples you gave could never be dealt with by XML mods. You would have to get into the code to do that, and even then I think you couldn't achieve that kind of result very easily.

I think your desire to mod out the prejudice against the human player(s) is more doable, but you would have to mod code to deal with all aspects of it.

Take a look at CIV4LeaderHeadInfos.xml to see the various properties setup for each leader. Some of the properties are self-explanatory (like iBuildUnitProb), but others are not so obvious.

 

[Naismith]

Naismith, I like your ideas with religion, would you say it's worked?

It effectively nerfed religion as a positive force in diplomacy, so if that's what you're after, it works pretty well. Just set iSameReligionAttitudeChange and iSameReligionAttitudeChangeLimit to 0 on every leader, and you too can throw off the shackles of religion. Some people might say it removes an interesting and subtle part of the game, but I think they're wussies. Seriously, I may have overdone it a bit, but I do believe that religion plays too strong a role in diplomacy.

What got me rolling making changes, is when 2.08 patch was released, and on Monarch or above, the ai starting getting riflemen in around 1300 ish on average. Then with a space race being in full swing often in the 1700s, I realised the tech line wasn't long enough anymore, with the ai being so peaceful and teching so fast. Thats what drove me to make all tech 50% more expensive, which is the basis of my mod, but that leads to an unbelieveable amount of small and large changes elsewhere. And I was just adjusting for one overall lvl (somewhere very vaguely around Monarch, but handicaps don't mean much when some Ive changed to extremes).

It plays a much better game than I was getting with standard 2.08, but there's stuff Im still not completely happy with.

Well, trying to mod this thing gives you a new appreciation of what the Firaxis programmers have to face.

You mentioned the ai wanting to go somewhere and couldn't sometimes starting a war, but Im not sure if wars of this type are of any significance. Agg ais also lowers the relative strength comparison the ai makes, before it decides to go to war, in other words "it gets braver" (or stupider )

Yeah, that's what I really want - significant wars. It's very difficult to analyze exactly what affect your XML mods have, sometimes. Especially if you've made multiple changes.

The AI also gets braver when you mess with things like iMaxWarNearbyPowerRatio and such in the leader head infos XML file. Of course, I didn't confine myself to just modifying that, not by a long shot.

I know one of Blake's earlier better ai mod versions, slowed tech down by building ridiculous nos of troops, but apparently all that mostly realised was a large "cold war" with ais and human having masses of troops, but reaching some kind of permanent mexican standoff. So obviously that wasn't the way to go.....I have real hopes for BTS, if they have extended the techline a bit, and made ais more war ready in general. I don't want a version of Rome Total War, but often having ais not fighting a single war ever in 6000 yrs of history is just silly...

Yes, the AI just building a lot more units isn't the answer. I don't want a simple war game, either. Strangely enough, I'm more of a builder, believe it or not. Like you, I seemed to have way too many games where I was the only one declaring war.

 

[DrewBledsoe]

Strangely enough, I'm more of a builder, believe it or not. Like you, I seemed to have way too many games where I was the only one declaring war.

Yeah, Im a builder too. Well, with the explaination that I'll fight hard for land early on, often laying out a city plan of around 18 or so cities as soon as I've explored the land. Ok, I can probably only settle about 8 till my science catches up, but I'll fight any ai who gets too greedy, and starts settling on "my" land, often when they are on the way there (an axe vs an archer in the open is 95%+ a win, if you let them settle, then with the garrison bonus, that goes down to probably 30% or less).....so Im often quite unpopular with several neighbours early on.

I've never been a mass conquerer, 18-21 cities is enough for me (huge maps), and Ill build to my hearts content from then on, unless someone attacks me that is

 

[Arkatakor]

Use Always War option. It isnt working perfectly but it works.

This is lame but at least you are proving my point by showing what drastic measures you are reverting to just to see some action.

I wanna make a note to everyone who has modified their settings / using aggro / big maps / etc like crazy. As I have mentioned before in my posts no other version of Civ required you to modify anything in terms of settings (more civs / aggro / mods) just to see some action.

Furthermore, no other version of Civ had ALL THIS ANTI HUMAN BIAS BULLSHlIT. As far as I am concerned, this game has been shelved and I will entertain myself with other games until this gets fixed. I will not buy BTS hoping that this issue is resolved like i did with Warlords. Instead I will be checking the forums every few months to see if Firaxis bothers to address this core issue that so many of us have stopped playing the game for. Should it be addressed in the expansion, then yes, I will go ahead and buy it. Otherwise Civ4 is history.

And please people, stop making posts like "oh i just played a game and there were wars all the time". Those are lame posts and do not reflect the experience of ALL players who play 100's of games on DEFAULT settings. If you have played that many games on default its statistically impossible to have had so many games with so many wars. Just coz you have one occasionally does not mean its enough. Yes there are inter AI wars but about 50-70% less often then in Civ4. Thats huge.

 

[Drago Askani]

Religion plays a big factor in that whole not as many wars thing. the change in BtS where everyone can found a religion when they research the early religious techs should go a long way to balancing that. When in theory each ai and you will have founded a relegion (standard map size on that theory)
Everyone having a -1-4*modifer from religion should be fairly handy for early wars....

 

[CivCorpse]

Religion plays a big factor in that whole not as many wars thing. the change in BtS where everyone can found a religion when they research the early religious techs should go a long way to balancing that. When in theory each ai and you will have founded a relegion (standard map size on that theory)
Everyone having a -1-4*modifer from religion should be fairly handy for early wars....

Actually I think you get to pick which religeon you found if you are the FIRST to research the tech.

 

[Drago Askani]

**Any changes to religions?
<alexman> There is a new option that allows you to found any religion upon discovering a religion technology** - from dev chat.

You may be right after rereading the way he says it, perhaps the discoverer is the one that gets it first.

 

[klokwerk]

If you go in the Creation / BetterAI forum, you can download a new unofficial version of BetterAI I juste made where wars are very frequent, between AI and with the player. The last 'official' version of BetterAI is known to be much too passive.

Any feedback about this version would be nice.

 

[Naismith]

@DrewBledsoe -

I did another XML mod, making units cheaper to build on Epic, just as they are on Marathon. I also elected to play a large map (huge is too much for my machine). I'm playing a test game now. As usual, with my other XML mods there are plenty of wars, but this time, Cyrus completely rolled over Napolean in record time. It wasn't an axe rush, but a mixture of axes, swords, horse archers, and immortals. I think he took somewhere around 6 or 7 cities. I've never seen anything quite like this (except once) in any of my unmodded games.

I now believe that playing larger maps and reducing unit build cost are the two most important factors in encouraging effective AI vs. AI wars.

You were right.

Edit: I was playing BetterAI as well - but not the newest version - I'll have to try that!

 

[DrewBledsoe]

@DrewBledsoe -

I did another XML mod, making units cheaper to build on Epic, just as they are on Marathon. I also elected to play a large map (huge is too much for my machine). I'm playing a test game now. As usual, with my other XML mods there are plenty of wars, but this time, Cyrus completely rolled over Napolean in record time. It wasn't an axe rush, but a mixture of axes, swords, horse archers, and immortals. I think he took somewhere around 6 or 7 cities. I've never seen anything quite like this (except once) in any of my unmodded games.

I now believe that playing larger maps and reducing unit build cost are the two most important factors in encouraging effective AI vs. AI wars.

You were right.

Edit: I was playing BetterAI as well - but not the newest version - I'll have to try that!

Maybe not completely conclusive but

The random personalities, really can be a factor too...recent game (huge/roughly monarch(my own "mod" thing)/marathon/random pers, 14 total civs)....found myself on a continent with 6 others including Caeser, Ragnar, and Mansa, and I swear they ALL had a Monty personality...I was Zulu and thank god for the fact I could build shock axemen straight out of barracks.

It was utter mayhem, Caeser was virtually permanently at war with me, and dozens of Praets is hard to keep on facing. I ended up with 25 shock axes in one city he kept after. It was war for a while, then eventual peace, then ten turns later "trumpets" and hes after me again. I spent the whole of the first millenium AD building nothing but troops in virtually every city, and neither us of us could make any headway at all. It was just mass combat to a standstill, while I had to use every ruse to try and stay ahead of him in tech, and using crossbows I eventually beat him off and took a couple of cities.....phew that's over I thought, but the very next turn, the other psychos on my continent stopped briefly fighting each other to turn on a now weakened me....

I don't mind tough, but this was just unfair, and I eventually gave up and went to gibber in a corner

With more civs, wars tend to start earlier I've found, but any more than 14 for even the largest map turns into whoever can rush their nearest opponent first, which to me is a)boring b)very repetetive, and c) doesn't really demand much skill or thought at all.

13 or 14 total civs for a huge map is just about right, at least there is a building /settling/development phase first, (sometimes of course its instiwar), how many you play on a large map?

 

[KMadCandy]

for anyone interested that has the time to read examples that to me were very entertaining and enlightening, RB just finished an event that was set up like so, in Vanilla:
Difficulty: Prince / Civilization: America / Leader: Roosevelt / World Size: Large Pangaea / Opponents: Eleven / Options: Aggressive AI has been turned on / Victory: Cultural is the only enabled victory condition

we obviously could only win by culture. so our goal wasn't to go kill everybody. rules were you could never set culture slider above 20&#37;, and you earned +4 points for each AI alive at the end of the game, so some folks that wanted max points tried their best to save AI lives, altho another rules was you could not bribe anybody to sign peace.

the sponsor (Sulla) did some testing and tried to set things up to encourage AI vs. AI wars, not just AI vs. human wars. he changed the map to make the non-warmonger AIs have completely terrible capitals, and chose some very warmonger types and boosted their own lands. his description of how he thought this would lead to max AI vs. AI wars, and report of his own game, is here.

reports from other games are here. you're looking for Epic11 reports.

the way our games unfolded varied WILDLY. some people lost track of the amount of AI vs. AI wars. some people lost count of the times AIs declared on them, up to at least 4 at once. two people had isabella declare on them before 2000 BC. then there's oddball me, i saw multiple AI vs. AI wars but only ever the warmongers vs. the non-warmongers, and the only AI to declare on me was Monty, he doesn't even count!

if that sort of "case study" in how one particular save, with those artificial map/rules things in there, can turn out depending on luck and how the different players approach the game, go check it out. if not, don't, some are quite long. avoid mine--it's probably the longest, and not even done yet *giggle*.

--

i played my first Random Personalities game recently. that was bizarre. isabella DoWd somebody who shared her religion, and she was only cautious with him while being annoyed with two heathens she left alone the entire game. later she attacked someone else (different religion) who capitulated to her a mere 10 turns later. i knew in my head she wasn't really isabella, but that was "oh wow, she is NOT isabella!"

 

[Naismith]

It was utter mayhem, Caeser was virtually permanently at war with me...

I don't mind tough, but this was just unfair, and I eventually gave up and went to gibber in a corner

You have my sympathy. In all the test games I've played with my XML mods, I've been playing Monarch (which is really more like Prince with my mods). At that level, it's fairly easy for me to keep well ahead in power, which is what I wanted. After all, I'm testing AI vs. AI wars, so I'm trying to take myself out of the equation as much as possible. If I were to play a competitive game, this is the sort of thing that could very easily happen to me, especially since I doubled the chances of an AI deciding to dogpile.

13 or 14 total civs for a huge map is just about right, at least there is a building /settling/development phase first, (sometimes of course its instiwar), how many you play on a large map?

On a large map with low sea level, I started with 10 Civs. I think that was the default + 2, but I don't remember for sure. I suspect the AI can war more effectively if they have a chance to grow to at least several cities before they go to war. Otherwise, it tends to be pathetic little stacks of 2 or three units pillaging, or trying (and failing) to take a city. An AI with 4 cities and 20 units total is not going to concentrate their forces enough to take cities, probably. An AI with 8 cities and 40 units is capable of concentrating their forces and taking cities, especially if you're playing with the BetterAI mod.

I'm thinking maybe that a large map, low sea level, with 8 Civs might be about right for me. I mostly play pangaea.