How to make the AI civs keep up with you?

[Zaga]

Lately I've been wondering about the difficulty of the mod, and if it handles certain gamesettings better than others. Reason is that I'm structurally getting on top too early to keep the game interesting beyond the classic period, even though I'm not a particularly good player (builder-type and playing monarch in vanilla). Tried several things - 'minor civs' off, leaders with less useful traits, not using thieves/rogues offensively, no pillaging, stalling researching religious techs so other civs would get them, pangaea, continents, but the end result stays the same. The latest game was from the SVN downloaded on the 5th of februari, but it's something which seems to happen throughout different versions since before november.

Question: Given it's not widely mentioned on the forum, I guess not everyone has this issue. Anyone has advice on what might be the cure or what settings to avoid? Does the game handle prehistoric era badly or should snail-speed be abandoned in favour of normal speed, maybe? I can keep fiddling around with settings and start over and over again, but if a fellow C2C-er already discovered what works it'd save lots of frustration and time on abandoned games (I love how snail-speed gives a sense of developing and continuity, but it makes for testing going slowly).

Latest game, as example:

Spoiler :

Latest game was large map, continents, prehistoric, snail, immortal, no cheating, no minor civs, no exceptional resources at my capital (just figs and dates), and resulting the event of risky experiments for extra health the capital was reduced from 2 to 1 again. I'd expect it to be a setback on other civs. Playing Egypt with the spiritual/industrious trait, so the benefits would only kick in after the prehistoric period.

Instead, when getting to tribalism I already stayed on par with the surrounding civs score-wise, could go on to my religion tech of choice and found ngaism, had to keep holding off researching Shamanism so another could found and by the time Monarchy was researched the score was almost twice that of civ nr. 2. The rest of the civs are primitive beyond salvation then, went on for horsehockys and giggles to Republic but abandoned the game there.

edit: added save game.

I didn't expect to be asked to post it later on and played it just for my own fun, so it's not presented as the ideal testcase. To comment on things you might notice when checking it out:

Spoiler :

*If I'm reading the graph correctly, there's a lasting GNP(gold) lead starting as early as 7000 BC - well before founding the 2nd city and well before getting access to commerce-rich resources like marble or gold. I'm not sure if that includes beakers, but if so then it suggests the advantage starts there already and is not linked to room for expansion, access to resources or other civs cities being destroyed (since that's far later) but something intrinsic to the early game without much depending on interaction between different civs.

*There are three reloads, which I would have avoided of course if consciously making a game for uploading. The three reloads in the course of the game are once for pressing the wrong direction button (there happened to be an autosave that turn and - totally admitted - I'm pretty anal when accidentally sending a captured animal back into the wilderness by such a silly thing), once for accidentally founding Shamanism (which I wanted another civ to have to make for a more interesting game) and once for getting some sleep in between sessions. They make for a 'tainted' save of course, but I don't consider them cheating or influencing the game to get an advantage.

*At the time of building the first wonder in 3500 BC there's already a solid advantage and comfy first place in score for about 2000 years long when looking at the graphs. The wonder-building spree which starts a few centuries after Stonehenge is a result from that solid advantage, instead of causing it: had lots of spare hammers and not much to worry about. Ramses' industrious trait , though kicking in here, therefore isn't the cause of the success - he merely adds to a victory which is already obvious.

*The horses at the capital only appeared later as an immigrating herd. Starting resources were the figs and dates.

*Also, I guess I'd have picked another leadername, sorry.

 

[Sgtslick]

Ramesses is hands down the best leader in this mod in my opinion. Spiritual industrious are the best two traits by a country mile.

Also were you in a war?
AI tends to seek out war early days and often there will appear 1 or 2 super powers some really low civs and a bunch of below average ones.

Also keep in mind AI does ALOT better if they have a decent amount of land to begin with.
If they can only build 8-10 cities they are never gonna really compete with human in my opinion.
They need 20.

 

[Zaga]

@StgSlick: I agree, Ramses is really good, but only after the prehistoric era ended - wonders are starting to play a role only then (industrious trait) and his spiritual trait isn't much of an advantage in the prehistoric period either due to the guaranteed golden age of the cave national wonder. Given that advantage already clearly developed in the prehistoric, I don't think it's due to Ramses in this case. Usually I play Mesopotamian civs or Vikings, and having the same problems there.

Didn't get into war this time - I usually try to avoid it since I prefer building games. If getting into one, I tend to react defensively and just protect homelands unless a fabulous opportunity rises. But keeping a decent defensive force prevents the worst, in general.

20 cities per civ? Hmm... not sure my computer can deal with any map larger than Large But you find the amount of space also plays a role in that early phase of the game? In that case it would point to something I indeed didn't think of, since I tend to add two more civs to the default amount when starting a game (the more the merrier).

Also, if you get to the point of 20 cities I take it you play until quite late in the game. Care to say if you found a difference in AI-behaviour in later ages? Part of me wonders if the AI handles them better and would make up for a less impressive early period.

 

[Sgtslick]

20 cities per civ? Hmm... not sure my computer can deal with any map larger than Large But you find the amount of space also plays a role in that early phase of the game? In that case it would point to something I indeed didn't think of, since I tend to add two more civs to the default amount when starting a game (the more the merrier).

Also, if you get to the point of 20 cities I take it you play until quite late in the game. Care to say if you found a difference in AI-behaviour in later ages? Part of me wonders if the AI handles them better and would make up for a less impressive early period.

Well yeah he gets better and better as game goes on but national wonders +50% too
Try a deceiver or expansive or seafaring civ, they are all piss poor traits
I like to go Random - I got Henry the navigator this game, expansive + organized. I like organized.

What settings do you use anyway?
Try taking off city civic restrictions.
Don't have raging barbs on.
Maybe try no tech trading at all, it might slow down your game tho.
Revolutions off? I think off is better for AI.

If you don't wanna play a larger map just remove some civs I think. Not every civ is gonna get 20 cities just a few. AI does do alot better the more cities they have tho I find. Like there inherent bonus kinda kicks in more the bigger the empire.

I play emp and I have to tech trade whenever I can - often accepting terrible deals just to keep up.
Whooping immortal is pretty impressive tho. Try a new game with less civs and play till gunpowder they tend to really pick up the pace around theology etc

Can i've a saved game, curious what you did to get so far ahead on immortal.

 

[Koshling]

Lately I've been wondering about the difficulty of the mod, and if it handles certain gamesettings better than others. Reason is that I'm structurally getting on top too early to keep the game interesting beyond the classic period, even though I'm not a particularly good player (builder-type and playing monarch in vanilla). Tried several things - 'minor civs' off, leaders with less useful traits, not using thieves/rogues offensively, no pillaging, stalling researching religious techs so other civs would get them, pangaea, continents, but the end result stays the same. The latest game was from the SVN downloaded on the 5th of februari, but it's something which seems to happen throughout different versions since before november.

Question: Given it's not widely mentioned on the forum, I guess not everyone has this issue. Anyone has advice on what might be the cure or what settings to avoid? Does the game handle prehistoric era badly or should snail-speed be abandoned in favour of normal speed, maybe? I can keep fiddling around with settings and start over and over again, but if a fellow C2C-er already discovered what works it'd save lots of frustration and time on abandoned games (I love how snail-speed gives a sense of developing and continuity, but it makes for testing going slowly).

Latest game, as example:

Spoiler :

Latest game was large map, continents, prehistoric, snail, immortal, no cheating, no minor civs, no exceptional resources at my capital (just figs and dates), and resulting the event of risky experiments for extra health the capital was reduced from 2 to 1 again. I'd expect it to be a setback on other civs. Playing Egypt with the spiritual/industrious trait, so the benefits would only kick in after the prehistoric period.

Instead, when getting to tribalism I already stayed on par with the surrounding civs score-wise, could go on to my religion tech of choice and found ngaism, had to keep holding off researching Shamanism so another could found and by the time Monarchy was researched the score was almost twice that of civ nr. 2. The rest of the civs are primitive beyond salvation then, went on for horsehockys and giggles to Republic but abandoned the game there.

If it's too easy on monarch play on a harder setting!

Having said that I agree with you, and I'm playing on deity. I'm trying to improve the AI all the time. The best way to help is to look out for cases it is handling badly that you can see, and post observations about particular specific things it is stupid at (ideally with save games).

From v20 on the AI should be pretty solid at prehistoric, it has new hunting AIs and so on to try to make it hunt and use subdued animals much more efficiently than previous versions did. However, it seems to sometimes get stuck in non-expansive mode and just hunker down when it shouldn't (which I'm investigating)

 

[Sgtslick]

he said he doubles ai score on immortal

 

[strategyonly]

HUH, i'd like to have games like you guys have said about, i am ALWAYs and i mean "always" at least 15-20 techs BEHIND the AI in every game i play, but i dont go for any religious stuff either. If i get one i am satisfied. But i never get to Sed Lifestyle before AI, and from there i get so far behind its hard, infact one of the games i am playing i have my highest str of 15 and even the Barbarians have str over 26 already

 

[Hydromancerx]

I personally think that not having minor civ on is one of the problems. You can get way ahead too fast if you don't start at war with everyone. Having it so you cannot make peace until writing is a major factor in keeping you below the AI IMO.

I play Monarchy/Snail/PerfectWorld2f and I was almost eradicated pre-Renaissance but made a come back (barely) at late Industrial (Steampunk tech saved me).

 

[JosEPh_II]

Less AI. If you go over 12 AI on a large Continent/ Pangea Map only 3-4 will ever develop into a respectable opposition, imho.

I prefer Archipelago or Islands (with 1 for each AI Plus several extra). Then Lakes map with sea level high to form big lakes/small seas. I play Giant or Huge with 8-10 AI and for the most part even on Noble level I have fun interesting games. Sometimes I breeze and then sometimes I have the type of game SO describes. But each one has it's moments of tension.

These type of maps, with a reduced # of AI, gives the AI a good running start. And Navies are a Must.

JosEPh

 

[soltari666]

I have the same issue as zaga.

i usually play 25-30 ai gargantuan maps perfect world 2f with break pangea and minor civs on and emperor difficulty.

because of short distance capitals the ai "panicks" and chokes itself with military which he cannot afford and abandons a reasonable build structure.

another issue is that the ai rarely builds farms or other food improvements and concentrates on mining and camps so growth is considerable stunted.

furthermore i feel the ai hesitates very long swapping to chiefdom/ despotism and expand its city base, which is the final deathknell in the classical period

in my last game carthage spammed footpads as preferred unit ( which seems unreasonable).

i agree with joseph II that gameplay improves with less civs. more time for reasonable buildup.


finally i have to give a lot of praise to the new rogue/spy ai. this is a pain in the ass. i have to defend my improvements constantly and stay on guard.


p.s. multiple religions are pretty overpowered as well, be it only for the sake of building monasteries ( insane tech and culture boni ), which i feel should only be active if they belong to your state religion or you choose "secular" in the civics menu. or better even: as drastic measure add 1-2 dissatisfaction for each religion or major religious buildings ( monasteries, cathedrals and base wonder) in the empire

 

[JosEPh_II]

Also on the maps and set up I described every AI ends up having a Founded religion. So religious Victory is much harder to get. I also Never use Mastery Victory setting.


I Do play with Multiple Religions On and would die if it was taken away from me and the AI. And even with founding several Religions I will stay with State Civic well into the Ren/Industrial Era. I change to Free once I have the better Ag, Welfare, and Army Civics.

JosEPh

 

[Koshling]

The fix to AI attack stack behaviour I pushed to the SVN earlier today may turn out of have a (very) significant impact on AI challenge in the first couple of eras. We'll see...

 

[JosEPh_II]

HUH, i'd like to have games like you guys have said about, i am ALWAYs and i mean "always" at least 15-20 techs BEHIND the AI in every game i play, but i dont go for any religious stuff either. If i get one i am satisfied. But i never get to Sed Lifestyle before AI, and from there i get so far behind its hard, infact one of the games i am playing i have my highest str of 15 and even the Barbarians have str over 26 already

I forgot who posted it originally but getting to Sed Life 1st requires a series of set tech paths. By using it you can usually get there 1st And cover the important early techs.

This is the path: Cave Dwelling->Oral trad.->Tracking->Cult. ID.-> Tribalism( build 2nd city asap)->Hunting->Petroglyphs->Barter->Chiefdom->Sed. Life.

You can add a few other important techs along the way If you cluster your Civic changes. I generally add Gatherer tech before CD.

I suppose I should add this to the Tips and Tactics thread if it's not already there. The player that posted it plays the higher difficulty levels.

JosEPh

 

[Zaga]

Some great reactions, awesome Will be caught up in college too much this week to try a new game, but you definitely give some good ideas - thanks!

@Sgtslick: Not sure which settings are relevant, there are so many these days *grin* (which is a good thing - I like the fancy stuff). The ones I think are relevant here:

prehistoric start
snail speed
no tech brokering
revolutions on
no raging barbs
city civic restrictions
early and civic buildings
multiple production and research
advanced diplomacy

And sure, feel free to take a look - added the save of that game to the initial post.

@Kosling: I played C2C on immortal, I meant to indicate me being a moderate player by stating the level I usually play on with vanilla. Well, played. It's very long time ago I did vanilla. Also, I hope you don't see it as criticism on your work - I imagine it's a very tough job to keep the AI competitive when not just constantly having a lot of new units and buildings added but also brand new game concepts introduced. In fact you're my personal C2C hero, and usually greet your posts in the SVN thread with chirps and squeals of delight

@Hydromancerx: Are you sure about having 'minor civs' turned on is more difficult? I had the impression that 'minor civs' held the AI back - my previous game with vikings was with the minor civs setting turned on, and the other civs were tiring themselves out by sending armies against eachother and razing their new settlements, while I stayed on the sideline and had some nice defences in my cities. While they wasted their resources on waging war and replacing lost cities & killed armies, I just kicked invaders in their glockenspiel at opportune times if they came too close and kept directing my efforts to building. The end result was the same as now - having to actually avoid founding most of the religions and cheering when someone else did get one.

@Soltari: Glad to see I'm not the only one, I almost felt like a pompous ass when stating the problem of the challenge Koshling improved that choice of civic a short while ago, and what I saw till now it worked wonders on civs not staying in anarchy constantly. I still have to play a couple of games to get a complete picture of the change, but he's definitely seeming to be working on the civic stuff. On another note - I experienced that ill-placed focus on rogues/footpads/thieves too in previous games. I remember the indians waging war with stronger germans in a previous game, and the indians kept building thieves to send on suicide missions to my territory even though we were at peace while the germans conquered their cities. Silly indians.

@Joseph: I'll definitely give the islands a shot! Would also possibly eliminate some issues with inter-AI warfare, come to think of it.

 

[Epona222]

@Hydromancerx: Are you sure about having 'minor civs' turned on is more difficult? I had the impression that 'minor civs' held the AI back - my previous game with vikings was with the minor civs setting turned on, and the other civs were tiring themselves out by sending armies against eachother and razing their new settlements, while I stayed on the sideline and had some nice defences in my cities. While they wasted their resources on waging war and replacing lost cities & killed armies, I just kicked invaders in their glockenspiel at opportune times if they came too close and kept directing my efforts to building. The end result was the same as now - having to actually avoid founding most of the religions and cheering when someone else did get one.

I ran into a problem playing a previous C2C version with minor civs turned on - I was playing on a Gigantic Terra map with very few starting civs (but barbarian civs option on so that more civs sprang up, it's usually a good option for the Terra map so that you meet new Civs upon exploring the "New World").

What happened in this particular game was that the original AI Civs were churning out military units (I assume because we were all minor civs they were in a "war" mentality) but because they had no nearby neighbours - just loads of empty terrain - and there was no actual war going on, they were all supporting absolutely massive armies, never having any battles in which to lose units, and had their research rate on 0% because of the unit maintenance/supply costs. I hit the industrial era before a single one of the original AI Civs had got out of the prehistoric.

That was the only game I had massive problems with (in terms of the AI just not being set up for that particular situation), and I abandoned it - so if going for minor civs option I would make sure that there are enough AI civs (or small enough landmass) that they can actually wage war in a way that might be productive (the chance of actually conquering a city or pillaging some improvements etc), because otherwise they can end up getting all dressed up with no party to go to iykwim, and it holds them back supporting units they don't need. This was a few versions ago though, things might be different now.

 

[Koshling]

I ran into a problem playing a previous C2C version with minor civs turned on - I was playing on a Gigantic Terra map with very few starting civs (but barbarian civs option on so that more civs sprang up, it's usually a good option for the Terra map so that you meet new Civs upon exploring the "New World").

What happened in this particular game was that the original AI Civs were churning out military units (I assume because we were all minor civs they were in a "war" mentality) but because they had no nearby neighbours - just loads of empty terrain - and there was no actual war going on, they were all supporting absolutely massive armies, never having any battles in which to lose units, and had their research rate on 0% because of the unit maintenance/supply costs. I hit the industrial era before a single one of the original AI Civs had got out of the prehistoric.

That was the only game I had massive problems with (in terms of the AI just not being set up for that particular situation), and I abandoned it - so if going for minor civs option I would make sure that there are enough AI civs (or small enough landmass) that they can actually wage war in a way that might be productive (the chance of actually conquering a city or pillaging some improvements etc), because otherwise they can end up getting all dressed up with no party to go to iykwim, and it holds them back supporting units they don't need. This was a few versions ago though, things might be different now.

That behaviour is something I think I have now fixed. It happened when an ai player decided it was appropriate to attack a city that it knew about, but was some distance away, and outside it's current visibility (revealed, probably by a scout, but not currently visible). The bug was in the code that decided when it's attack stack size was sufficient to attack the city (at which point those stacks will head for it to attack). When the city was outside of current visibility it's defensive strength could not be determined, and an error in the logic interpreted that as meaning the stack was not yet strong enough. The result is an indefinite build up, that never actually rolls over into an attack unless the ai gains visibility of it's chosen target city by some other means.

After the fix (which was made to the SVN version last week, so will be in v21), once the attack stack reaches a basic viability threshold ( that is not dependent on the specific target) it will begin it's march. When it gets close enough to gain visibility of the target, if it finds it too well defended it will build and wait for reinforcements (before declaring war if it manages to gain visibility without breaking borders) to arrive.

 

[Epona222]

That behaviour is something I think I have now fixed. It happened when an ai player decided it was appropriate to attack a city that it knew about, but was some distance away, and outside it's current visibility (revealed, probably by a scout, but not currently visible). The bug was in the code that decided when it's attack stack size was sufficient to attack the city (at which point those stacks will head for it to attack). When the city was outside of current visibility it's defensive strength could not be determined, and an error in the logic interpreted that as meaning the stack was not yet strong enough. The result is an indefinite build up, that never actually rolls over into an attack unless the ai gains visibility of it's chosen target city by some other means.

After the fix (which was made to the SVN version last week, so will be in v21), once the attack stack reaches a basic viability threshold ( that is not dependent on the specific target) it will begin it's march. When it gets close enough to gain visibility of the target, if it finds it too well defended it will build and wait for reinforcements (before declaring war if it manages to gain visibility without breaking borders) to arrive.

Ah that's fantastic news, thanks for clarifying. In the game I was referring to, one AI civ had 50+ units just milling around just outside his city for thousands of years - every so often I was thinning them out in the hope that his research would stay higher than 0% for a few turns but he was just spamming troops and there was no way I was going to get a good game out of that situation.

I am so pleased to hear that has been fixed, so for future versions I should be OK to start with few, distant, civs? Sounds good - I like war later in the game rather than earlier, hence the few civs on a gigantic map startup - but it's no fun if they don't get developing

 

[Koshling]

Ah that's fantastic news, thanks for clarifying. In the game I was referring to, one AI civ had 50+ units just milling around just outside his city for thousands of years - every so often I was thinning them out in the hope that his research would stay higher than 0% for a few turns but he was just spamming troops and there was no way I was going to get a good game out of that situation.

I am so pleased to hear that has been fixed, so for future versions I should be OK to start with few, distant, civs? Sounds good - I like war later in the game rather than earlier, hence the few civs on a gigantic map startup - but it's no fun if they don't get developing

If you still observe similar broken behaviour in future post a save game please (or any other AI broken behaviour really) and I can analyse what the AI **thinks** it is doing.