AI at higher difficulties, kinda weak?

[shockk]

Anyone know when the ai bonuses start kicking in at higher levels? Turn 96 on the 2nd hardest difficulty, and am looking around at AI cities and i don't see any large modifiers on their resource productions. The 2 other world powers have armies ~1000 power, around the same i can field before going into negative income (and i rarely build money buildings). I am 133% bigger then Rome, and 2x bigger then Macedonia(the next largest) and that's playing as Nubia. Sure Nubia probably isn't that hard to play, but the poor terrain is slowing down the development of the country compared to the rest of the world. Am still teir2 government if that matters. Mainly want to know if it's worth continuing and i simply haven't triggered the higher AI buffs yet.

Also noticed in alot of the desert areas the ai sometimes is really bad at building it up. Alot of the Arabian Peninsula (which has been colonized at least 50 turns ago) has less then 5 buildings. Some regions only 1-2 building and 0 workers on production. Compared to my desert lands, that have 20+ pop, ran out of building zones and im starting to deconstruct infrastructure buildings (and Nubia has a ton of crappy land around it at the start). Not expecting the ai to be perfect, but its rather far behind and should have good bonuses at this difficulty. (will say Saba starting province is well developed but it also has some amazing starting buildings to help it)

Also Saba likes to send armies of ~5 troops into my fleet at Mara Sabaticum. My fleet has been parked there for 30+ turns, and its on its coast line so i assume the ai can see the fleet?

[MoLAoS]

Big issue you probably have is the AI likely doesn't understand the province system. Provinces, especially when you manage them as a player, drastically buff production. Developing individual regions, which likely most AI powers do, is quite inferior even if you do it optimally. 5 buildings isn't bad for a non-province region over 50 turns, especially Arabia which is very mediocre land. Of course I doubt the devs balanced buildings mathematically anyways. Most developers are not good at calculus or algebra in economic contexts. Similarly most devs are small and don't have an AI expert who also really understands their game. No way the AI can be expected to compete with you in the area of the world you are in. AI in better areas have more of a shot. Your numbers are not helpful. Bigger than Rome how? Also is Rome the biggest nation? Population, landmass, building usage? I won the game on turn 70 as Rome at the third difficulty. Mainly because large AI nations had bad luck giving me the easy early legacy lead win, which at the time I wasn't even aware of. Are you playing out battles yourself? That's an unbelievable advantage over the AI. You save thousands of gold if you never get stack wiped or even heavily damaged.

[shockk]

My point isn't why is the ai weak, the ai is always going to be weak to the player (unless machine learning becomes cheap and easy). The normal way to solve this is the ai gets a bunch of bonus to production at higher levels. Point was more of why is the ai not out preforming me when it should has alot of buffs at this difficulty.

The ai does understand the province system to a degree, at least from playing around with the auto builder. It smart enough to build as few building at a time as possible

When i was talking about size, i meant number of regions.

I don't play out battles in fog2 but even if i did that wouldn't explain why the ai armies don't out number me at high difficulties

When i said less then 5 i mean like 2-3 buildings. It should have more then this after 50 turns. Yes i understand the land is grabbage, giving only 2-3 food and industry production per worker. Also talking about the ai holding the whole province so it should be pooling production, even in not it should be doing better.

Off topic but i find it extremely hard to believe devs that make complex 4x/grand strat games don't understand numbers lol (expect TW games...). Programming at its most basic level is basically algebra so any programmer is probably decent at it

[loki100]

it doesn't get production bonuses as such, so no reason why it should generate much larger armies etc.

Its advantages come in at the level of Empire management and your difficulties fall into the same area.

So the barriers in the game to your expansion hit you earlier and the AI has more leeway.

I sort of think this is a lot like the AI in Pride of Nations, at one level its far far better but the key to that game was in effect you were dealing with your own internal challenges and not really with the AI controlled powers?

Edit:

My instinct is that at the level of region management it does a few things badly:

a) it really doesn't sell its slaves so tends to loyalty problems (despite its bonus in this regard) so not only revolts but also baseline production slips back. I find if I take a mature AI region, its a mess in this regard.
b) more generally it tends to ignore loyalty in its pop and building assignment ... I'd like to see the balanced auto system weight this a bit more

[MoLAoS]

Never seen developers that understood the consequences of their economic system. Not one single 4x or GS game. Never. Much less devs that could teach their AI to optimize in the system. CIVI is a prominent example. Massive game breaking exploits like the scythian thing. AI testing should have easily discovered this even if devs didn't but then they probably run shitty scripted AI. Same problem with space clone GalCiv 2/3. Even Pandora's famous AI was actually mediocre and mostly through static script. True machine learning would be one solution, but not really necessary. Smart human could make a functioning AI if they knew what they were doing.

FoG would explain why you had the ability to develop large and more resource intensive armies though, because you would have been massively more resource efficient. AI wouldn't outnumber cause they hadn't snowballed, and were constantly getting dogpiled after a major army loss. In a game with two players if one loses a whole army to AI or each other, may as well surrender, same with AI. Losing strong army is fine against shitty AIs who are also losing armies all the time but a smart human who never loses an army has a huge advantage from then on.

I don't understand your auto builder comment. You should always be building something in every single province. I would understand being starved for resources so you would argue that building one at a time means each building spends less time in construction, so you get two buildings after x turns but one you get halfway instead of both finishing at the end, however this should never be relevant. Never been in a position where I wanted to build say, only in 4 of my 7 regions. Maybe if you are really weak I guess and have no infrastructure. But gains are marginal at best.

Also what I mean by understand the province system isn't what to do when you have them, but how important it is to get them in the first place. Having a 7 region province vs having 7 regions among 3 provinces, which happens to the AI all the time, is a massive difference. AI is worrying about dumb shit like diplomacy. Player just wrecking whoever he has to wreck to get that sweet sweet province formation icon. AI is role player and player isn't. As a human having superior diplomat is a massive buff. As an AI its useless, not only because the player will ignore it but because the AI can't abuse it to keep everyone around him friendly while he wrecks anyone he chooses. AI in fact doesn't take advantage of traits at all in any pseodo-conscious way as far as I can tell. Having superior administrator as a player or superior military also 10x better as a player vs an AI. If there was a game button that would give every nation whether AI or human a top tier trait vs no one having it I would mash that button every time. Don't even care what trait.

I talk about many advantages of humans over AI in this old blog post:
http://axiomsofdominion.blogspot.com/20 ... bbing.html

The AI would probably get more armies than you depending on its size. But a 1-5 region or even 1-4 province AI should never be a threat to you by turn 96. And their armies would always be less experienced and weaker in other ways.

[loki100]

....
I don't understand your auto builder comment. You should always be building something in every single province. I would understand being starved for resources so you would argue that building one at a time means each building spends less time in construction, so you get two buildings after x turns but one you get halfway instead of both finishing at the end, however this should never be relevant. Never been in a position where I wanted to build say, only in 4 of my 7 regions. Maybe if you are really weak I guess and have no infrastructure. But gains are marginal at best.

...

the gains can be substantive when you are developing a poor province, don't ignore this option

[MoLAoS]

it doesn't get production bonuses as such, so no reason why it should generate much larger armies etc.

Its advantages come in at the level of Empire management and your difficulties fall into the same area.

So the barriers in the game to your expansion hit you earlier and the AI has more leeway.

I sort of think this is a lot like the AI in Pride of Nations, at one level its far far better but the key to that game was in effect you were dealing with your own internal challenges and not really with the AI controlled powers?

Edit:

My instinct is that at the level of region management it does a few things badly:

a) it really doesn't sell its slaves so tends to loyalty problems (despite its bonus in this regard) so not only revolts but also baseline production slips back. I find if I take a mature AI region, its a mess in this regard.
b) more generally it tends to ignore loyalty in its pop and building assignment ... I'd like to see the balanced auto system weight this a bit more

This is pretty similar to the Paradox method but Paradox is a lot easier for many reasons. The difficulty gate in Paradox is how fast you conquer and literally nothing else. At least once you hit a decent size. FoGE at least tries to take a step forward in that regard with decadence. However those issues don't really impact the player. Even completely hamstringing yourself by using auto-resolve, a player should always be untouchable at turn 96. Or even passed turn 20-40 really unless they pick some very specific nations. One key issue is the AI can't maintain itself very well. While a player could conceivably conquer the entire map before the time limit, this is absolutely unachievable for the AI. Even at max difficulty the AI is forced to interact with other AIs at max difficulty. I would be very interested actually in seeing AI only games where not all AIs ran at the same difficulty level. Having the AIs pretend like the player is one of them significantly harms their ability to challenge the player even above and beyond their own issues where raising the difficulty hurts them as much as it helps.

[MoLAoS]

....
I don't understand your auto builder comment. You should always be building something in every single province. I would understand being starved for resources so you would argue that building one at a time means each building spends less time in construction, so you get two buildings after x turns but one you get halfway instead of both finishing at the end, however this should never be relevant. Never been in a position where I wanted to build say, only in 4 of my 7 regions. Maybe if you are really weak I guess and have no infrastructure. But gains are marginal at best.

...

the gains can be substantive when you are developing a poor province, don't ignore this option

If you have more than one trashy province you are almost always incentivized to ignore culture and income and minimize food to max out infrastructure pop use. Of course if you get a mediocre draw on buildings for your purposes maybe waiting a bit to mash shuffle is useful but otherwise I've never been at that point. Of course the value of this strategy might go up enough to offset the mental upkeep on very high difficulty levels where you are also purely auto resolving. You have a lot more free time left over for management minmaxing when you don't have to take out 15-75 minutes for FoG2 battles.

I suppose I might do this if I was going for a WC, the same way people like DDRJake in EU4 would micromanage diplomats to gain excess relations. Every 1-5% advantage would matter, then.

[loki100]

actually disagree with both your responses.

the point about not building in all but one region is that all the infra goes there, so you can make progress in developing the region rather than waiting 50+ turns for the simple buildings to complete. Nothing to do with trying a World Conquest (or not)

If you play Empires into the second half you'll quickly realise that it becomes hard to expand for very good in-game reasons. Again nothing to do with how Paradox tries to reflect this (do they in any case as their games are all about world conquests)

[MoLAoS]

actually disagree with both your responses.

the point about not building in all but one region is that all the infra goes there, so you can make progress in developing the region rather than waiting 50+ turns for the simple buildings to complete. Nothing to do with trying a World Conquest (or not)

If you play Empires into the second half you'll quickly realise that it becomes hard to expand for very good in-game reasons. Again nothing to do with how Paradox tries to reflect this (do they in any case as their games are all about world conquests)

Actually world conquest in EU4 is extremely difficult even as a top tier nation. And they do try to stop it but extremely invested players working together always figure out a way to get around the roadblocks. Most people couldn't do it even following a decent WC guide.

As far as getting into the second half, on very high difficulty levels I imagine it would be difficult to do a WC, but are you counting the second half as the last 250 years? Can't imagine not winning by then. The AI can't handle legacy gain well enough for you not to win early that far in.

WC relating to province min-maxing is about time investment. Or maybe I just haven't conquered provinces where it mattered. I'll do the math on finish a building early sometime but the gains aren't that massive except over extremely long time periods in mediocre provinces.

[shockk]

Well that's rather disappointing to hear. Wounder how hard it is to mod in ai buffs

At the start of the game being able to pool production to 1 specific building at time per province is huge (normally into infrastructure or food to take more workers off food gathering). Also important if you claim any un-colonized lands. In the case of wonders it will only take 4ish turns to make instead of 20. Until i get to 300-500 infrastructure per turn per province i rarely building new buildings at every region in that province. (and you never have to worry about over building infrastructure with the 2 free build slots you get, and you can tear them down later). This type of growth is exponential so it really adds up over time

If the games not hard, i can't seem to enjoy it, which would be a shame because i like this game alot so far. Will have to try on impossible as a harder starting nation and hope it gets harder. Though from this new info sounds like it won't and just become a race vs a clock as the AI gains points at higher rates. Instead of having the ai interact with the player

O well guess there always multiplayer

[shockk]

To reply to the comments of saving troops via FOG or the ai losing troops to often. Losing troops doesn't really matter that much. Most of the upkeep are around 1/5 of the unit cost (with the exception of the units with increasing unit costs). So any losses can be bounced back from in 5 turns if no major econ damage was done in that time frame. Your countries econ is about how large of an army you can maintain, damage to the army matter little. Because of how decadence works, spending time to recover your army doesn't matter much as you can't contently expand at the speed of light.

As for human vs ai, yes the human is going to be better, never argued that. Sure small AI nations won't be a threat, but world powers the player never interacted with should still be a threat until they do so. (assuming the player doesn't have like 1/2 the would under their control or something crazy). In my game iv only killed 2 AI, it shouldn't be over at this point

[loki100]

...

If the games not hard, i can't seem to enjoy it, which would be a shame because i like this game alot so far. Will have to try on impossible as a harder starting nation and hope it gets harder. Though from this new info sounds like it won't and just become a race vs a clock as the AI gains points at higher rates. Instead of having the ai interact with the player

..

I'd ignore the comments from MoLAoS, think he has a strange conception of how the game works.

My view is the AI difficulty levels work well as its impact both you and the AI. What it does is to set a natural limit to your expansion and then starts to tighten on this till you tip over into decline. There are a lot of tools in the game to control and manage this process but its what happens.

My current Epirus game is around T180 and I am fairly stuck (in a good way), I need to weaken some opponents and I need to think very carefully about how to do this. Even selling slaves and controlling domestic population, I have a few regions <50 loyalty and starting to make some use of the decadence for loyalty buildings. So I know where this is going to go.

I can prob get into a legacy lead before decline really kicks in and manage the end game chaos from there.

All this is on very hard, going further up the difficulty scale makes this worse, down the scale creates more flexibility for the player.

But practically, I am already fighting most of my wars defensively

[MoLAoS]

To reply to the comments of saving troops via FOG or the ai losing troops to often. Losing troops doesn't really matter that much. Most of the upkeep are around 1/5 of the unit cost (with the exception of the units with increasing unit costs). So any losses can be bounced back from in 5 turns if no major econ damage was done in that time frame. Your countries econ is about how large of an army you can maintain, damage to the army matter little. Because of how decadence works, spending time to recover your army doesn't matter much as you can't contently expand at the speed of light.

As for human vs ai, yes the human is going to be better, never argued that. Sure small AI nations won't be a threat, but world powers the player never interacted with should still be a threat until they do so. (assuming the player doesn't have like 1/2 the would under their control or something crazy). In my game iv only killed 2 AI, it shouldn't be over at this point

FoG2, likely because it's designed to present battles with equal footing for human and AI and with intelligently designed armies is not capable of handling the FoGE battles. Literally you can massacre a dozen medium and heavy infantry with skirmishers in ~17 turns and take 0 casualties if the enemy army has a bad composition such that it fears your infantry too much to move. It will sit and get annihilated without once counter attacking you. Accounting for battles where the FoG2 AI glitches out and lets you take no losses or win 20-40 to 1, vs battles where you win 5-10 to 1 when it functions properly, you can fight the same battle where the auto-resolve would stack wipe you and you might have 1 or 2 infantry move to orange status, if that. The AI is just lucky you are forced to auto resolve sieges due to presumably pathfinding/AI constraints, also Fields Of Glory and not Slums Of Glory after all. Now, AI brainfarts aside, at the top 2 difficulty levels you might be hampered a bit but you have to remember something about the difficulty levels and AI, as noted below.

The game is specifically designed to treat AI and players equally. Consequently, because the AI is inferior to humans in many ways, large AI nations run into the same problems humans do and as far as I know, they are incapable of planning for them. As a human you can optimize territory gain province wise to have minimum decadence and maximum resource pooling, attack whoever you want whenever you need to, manage your ascension through the government ranks so you don't sabotage yourself, and do many other things. The AI is not capable of this. The AI will pick up progress tokens in a totally naive way which will cause large shifts in its position that it can't deal with as far as decadence and so forth.

The game is over after 50 turns. There's just no way around it. You may not be in possession of a de jure victory yet but you've de facto triumphed. The economic system is exponential so you conquer what you like to be the baddest military power on the block and then you min-max culture. Even in the case where you don't pick an easy nation like Rome, which roflstomps at all difficulties if you don't autoresolve and is still amazing if you do, after the first 50 or so turns you still know if you've won or not. Taking a weak nation in a difficult spot geographically is effective until you become a regional power, the same as in EU4 where people do The Three Mountains, or in Total War. The AI is never going to rack up 5000 Legacy points in 100 turns. But as the player you can do that through min-maxing culture output.

If you want to have a challenge you really have to set up a scenario. Because the game treats the AI and players equally, AI gain nothing from upping the difficulty. Consider the Successor states. If you up the difficulty for the Seleucids you are also empowering Maurya, Egypt, and Anatolia as well. So the AI nation now faces enemies equally boosted. Sure you as the player get penalties. But that is ameliorated by the AI conflicts.

[MoLAoS]

Well that's rather disappointing to hear. Wounder how hard it is to mod in ai buffs

At the start of the game being able to pool production to 1 specific building at time per province is huge (normally into infrastructure or food to take more workers off food gathering). Also important if you claim any un-colonized lands. In the case of wonders it will only take 4ish turns to make instead of 20. Until i get to 300-500 infrastructure per turn per province i rarely building new buildings at every region in that province. (and you never have to worry about over building infrastructure with the 2 free build slots you get, and you can tear them down later). This type of growth is exponential so it really adds up over time

If the games not hard, i can't seem to enjoy it, which would be a shame because i like this game alot so far. Will have to try on impossible as a harder starting nation and hope it gets harder. Though from this new info sounds like it won't and just become a race vs a clock as the AI gains points at higher rates. Instead of having the ai interact with the player

O well guess there always multiplayer

Probably the easiest thing you can do for difficulty is mod the game. Give the heavyweight nations more buffs. Modify their objectives so they aren't in conflict, etc. Basically buffs that benefit specific nations rather than buffing the AI as a whole.

[shockk]

"The game is specifically designed to treat AI and players equally"

Well that just bad game design. Also if it was true there would only be one difficulty.

"The AI is never going to rack up 5000 Legacy points in 100 turns. But as the player you can do that through min-maxing culture output."

Turn 122 in my game and the ai has 8.5k Legacy points. Not sure why your bring up TW, its "campaign" system has always been atrocious. Yes starting as a tiny nation in eu4 is fun and challenging (bit less nowdays with so much mechanics allowing way to much min/maxing). However there was always a local power for you to overcome that would be a challenge. There needs to be something to fight mid/end game, which id expected to rome, but the ai needs buffs.

"If you want to have a challenge you really have to set up a scenario. Because the game treats the AI and players equally, AI gain nothing from upping the difficulty. Consider the Successor states. If you up the difficulty for the Seleucids you are also empowering Maurya, Egypt, and Anatolia as well. So the AI nation now faces enemies equally boosted. Sure you as the player get penalties. But that is ameliorated by the AI conflicts."

Disagree, a general buff to the ai would work just fine. The ai being at war most of the game doesn't matter that much, as it only take 5 turns to recover their army. The ai isn't going to have their econ ruined from constant war. The earlier point i was trying to make is taking loses doesn't impact much unless it means losing the war. ( So saving a troop here or there with Fog2 has little impact on the game, sure if you abuse it ya things will break. Player shouldn't be using FoG2 to be getting an edge in FoG:E, but only if they feel like playing a game of FoG2.) Now if the ai is in a bunch of wars at the same time as the player, sure its going to make war easier. However at higher difficulties it seems like every ai by the player declares war on them so its not like the player lacks other enemies at the same time

[loki100]

...
The game is specifically designed to treat AI and players equally. ...

Everyone is entitled to their opinions and views of how the game works. Some have the wit to listen to others when they suggest your views are wrong.

but this statement isn't just a flawed opinion, its so wrong as to indicate you have little or no understanding of the game.

[MoLAoS]

...
The game is specifically designed to treat AI and players equally. ...

Everyone is entitled to their opinions and views of how the game works. Some have the wit to listen to others when they suggest your views are wrong.

but this statement isn't just a flawed opinion, its so wrong as to indicate you have little or no understanding of the game.

Based directly on comment by the devs. Unless you are misunderstanding what I'm saying. Difficulty levels do impact the player but the AI itself isn't engaged in some grand conspiracy against you, unless progressive difficulty is on. Although that setting is a bit confusing since it doesn't explain what the game considers to be "winning".

[panzeh]

"The game is specifically designed to treat AI and players equally"

Well that just bad game design. Also if it was true there would only be one difficulty.

"The AI is never going to rack up 5000 Legacy points in 100 turns. But as the player you can do that through min-maxing culture output."

Turn 122 in my game and the ai has 8.5k Legacy points. Not sure why your bring up TW, its "campaign" system has always been atrocious. Yes starting as a tiny nation in eu4 is fun and challenging (bit less nowdays with so much mechanics allowing way to much min/maxing). However there was always a local power for you to overcome that would be a challenge. There needs to be something to fight mid/end game, which id expected to rome, but the ai needs buffs.

"If you want to have a challenge you really have to set up a scenario. Because the game treats the AI and players equally, AI gain nothing from upping the difficulty. Consider the Successor states. If you up the difficulty for the Seleucids you are also empowering Maurya, Egypt, and Anatolia as well. So the AI nation now faces enemies equally boosted. Sure you as the player get penalties. But that is ameliorated by the AI conflicts."

Disagree, a general buff to the ai would work just fine. The ai being at war most of the game doesn't matter that much, as it only take 5 turns to recover their army. The ai isn't going to have their econ ruined from constant war. The earlier point i was trying to make is taking loses doesn't impact much unless it means losing the war. ( So saving a troop here or there with Fog2 has little impact on the game, sure if you abuse it ya things will break. Player shouldn't be using FoG2 to be getting an edge in FoG:E, but only if they feel like playing a game of FoG2.) Now if the ai is in a bunch of wars at the same time as the player, sure its going to make war easier. However at higher difficulties it seems like every ai by the player declares war on them so its not like the player lacks other enemies at the same time

The only thing that's a lot harder about the harder difficulties is staying high in CDR because i'm pretty sure they get less decadence and since it's a relative measure that matters. Otherwise, fighting the wars isn't that hard, especially if you can get a solid province to build units in.

Also, yeah, you don't need FOG2 to beat the AI at wars, just kinda a willingness to spend your money.

[loki100]

...
The game is specifically designed to treat AI and players equally. ...

Everyone is entitled to their opinions and views of how the game works. Some have the wit to listen to others when they suggest your views are wrong.

but this statement isn't just a flawed opinion, its so wrong as to indicate you have little or no understanding of the game.

Based directly on comment by the devs. Unless you are misunderstanding what I'm saying. Difficulty levels do impact the player but the AI itself isn't engaged in some grand conspiracy against you, unless progressive difficulty is on. Although that setting is a bit confusing since it doesn't explain what the game considers to be "winning".

forgive me for asking but have you read section 4.1 of the manual?