Controlling AI Underexpansion (Rant and Question)

[TheMeInTeam]

In my last 10 games or so, I've lost three times. In one game, my first attempt at PC nappy, an AI vassaled to a war ally who was stronger than me despite me also having sufficient power and capturing 5 cities to that AI's 0 captured. However, as crappy as this mechanic is, my other two losses bother me more:

On a 7 civ map, one AI gets 25-33% of the land without a single war.

Yes, a runaway AI is the leading cause of my losses on immortal (and espectially deity), but here's the thing: can it be controlled?

After a debacle on terra in my LP and a solid showing in BOTM 29, I tried an offline terra. One of the problems with this mapscript is that it gives extra land for bonus AIs to grab, but I figured I could handle it.

Then I met sury, izzy, pericles, and several others. Around the 1-200 AD range I look at the trade screen for sury.

15 cities. What the hell? I expanded in his direction!

I then look at pericles' city count. THREE. Three effing cities for an immortal AI in the ADs. No wars yet. Pericles was not on some tiny peninsula that blocked expansion. For @#$% sake, sury didn't even block him. I met them both by turn 10, and one AI simply chose to hand the other 3/4 of its wide-open land!

By the time sury declared on pericles and his 3 city hell, he had over 30% of the world's land already (I can only expand so much, since sury/izzy/pericles all had different religions and i was between sury and izzy, so had to mind military and maintenance but still got >6 cities with some more blocked)

That's where I winced. When there are 7 civs on a map, each one should ON AVERAGE be ~14.29% in land once peaceful expansion is done, unless somebody gambit rushes or chokes. On higher difficulties, some of that land falls short for the player since it's hard to keep up with AI expansion.

But 32% land for one AI without war is turd garbage for balance. That is nearly double the land an average AI should be getting if the player does nothing and it were to be split SIX ways !!!!!!!!!!!!!!.

This can only get compounded, when 1-2 AIs peacevassal (aka de facto permanent ally) to that sucker and his supposed pop/land is nearly 40%, all without ever having military leave borders. This is easily the #1 cause of losses for me, way more typical game over than surprise DoW or something more legit.

So my question to the forum: Is there anything that can be done by the human, short of early DoW (which isn't going to work if you're not there), that will either 1) slow down the fast expanders so that the AI distribution is more balanced or 2) speed up the underexpanders so that one civ isn't given "super AI" status on a silver platter? I noticed @#$%ttycles founded early religion, and underexpanding AIs seem to have a pattern of early religion...but that's not exactly something that one can consistently deny on immortal+! Is there anything else that can be done one way or the other?

I guess my point is that this system of severely imbalanced land gameplay skews game difficulty badly, probably it can go +2 or -2 difficulty levels at the extreme ends of this garbage. If there's an actual in-game way to curb the AI from tag-teaming and losing on purpose so another AI is stronger, I'd love to hear it.

Otherwise, I thumb my nose at whatever supposed playtesting/start code script was done for civ IV when balancing map size/tiles vs maintenance. If building in more-than-double-land starts was intended, whoever is responsible is contending for the worst possible balance decision ever made in civ IV. Looking back at my L's, I would happily enable EVENTS again if I could be guaranteed that this were corrected, it's that bad .

And up front, I'd like to point out a few things:

- This problem manifests itself in far more extreme fashion on higher difficulties, where the AI can afford its triple city count. I'm not interested in advice for lower difficulties where early war is consistently cost effective and the player can actually ICS if he knows what he's doing. It's nothing personal, just that the reality of the situation changes as you get higher in level, as it becomes increasingly impossible for the human to match AI expansion rates on big maps, while the AI suffers for the same expansion less. I'm looking for at least emp+ advice on AI manipulation here.

- This is assuming some non-standard settings, where standard ones were apparently the only thing tested. I know how to beat fractal, pangaea, archipelago, and continents maps on standard size. Although the AI occasionally still gets >25% land without lifting a finger on those scripts , in terms of city count that is more like 3-4 extra cities rather than 10-15 extra cities. I want to know how to beat immortal or deity AIs with 30-40 cities total +, not 15-20 cities max.

- Add a monkey wrench for plains-filled starts/maps. Can you REALLY pull rush/early war when the nearest AI is 20 tiles away but you're covered in plains? I'd be delighted to hear it.

- I would be delighted to see a deity or even immortal walkthrough where a) the player meets someone with >33% land PRE-vassal on contact (without war) and b) the player manages to win via something other than culture or diplo.

- But hell, I would be happy just to see some mechanic manipulation tricks. Crap like "start SH so AI delays it and doesn't cut it's expansion" or "selectively open borders to change expansion rate". ANYTHING.

- Or just tell me how to get an equal ratio of cities on immortal+ on huge 11 civs as one might get on standard 7 civs .

 

[stopstopp]

Doesn't pericles not expand fast? I don't have BTS so I'm not completely sure but I've seen a few games and it doesn't seem like pericles gets very big.

 

[Kadazzle]

Choose different leaders? Elizabeth, Pericles, Willem and leaders like them very often don't get passed 6 cities in my Immortal/Deity games, while REXers like Charlie, Cathy and Survy nab a whole bunch of cities. In my opinion this is what make Immortal/Deity, and this isn't much you can really do, except buckle down and tech like a champ, or DOW early. You could run around with spies for a little bit, but it won't stop them from expanding.

 

[Zexx]

I've never noticed an expansion rate difference depending on leaders. They all tend to prioritize something ingame, and go after it. Sometimes it's REX, sometimes it's not.

 

[TheMeInTeam]

Choose different leaders? Elizabeth, Pericles, Willem and leaders like them very often don't get passed 6 cities in my Immortal/Deity games, while REXers like Charlie, Cathy and Survy nab a whole bunch of cities. In my opinion this is what make Immortal/Deity, and this isn't much you can really do, except buckle down and tech like a champ, or DOW early. You could run around with spies for a little bit, but it won't stop them from expanding.

Tech like a champ? Even 1400-1500 AD tanks would struggle against 30+ deity cities...the AI will often have infantry/arty by then! While tanks are better than that, they DON'T flank it, and massed stacks of >80 units tend to rape you silly unless you get nukes...remember that at least half of those cities on deity are unit/turn factories. On immortal, the AI is "only" getting about 10-15 units/turn there .

There IS no "DoW" early on long-distance spawn deity. Probably not even on immortal if we're talking 15+ tiles.

Maybe the simple conclusion is that some difficulties were designed such that on larger/less usual maps they're not intended to be consistently winnable, but I'd like to avoid that conclusion if possible.

I point all this out specifically because I see so few high-level games outside of controlled environments. That screw-job I got in my let's play was really painful...but probably winnable. It's the ridiculous start deviation I'm trying to curb, it seems you don't have an answer. Hopefully someone has one.

 

[shyuhe]

Maybe the simple conclusion is that some difficulties were designed such that on larger/less usual maps they're not intended to be consistently winnable, but I'd like to avoid that conclusion if possible.

Huge maps aren't very friendly at higher levels because the AI can expand so much more than the human. Then again, if you drop to marathon, you can just continue drilling the AI while you whip courthouses and continue on your military march. The maintenance costs do scale to map size but I don't think it balances out very well at the higher levels - hence I rarely play those settings. The only other suggestion I can think of is to gift cities to the AI to tilt their expansion tendencies, although even that may not engineer the game in the right direction for you, not to mention the high hammer cost.

 

[timcrewgreen]

I don't think there is anything you can do. I say quit and start a new game, and hope that this is fixed in Civ 5. This game's pretty much been "figured out" by the community now and there isn't much room for original strategy any more.

 

[DaveMcW]

- This is assuming some non-standard settings, where standard ones were apparently the only thing tested.

That basically sums it up. Non-standard maps are easier to break, and not always in the human's favor.

 

[amanbe]

My feeling is that the game is not well polished/balanced for higher levels, hopefully this won't be the case in Civ5.

 

[Metal Bunny]

I just started immortal, and I'm only used to playing huge maps on marathon, but what I have noticed is that the civ flavors, much like Catherine and Sury, will always expand as much as they can, much like Shaka will at one point go bat**** crazy and DOW even his closest friends.

What I have seen is that, if you want to play within mechanics, is that you simply artificially choke the AI. You can't do this in-game, except if you scout early and have a really lucky map lay-out.
What is possible is pre-game manipulation. Archipelago, hemispheres, big and small and medium and small (my 2 favorites), as well as any kind of map that has a lot of islands, tend to work great.
Simply because even though they make a lot of cities, the cities themselves are crap. AI is hopeless in re-directing troops from one island to another and basically just floats its trireme stack of useless near its capitol. The cities they so relentlessly settle will be able to contribute a lot less to the main capitol or to the 'war effort'.

Not only that, AI's on map styles such as medium and small and archipelago are kind of easy to choke. Well, at least on emperor and then only 2 AI max. I haven't tried it yet on immortal. Then again, being able to choke 1 AI, if you pick a good one, such as Cathy or Sury, would be helpful enough to at least get some normal semblance of 'balance'.

But these are all controlled environments, the only thing I can think of is getting 2 REX lovers to DoW on each other.
Or, to play pangaea with any kind of rush unit, or even axemen/chariots, and to ruthlessly expand and use all the money to get to currency, while avoiding investing hammers into settlers.
Granted, I only did this once, and it was with a financial leader and I captured the mids.

But, I just started immortal and I may not be experienced enough to actually be able to contribute a lot here

 

[raxid]

Frankly I don't see the problem here at all, lets take your gripes one at a time. If you don't want to read it all TMIT it all boils down to the last sentence really.

1) "an AI vassaled to a war ally who was stronger than me ..." So the only reason your upset is because you know how the AI is "supposed" to work (with checks on war success and what not)? The right choice (however the AI were to get to it) was to vassal to the more powerful (and thus better able to protect them) civ, this wasn't you in this case.

2) Average civ gets 15% and sury got 32% AND started next to you. Pericles (stupidly) went for early religions and thus can't support as fast an expansion as Sury. Not to mention that the early religions gave him new priorities/options. So of course he underexpanded and it had a snowball effect. According to your numbers here, Pericles still had 10% land (or at least his death allowed sury to get that much more). So Sury got an extra 5-10 (remember Pericles was only "entitled" to 5% more than he had, but he too was blessed with lots of land from the map generator, he just didn't take advantage) from Pericles's underexpansion and probably close to 5 from you (humans can't expand as fast) so your looking at 25-30% right there. So Sury got 3 times lucky in the stat, lots of room, and 2 separate neighbors which are easy to expand against. I really dont' see a problem, one AI played poorly and the other got some luck and played well and took advantage of the situation, this is exactly what you and I do when we play.

The only way to "fix" that problem is making terribly predictable maps that are more 'fair' or making it so that one can only take advantage of another's mistake so much. The latter is accomplished generally by making the expected % of land for one person to be a lower number.

3) Peace-vassaling. You seem to have a big problem with this, but honestly I've never understood why. Except that it makes the game more difficult, but if that were a reason to dislike something/want it out of the game then there would hardly be any features in the game. It makes perfect sense from both sides why they'd peace vassal (closer to dom limits/free resources/other vassal benefits and the vassal obviously gets protected so it can go for a peaceful victory type). I mean, when is the last time you rejected a peace vassal? And if you did b/c they were going to win if you did (pretty rare case for a human to be in) just 'imagine' the AI turns down other AIs asking them to peace vassal all the time.


Basically I'm saying you are getting upset at the game because of code diving and assuming the game should work one way or another (at least for the two vassaling issues).

 

[Jaybone]

what would the point of playing be if you always win?

 

[amanbe]

A chance to win is the discussion here I believe...

 

[Jaybone]

A chance to win is the discussion here I believe...

but you give excellent players like TMIT a chance to win, meaning niot a major unbalancing, and they will win pretty much every time.

Somedays you eat the bear, somedays an aryan vedic invasion crushes you

 

[Negator_UK]

I'm only a struggling emperor, so my obervations will be less relevant than imm/dietys, but...

1. Part of the joy of civ is the randomness, and the discovery that goes with it. - Once in a while you will meet that big civ, or religious love-in on the other side of the world too late to win the game, it goes with the territory - play more predictable map scripts to avoid this I would say.

2. For vassaling, please confirm that you had no chance to vassal the AI yourself - sometimes I have noticed a AI vassal to another when I was trying to annihilate them - you can't blame him for vassaling under these circumstances.

3. Civs UI has always been pants as I have ranted elsewhere - but maybe its time to make some request to the BUG guys to update their product to permit finer obervation of the vassal-to-me / vassal-to-him question like they did with WHEOOHRN. Not that this will help those of us that don't use mods...

 

[vicawoo]

Back when I could still play, my computer could barely handle standard maps.

So I'm going to conjecture without any proof that "relaxing" immortal Montezuma game someone played awhile ago, abusing diplomacy to use all the AI's as their attack dogs.

Rushing someone 20 tiles away is incredibly inefficient, if you don't raze it. Each city is going to cost over -7 to your empire with standard size maintenance. Plains heavy starts, try wonders, especially ones that scale.

 

[Semmel]

I dont think that one AI "donated" its land to the other just to piss you off. Well, it pissed you off but just because one AI played week. I had a similar situationin my last normal archipelago map on emporer (random events, goody huts and vasseling off). Charly and Churchill were next to me. The map was nice so I was able to reach all islands in nutshells. The thing is: Charlemagne only got 4 or 5 cities with plenty of space left to expand, but he didnt. No other civ claimed the islands either because there just wasnt any in this direction. I dont know why, but maybe because he didnt had any copper or iron. That was the only thing that cought my eye when I was invading him.

 

[TheMeInTeam]

Basically I'm saying you are getting upset at the game because of code diving and assuming the game should work one way or another (at least for the two vassaling issues).

I did read it all, but probably people don't want to read it twice so I'm just quoting this.

Remember, if the human were to delete his initial settler and just watch (assuming that didn't lose of course), AIs would average ~16 or 17% land. Granted, the gap created would give someone extra room, but remember that I expanded IN HIS DIRECTION, not away from him, noting that he is a generally solid AI and more likely to be a long-term issue than izzy. His access to >30% land came solely from the asp-hat other AIs giving it to him w/o fighting over it.

I don't have an issue with peace vassals when it actually benefits both sides. However, your assertion that it does so is not consistently accurate. My #1 problem with peace vassal is when a powerful AI deliberately protects a culture or space whore, even feeding tech to the former. This is not a mutual relationship, it's a parasitic one, with the master effectively actively trying to lose and take the human with it. I would have far less issue with this mechanic if it weren't so damned common due to land imbalance.

Speaking of that, this case with sury is not the norm. Usually, the AI that gets 30% land without war is *not* near me (they are often on the other continent making it harder for me to stop early), so you can throw out the component of your argument that a human can't keep up with an AI's expansion (I actually can below deity, and if the diplo situation is secure enough that I can skirt a DoW, will hit double digit cities BC on occasion to deny land if it is good).

But my #1 issue with the post is that you only addressed the rants, not the question . This is an unfortunate pattern in the responses so far. It seems like if I don't want to pre-select opponents or avoid some map scripts entirely, the only other option suggested is to resort to marathon-based military chokes or attacks. At least so far that's what I'm hearing.

 

[budweiser]

Its not a perfect system. Its a very complicated game, hard to balance eveything. You have to remember the AIs are playing according to their script, not the same game as you at all. They dont think, they follow a routine.

Don't expect perfect balance every map script. The standard normal map is good, but has its limitations as well, there is only so much variation in the system. I mostly play continents because I think archpelago is cheese and pangea is too easy. But, I still notice that there really are only a few variations on the standard continents script; isolation, duel, 2,3,4 AIS on your continent, etc. It can get repetetive. If I end up with a scenario that i don't feel like playing out because of time or effort, I just re-roll.

I think the best way to play is large continets epic (second slowest speed), but I dont play that way because of my computer. I think this map should aleviate some of the travails of handling the big guy on the other land mass.

I find the vassal thing to be annoying sometimes. It happened to me the other day. Sal invaded SB from across the sea, so you know it was lame. then I invaded SB who was on my continent and took 6 cities (half his empire). Sure enough, some time later SB vassalized to Sal. I'm not sure the rational, but I think it is supposed to simulate them looking to the master for protection. I didnt look, but I am 99% sure that the turn before he capped to Sal, he would have also capped to me, but I was going for the knock out. The thing is, its a historical simulation and you just dont see that many instances of 1 civ wiping out another civ completely in modern history. Capitulation has to be a fail safe built in or just sub-optimal human strategy that triggers it.

 

[Metal Bunny]

Well, I'm all for marathon, part of my childhood

But, I think that the AI simply prioritizes vassals in a completely different way than we do, as well as the status of friendly.

I realize that some peaceful vassals are nothing but parasites, especially when mansa musa voluntarily vassals to any sort of warmonger, but I believe the AI only sees the immediate leap in power, score and land, basically seeing it as some free gift of power.

The AI then has a higher chance of becoming friendly with each other, not only because the AI only peace vassals to people they at least like (at least, that has been my experience so far), and thanks to the defence pacts, they'll get more brownie points and maybe jump to friendly.

When the AI is friendly, s/he will vote for you for a diplo victory, thereby actually conceding the game to you, prioritizing the friendly relationship above winning the game.

What I see is that, especially in standard games, because you have less civilizations (thereby increasing the random factor; if you have more civs, the chances increase that 'normalization' occurs) you're going to actually have an increased chance of getting a runaway AI.

I have almost never had this sort of situation in my games, in about 50 of them, this happened to me about 2 times.

I think it's just part of the coding, that the AI simply will behave in this way.
Other than playing a pre-game controlled environment, or to play a relentless genghis khan game, I can see no other solution, at all.

So yeah, if someone has an answer, that'd be sweet, lol.