That one Runaway AI ...

[arthurF]

So in the past couple months i have started multiple games with the following settings: immortal, large, ancient, abundant lux(where possible), no tech brokering, no events, conquest/tech victory only, earth or continents, no vassals.
In each game, i have played to at least around 1400AD, and in every single one, whether the early game had been good for me (where i was winning compared to nearby rivals) or not, I would eventually encounter one civ that had used the time better than all of us. This civ, the runaway AI ( seems independent of exactly what civ, i have seen mongols, arabs, poland, etc in this position), is suddenly somehow 9 techs higher than everyone. "Uh oh", i say to my self, and i imagine the other AI's mutter algorithmically to themselves, and try to speed up my own progress.
But how to speed up? I thought i was doing everything right? I had staffed libraries, I didn't cut down forests or jungles in preparation for universities, i even went after wonders that had effects on science, and tracked down any luxs that would help too. I tried hard to keep my happiness near 10, i took progress, and maybe authority (for science kills and happiness and the early settler). I tried to keep my net gold intake positive.
Meanwhile the Runaway AI has jumped from a lead of 9 to a lead of 13. WHAT?? Much gnashing of teeth and attempts to stall by way of alliance or UN sanctions.
The Runaway AI now has a lead of 20 and has turned hostile. Game over.
Seriously this happens every time.
So my question is this: why? how?
Also, how to deal with it

 

[PapaRockett]

Yup. In my opinion this ruins the fun. I've met civs who are so far ahead in science that it's practically impossible to keep up. They get to atomic era when the rest of us are still in late industrial/early modern. They have tanks and nukes while we're still using lancers. I really wish it didn't happen and I'd also love to know how to deal with it. World Congress resolutions don't seem to be helping that much, spies always get killed or take 200 turns to do anything because that civ has the correct infrastructure to prevent spies.

 

[Txurce]

This is already being discussed in a thread started about an hour before your. You may also want to check in there.


https://forums.civfanatics.com/thre...ng-in-industrial-modern.626222/#post-14968275

 

[LukaSlovenia29]

Txurce, he said this has been happening for months. So I think his experience goes beyond just the introduction the of new AI bonuses etc.

Arthur, I'd recommend trying to play games with older versions (before the new AI bonus system) and on Pangea/Oval, with advanced start selecting which civs can or cannot be in your game.

Pangea/Oval -> playing on continents it's much easier for the AI to steamroll because it's harder for the human player/other AIs to curb that AI by military action etc., and it's also easier for that one AI to swallow up other AIs on that continent.

Manual selection of civs -> If you put/allow Venice in the game with the Huns and Aztecs, chances are the Huns/Aztecs will 1. settle into the territory around Venice and 2. Conquer Venice before it can leverage its merchants. It's similar with having e.g. Korea in the game with Huns and Aztecs -> chances are Huns and Aztecs will severely limit Korea before Korea's bonuses start firing up. In my experience, extreme warmongers with early uniques don't mesh well with peaceful civs without early UUs. So I usually select all the civs to be more or less of equal nature (1. all warmongers; 2. all civs with peaceful bonuses; 3. all civs that are kinda in between).

Since I started doing both of these things, the rate of games with a huge AI runaway has decreased a lot. Hope it helps.

 

[LukaSlovenia29]

P.S.: Just saw that you also play on "Large". I think playing on large vs. playing on standard increases a chance of one AI getting a big early advantage (in settling/war over another AI) and then leveraging that to steamroll. So my advice would be to play on a standard size map.

 

[Infixo]

I think we should however remove or fix the root cause of the problem. Limiting ourselves to specific options is not the way to go. Can’t play large, can’t play continents (btw, the most standard map ever), other thread - can’t play epic or marathon (re: xp discussion), etc. Again - not the way to go.

 

[LukaSlovenia29]

Well, I think historically speaking vastly different levels of technological/military development between places far apart are the norm and not the exception -> see indigenous peoples in Americas and Australia before the arrival of the Europeans. That's why I'd be ok with large&continental maps were a bit less balanced than standard/Pangea (being a good representative of Europe as a continent, where everyone approximately the same in technology/military level throughout the history).

And I think the most balanced the non-standard games (speed, size,...) will be is when the VP will be in the final stages of its development, when the standard game parameters will be set "in stone" by Gazebo and the community will be able to focus on non-standard

Lastly, I'm suggesting this only as a temporary "solution" until the new parameters are figured out.

 

[Stalker0]

There are really 2 ways to go about this:

1) War: Take them out before they get too strong.

2) Diplomacy: Use the power of the World Congress to take them down a peg.

Regardless of which way you choose, first....you should focus your spies and trade routes on the super civ. These will help you catch up on science.

War is war. Diplomacy wise, the key are the following policies:

1) Everyone gets a bonus on a tech for each civ that has researched it. Again will reduce the gap a bit.
2) The policy that gives science/food bonuses to people lower in score (aka you!). Again will help reduce.
3) Work on getting yourself circles of influence with CS to build a greater diplomatic coalition.
4) Put the nerf on spaceship parts if that is is the victory he seems most keen on.

Its suprising how much these can do. I have come back from 13 tech deficits to tech parity with these techniques. Not every time, but several times.

 

[Deadstarre]

earth or continents... I would eventually encounter one civ that had used the time better than all of us

I love continents too, but what I don't like is having them separated by deep ocean, necessitating that you play until 1400ad before discovering theres a big problem going on. the sooner you met them and saw it happening, the more likely it is you could have done something about it.

my solution has been playing on a modified continents++ map with the water level low enough to always guarantee there is a shallow water channel that connects the continents on both ends of them. it keeps scouting relevant and makes the embarking technology a real consideration early, as i always know there is a path to find my way across the continents and meet everybody.


I edited the water with standard map size specifically in mind so it may not be any use to you but ill upload it here.

note: leave the water level on standard when starting a game (and i usually pick 5 billion years for earth age, i dont like so many mountains)

 

[CrazyG]

I think we should lower the science on public schools. Most AI enter that era through this tech, they are so strong that whoever gets them first will snowball ahead ( I snowball when I do this)

AI bonuses for great people or archeologists might play a part too, but this has been debated and I don't really understand what the code really does.

A spy taking 45 turns to steal a technology and then getting killed is infuriating. I don't think spies are doing their job as a comeback mechanic, like at all, come industrial I'm surprised if I manage to steal one tech the rest of the game.

 

[Deadstarre]

A spy taking 45 turns to steal a technology and then getting killed is infuriating. I don't think spies are doing their job

it's just really hit or miss. when he doesnt get killed and he levels up, if he stays in a lucrative city i've often seen my guy continue to come back in very reasonable time frames with new tech (10-15 turns even later on) but if he got killed just once, thats the end of free techs in the game for me.

 

[Stalker0]

I
A spy taking 45 turns to steal a technology and then getting killed is infuriating. I don't think spies are doing their job as a comeback mechanic, like at all, come industrial I'm surprised if I manage to steal one tech the rest of the game.

This is a King perspective, but usually my spies work pretty well at getting me caught up. I don't generally see those kind of turn numbers until I've mostly caught up, or I'm simply targetting the wrong civ.

 

[arthurF]

Anyone know HOW the AI does this, though? If we knew how ( and why it is only 1 of them and not all of them), then I should be able to do what they do right? The runaway AI usually had really big cities as well. Populations that are 150% to 300% bigger than anyone else, and still is apparently happy somehow.
How do they manage that? I am often put in the position of "locking" my smaller cities while they build buildings, so my happiness doesn't fall through the floor.

 

[CrazyG]

This is a King perspective, but usually my spies work pretty well at getting me caught up. I don't generally see those kind of turn numbers until I've mostly caught up, or I'm simply targetting the wrong civ.

I usually can only steal from one civ, the runaway, and his capital and most cities usually have these times. It takes like 8 turns to travel and set up surveillance so I usually don't bother trying to check the other cities, maybe that is a mistake?

 

[Stalker0]

I usually can only steal from one civ, the runaway, and his capital and most cities usually have these times. It takes like 8 turns to travel and set up surveillance so I usually don't bother trying to check the other cities, maybe that is a mistake?

Is this on Standard? I don't usually see those kind of times at all. Much shorter for me.

 

[CrazyG]

Is this on Standard? I don't usually see those kind of times at all. Much shorter for me.

Loaded a recent game, its 38 turns in industrial era on standard. Its 40+ after you factor the time to move your spy there

 

[Infixo]

Anyone know HOW the AI does this, though? If we knew how ( and why it is only 1 of them and not all of them), then I should be able to do what they do right? The runaway AI usually had really big cities as well. Populations that are 150% to 300% bigger than anyone else, and still is apparently happy somehow.
How do they manage that?

1. Human probably would not be able to follow the same strategy simply because we dont get same bonuses as AI.
2. Good questions here. Why 1? I am wondering, could be difficulty related. On king/emperor I saw 1 but sometimes 2 were quite strong. Maybe on higher diff one of those 2 would eventually snowball more.
3. Characteristics. Let’s get more real game examples, to compare. In my last game I had Pachacuti, who had the most cities and biggest population. At ~T300 his cities were on average bigger than 2nd AI and mine, by 5-7 pop, like 25%. He was leader in policies, adopted Ideology, ahead of me by 4 or 5? policies, and I was 2nd. Conquered 2 other civs that were on his continent. Huuuge army, really. I even checked - almost 140 units. And stacked like 20 wonders? Long list, anyway. Also, what he did NOT do super-good: city states. Few allies, nothing spectacular.

 

[PapaRockett]

1. Human probably would not be able to follow the same strategy simply because we dont get same bonuses as AI.
2. Good questions here. Why 1? I am wondering, could be difficulty related. On king/emperor I saw 1 but sometimes 2 were quite strong. Maybe on higher diff one of those 2 would eventually snowball more.
3. Characteristics. Let’s get more real game examples, to compare. In my last game I had Pachacuti, who had the most cities and biggest population. At ~T300 his cities were on average bigger than 2nd AI and mine, by 5-7 pop, like 25%. He was leader in policies, adopted Ideology, ahead of me by 4 or 5? policies, and I was 2nd. Conquered 2 other civs that were on his continent. Huuuge army, really. I even checked - almost 140 units. And stacked like 20 wonders? Long list, anyway. Also, what he did NOT do super-good: city states. Few allies, nothing spectacular.

Just out of curiosity, how did that game end, did Pachacuti win?

Also, one observation I have is that a civ who adopts the "Co-operation" follower belief is most likely the runaway

 

[tu_79]

Why does it happen? There's a snowballing mechanics built upon another snowballing mechanic.

Current handicap bonuses are awarded to AI civs that accomplish something, like increasing cities, getting to new eras, and birthing great people. A civ that get more cities is, in a way, taking a lead at controling territory and resources, and also get more bonuses for free. A civ that get soon to a new era has already a tech lead, that can be used for getting stronger stuff, but also get more bonuses for free. A civ that gets more great people than the others already has an advantage in terms of historical events and their bulbing effects, but also get more bonuses for free.

Why in industrial? Probably CrazyG is right and the culprit is Public School. But there is other explanation. AI has already a bigger population, and there are several things happening at industrial age concerning population: Farms give more food, there's agribussiness, there are several buildings with a bonus on population (Public School, Factory, Museum).

 

[Infixo]

Just out of curiosity, how did that game end, did Pachacuti win?

I stopped playing it, not sure if I want to continue.