[R&F] How do you play if you want the AI to have a fair chance?

[Trav'ling Canuck]

After many months of not playing Civ 6 to try and win, I decided to play a game with the intent of winning, but with self-imposed rules to try and make the game more fair to the AI.

Much of the talk on this forum about why the AI is so easy to beat in Civ 6 seems to centre around the following three items:

• War is too easy, and thus the human player gets a disproportionate benefit to investing in military units.
• Chopping is too powerful because of "exploits" like overflow.
• The AI's bonuses are mostly front-loaded, so once you catch up, Deity isn't any different than playing Prince.

So here's the set of rules I imposed on myself to try and take these out of the equation:

• No conquering AI cities or city states.
• No interfering with the AI by using Spies against it or capturing vulnerable Settlers/Builders etc.
  
• No using chopping overflow techniques.
• No using Magnus (not for his chop bonus, or his almost as good second ability of population free Settlers).
• For good measure, no using Public Transport policy either, as that card is so OP it's stunning.
• I didn't have a good way to change the AIs' bonuses, so I gave them an even bigger head start by forcing myself to:
  • Build a Holy Site as my first district
  • Delay my own expansion while I run projects until I get a Religion

On a standard speed, standard size continents game under these rules I won a science victory on T274 (rolled Trajan, with Seondeok, Genghis, Qin, Jayavarman, John Curtin, Frederick, and Cyrus as opponents). That's still 25 to 50 turns before the AI would be able to win (no other civs had finished the Moon landing when I won or completed any of the third stage research techs).

I'd estimate that, other than the rules related to not taking AI cities, the other rule changes (chopping nerf and delayed expansion) cost me about 25 to 50 turns combined. In the absence of these, I'd normally win between T225 to T250. I can't say what impact the not conquering had on my victory time because I honestly can't remember the last time I took an AI city.

I'm curious if anybody has any thoughts on what rule changes would be required in order to render the AI competitive? Hopefully it'll be redundant if there's a new expansion that improves the AI's ability to win the game, but any changes to the AI in that expansion could be based on feedback from players' experience now.

From the above, despite all the noise that overflow and Magnus get, I'm not sure that they're the reason why the human can win so much quicker than the AI. Similarly, just not taking advantage of the AI militarily doesn't make the AI competitive either. So does anybody have any ideas on what mechanisms could be changed to slow the human player down to the point that the AI offers a real threat of winning first on higher difficulty levels?

 

[FearSunn]

Remove wonders from the game.

 

[Trav'ling Canuck]

Remove wonders from the game.

Because the AI wastes too much production building them?

 

[FearSunn]

Exactly. Production is by far the most important yield in the game. AI gets production bonuses on higher difficulties and wastes mostly all on useless wonders. Ban wonders building for AI and you will get much better AI.

 

[Trav'ling Canuck]

Exactly. Production is by far the most important yield in the game. AI gets production bonuses on higher difficulties and wastes mostly all on useless wonders. Ban wonders building for AI and you will get much better AI.

So flipping that around, I could force myself to build a wonder each era. That could replace the "force myself to build a Holy Site and get a religion" rule, absorbing production not just in the ancient era, but throughout the game.

And from the development team's perspective, the take away should be that either wonders need a general cost reduction or the AI needs to be taught to more selectively pick which wonders are helpful in any given circumstances.

 

[FearSunn]

And what if you run your game test (it is you who run game test right?) to see how much removing wonders really helps for AI to win quicker?

 

[Trav'ling Canuck]

And what if you run your game test (it is you who run game test right?) to see how much removing wonders really helps for AI to win quicker?

I could do that, but I'm not sure if the output will be worth the effort to generate it? I doubt anyone wants to take wonders out of the game, so what we'd be testing is whether the AI is actually weakened by building wonders, which in turn would suggest either that wonders are generally overpriced or built in the wrong circumstances by the AI. Which in turn would then require another set of tests to see if the results differ if wonders are reduced in cost by 20%, say.

I'd be willing at some point in Civ 6's development to make that effort, say as part of a community balance patch when I knew the output would be considered when making development decisions. I'm doubtful Firaxis would base their own balance patches off any work I do.

For the moment, I'm inclined to test it in the more round-about way of forcing myself to build wonders, and see how much that slows me down even if I pick the best wonder I can for my current situation. That kind of cuts directly to the question of whether wonders are worth their cost.

 

[FearSunn]

For the moment, I'm inclined to test it in the more round-about way of forcing myself to build wonders, and see how much that slows me down even if I pick the best wonder I can for my current situation. That kind of cuts directly to the question of whether wonders are worth their cost.

Ok. Looking forward for your impressions.

 

[c4c6]

So flipping that around, I could force myself to build a wonder each era.
[...]
And from the development team's perspective, the take away should be that either wonders need a general cost reduction or the AI needs to be taught to more selectively pick which wonders are helpful in any given circumstances.

And flipping that around again, I conclude (for myself, just in case no magical AI change happens in this sector): wonders need a general cost reduction along with restrictions (aka house rules) for the human, because 'none wants to take wonders out of the game'.
Most simple would be 2 different 'price lists' for AI and human.
Or build restrictions for the human: just one or two wonders per era ...
maybe random wonders ... I did this in my last civ1 games - assign available wonders to a dice and (rock&) role

 

[Mahi]

As I always do. By peaceful victories. War as you say is too easy

 

[chazzycat]

Basically role play as a benevolent leader. No aggression, worker stealing, or anything that could be construed as evil. Focus on amenities & housing to give my people the best life. Try to be peaceful with everyone. Bind the world together with alliances for stability, but maintain a strong military to keep the peace. Play world police. If anyone gets aggressive, respond with righteous wars of liberation. Pursue culture & science victory at the same time so my people are as enlightened as possible.

 

[UWHabs]

I'm not sure if there's anything to bring you right in line with the AI, unless if you also add rules like you have to keep forgetting to improve your terrain, and need to make sure to always build districts, especially harbors, in the worst possible spots. And probably something about choosing the worst government, or using a policy card that doesn't make sense (ie. Kongo using the double holy site adjacency card). And of course, as mentioned, the AI's weird obsession with building a wonder even if it doesn't help (Petra on the only available desert tile, etc..)

But overall, yes, avoiding war is probably the #1 thing to get close to the AI. I would also suggest doing as best as you can to "role-play" your leader using their agenda as a guide. So if you play as Brasil, then you tell yourself that you won't chop the jungle. You need to set yourself a goal of using every one of a civ's ability to the maximum - as Arabia, that would be spending all your effort to convert foreign cities to your religion, for example.

 

[liv]

Look forward to hearing about your experiment

It is a game of expansion so if you limited yourself to a few cities you would give the AI a chance to win. I am not sure what that number could be. Maybe 5?

 

[Trav'ling Canuck]

It is a game of expansion so if you limited yourself to a few cities you would give the AI a chance to win. I am not sure what that number could be. Maybe 5?

For my skill level, I'd guess 5 to 6. Better players can likely do it as a single city challenge. That's probably the simplest single factor to control.

 

[Archon_Wing]

By actually growing your cities

Or only building izs lol

 

[Boris Gudenuf]

I'm not sure if there's anything to bring you right in line with the AI, unless if you also add rules like you have to keep forgetting to improve your terrain, and need to make sure to always build districts, especially harbors, in the worst possible spots. And probably something about choosing the worst government, or using a policy card that doesn't make sense (ie. Kongo using the double holy site adjacency card). And of course, as mentioned, the AI's weird obsession with building a wonder even if it doesn't help (Petra on the only available desert tile, etc..)

But overall, yes, avoiding war is probably the #1 thing to get close to the AI. I would also suggest doing as best as you can to "role-play" your leader using their agenda as a guide. So if you play as Brasil, then you tell yourself that you won't chop the jungle. You need to set yourself a goal of using every one of a civ's ability to the maximum - as Arabia, that would be spending all your effort to convert foreign cities to your religion, for example.

My solution to make certain that I do all these things and 'even up' with the AI is very simple: drink about a liter of vodka and then try to play when the screen is dancing joyfully in front of my eyes and I cannot remember from moment to moment which game I'm playing...

 

[Bitterman]

I use this: https://forums.civfanatics.com/threads/smoother-difficulty.603093/

It gives the AI constant bonuses which escalate through the game. That means that they will have a higher science/culture/production/faith/gold bonus the higher the era is. That means that catching up is more difficult the more advanced the game is. You will frequently be neighbouring some high-science AI (damn you Sumer) so you'll have to keep that in mind.

Also war can be tricky, since the enemy units also get flat combat bonuses (which can escalate or not, depending on the difficulty) and high production/science bonuses too, forcing you to create certain strategies or having to exploit some advantageous situations to have a reasonable chance at it.

It also adds two new difficulties, Demi-god and Omniscient, although I stick to Emperor beacuse skill.

 

[acluewithout]

My 2 cents.

I mostly stopped playing shortly after the last big patch. To me, the issue isn’t so much the AI, more that there are too many mechanics that just don’t work because of balance issues, which really narrows the range of fun strategies. That plus the Governors and the Government Plaza / Tier 1 buildings being so “boring” (yet also unavoidable) and I just lost all momentum.

But. Before I did run out of steam, I did play with various rules to make the game more interesting.

1. Science. There are two problems with Science. First, it’s very easy to race through the tech tree. Second, relatedly, it’s very easy to massively out pace the AI. I suggest limiting yourself to no more than three campsuse. That makes things more competitive, but also makes generating science more challenging - you need to lean on Eurekas, trade routes and projects much more. It also helps if you use a mod like 8 ages of pace. I haven’t tried it, but mods that give the AI progressive bonuses over time might also help.

2. War and ICS. I don’t recommend playing peacefully. Too much of the tactical challenge of the game is tied up with war (including diplomacy), that just having a blanket now war mechanic makes the game too boring. War is also often the only way to deal with forward settling. Instead, the important thing is to limit the number of cities you have. I usually limit myself to 8 or 12, with some exceptions for late game cities and loyalty flipping. Relatedly, you may want to limit your military - eg max x melee / heavy cav units, or no using professional army. I’d even had a few games where I rerolled because horses or iron were too easy to get.

3. Use more mechanics. Some mechanics just don’t give great value for their investment - religion, some early wonders, maybe alliances. But if you just ignore these mechanics, again the game is very boring. Spending some hammers on things that aren’t strictly necessary will also help you avoid outpacing the AI again. My suggestion is to just make using certain mechanics a pre-req for your victory type. e.g. you must get a religion, or you must build x wonder.

4. Culture victory. @Trav'ling Canuck , I know from past posts that you’re not super keen on culture victory, which is totally fair and I accept CV does have some problems. But CV is probably the victory that best ties in the various mechanics. eg there’s some value to war without conquering, eg to reduce your opponents culture or steal great works, or curry favour with other Civs; diplomacy has some value because you need open borders; late game cities have some value for seaside resorts and trade; wonders and religion have some additional value; more techs, civics and policy cards are important. The AI is also quite good at defending culture victories given that just requires the AI to produce culture. So, overall, I think CV is sort of the best victory type for the player to pursue if you want the AI to have a chance of winning.

5. Play less powerful Civs. It’s generall easier to have fun if you play less powerful Civs. This is maybe where I’ve get a little fed up with the game though after England got nerfed. A lot of the non-warmonger, non-turtle less powerful Civs suffer from some key mechanics not working as intended, closing off interesting strategies for these Civs. So, for example, Norway can be a lot of fun, but balance problems with Pikes and Military Tactics become really frustrating after a few games. Their religious building is also a pain - if you don’t have a religion, you’re sort of always getting pushed into GrandMasters chapel if you really want to use this building, which gets boring fast. Japan and Georgia suffer from similar problems. Likewise Germany, which isn’t that poweful overall, can be very frustrating because of Pikes / MT again, but also the weakness of IZ buildings. But overall, you need to play weaker Civs to give the AI a chance.

So, they’re my thoughts. But can I add that, looking at all those points, it’s clear the game isn’t in great shape. I don’t mind playing with some house rules or one arm tied behind my back. But the extent to which I have to house rule the game is crazy. And even then, you’re still stuck with some real mechanical drags, eg government plaza and governors.

I really hope the next expansion addresses the remaining balance / gameplay issues, and reworks a few of the RnF mechanics so they are just less boring. If the expansion also integrates / ties up the various mechanics too, then that would also be great.

And. You know. Fix England.

(...Hmmm. That's a bit more than 2 cents. More like a buck fifty in change. ... oh well...)

 

[Duuk]

My solution to make certain that I do all these things and 'even up' with the AI is very simple: drink about a liter of vodka and then try to play when the screen is dancing joyfully in front of my eyes and I cannot remember from moment to moment which game I'm playing...

I literally opened this thread to reply "drunk" but you beat me to it.

 

[Quoth the Raven]

Close Civ 6. Open Civ 4.