Why does civ 6 feel so much easier than the older civ games?

[Eddie Verdde]

It's funny that someone brought this up because over the last few days I've been trying some mods that aim to improve the AI (AI, AI+ and Real Strategy) and I've been asking myself the exact same question:

"Why does Civ6 feel so much easier than Civ games?"

I think the answer is simple: the game became so complex, with so many different interacting systems, that it also became harder to program AI behaviour.

Units are divided in melee, ranged and siege; there are health bars, promotions and flanking bonuses. Moreover, there's a limit of one unit per tile. When going to war, the player has to consider how strong the opponent is, which type of units he has, the resources available and the type of terrain where the battles will be fought. The player has to figure out which units to build, which units to send ahead, how to place them, when to retreat them, which promotion is more convenient and if it's better to retreat a damaged unit or pillage a farm and gain health.

When settling cities, the player has to consider what to expect from the new cities: gain access to a strategic or luxury resource? Build a district with strong adjacency bonuses? Use the city to produce military units and settlers while you build an wonder in the capital? Or just build a city to create a buffer zone between your capital and an expansionist neighbour? And then, with the loyalty system, the player has to consider the distance to its capital and the proximity of a neighbour in order to assess the probabilty of rebellion: will he be able to increase the loyalty by assigning a governor, buying a monument or unlocking the policy that increases the loyalty with a garrisoned unit?

When constructing districts, the player has to consider all the possible adjacency bonus (present and future) in order to find a good spot for the district.

Moreover, every now and then the player has to assess its progress when compared to the opponents and maybe shift the strategy. Should he focus on religion or culture? Is he lagging behind in technology? Should he maintain peace with a neighbour or go for war?

AND all of this has to be made in a coherent and efficient manner, considering the agendas of each leader and its specific abilities, buildings, improvements and unique units.

So, it goes without saying that the human player will be consistently better than any AI. Still, I think it's possible to create a challenging AI without too many bonuses or "cheats". But in order to get that, the devs of the game would have to put a lot of effort into it. And they won't be doing that as long as the mainstream gamer stays happy with just more civs and more leaders to choose from, as well as more unique units, eye-catchy graphics and modes that incorporate vampires and zombies into a Sid Meier's civilization game.

For instance, I quite enjoy the Barbarian Clans Mode and I think it was a nice ad-on, but the AI doesn't seem to be able to take advantage of it as I do. I will hire Barbarian Horsemen from a clan and use them to roam througout the continent, raiding other clans and then using the money from the raids to hire even more Horsemen until I have a strong army almost exclusively made up from Barbarian mercenaries. The AI never does that. Therefore, although the mode is nice, the AI was not programmed to deal efficiently with the Barbarian clans, which suggests that a challenging AI is not a big concern for the devs.

 

[SCBrain]

The most important part of 2-4 units per tile as a possibility is that it avoids (or moves to a higher level of abstraction) the sliding puzzle problem. The AI’s archer wont be boxed in because of two mountains and 3 other units in the way. There is more “give” in the map. And the dumb AI won’t trap itself. The AI is actually pretty good in the first 50 turns, when it only has enough units for 1UPT.

All that said, when I read my first post at the top of the thread, I see that I wrote that 1UPT and the science bug are the usual reasons given for AI weakness. I also wrote that the AI becomes very passive after turn 100 or so, and it might be before that. It’s a leap of logic to say that this weakness is intentional on the developer’s behalf, because most gamers like easy games. Who knows if that’s true?

 

[Alaindor]

It’s a leap of logic to say that this weakness is intentional on the developer’s behalf, because most gamers like easy games. Who knows if that’s true?

I don't believe that, myself. If that were true, it wouldn't behave incoherently the way it does, it would just cap some numbers (so many units, stop attacking after x,...). Instead of that it does some sound moves, and it does some stupid moves. That is not intentional.

Rather, I believe the devs did not put too much effort into the AI. "Teach" it the rules, put some simple if-then conditions, and that's it: good enough for what the general public wants.

If you want a truly "smart" AI for this type of games, I believe the way to go now is by using machine-learning. Look at this old article back in 2016: https://techcrunch.com/2016/12/06/a...-human-players-at-complex-civ-strategy-games/
... sure, it may not be as complex as Civ6, but that was a massive 5 years ago, and a lot of progress has been made since in that field. If you're worried about CPU power, you could have an option to use centralized computing power, where your device would query some central server for the next move and deal with graphics and UI. A complete different way to handle the game AI and a different business model you may argue, sure. But maybe some of us would be ready to pay for a subscription (I would, if the price is reasonable).

 

[Victoria]

Not intentionally dumb it down, but limit its abilities with lack of fine-tuning.
For example, just before the last patch there was a lot of discussions on how the AI was easy because it did not concentrate on science earlier. The patch now has AI doing that by spamming campus early and running out of gold.
How easy would it be for the AI to slot the double science adjacency bonus… but they do not, how easy would it be to say, hey we are running out of gold, how about some commercial hubs?
Either just programming in high campus starts was all they could do, or it was left intentionally so. It could well be the former, too many bugs and not enough developers? Certainly makes sense.
I may be wrong about the easy thing, it is a proposal, another one I have posited in the past is the number of artists required limits your development budget.

 

[Alaindor]

Well, regarding science, there seems to be a consensus that the "AI-science-crazy" is probably the result of a typo... and it is reasonable to infer that by exaggerately prioritising science over everything else, the commercial hubs and other districts get left out, as the AI logic is too simple (based on simple if-then assertions). Hence the AI runs out of gold, disbands military, etc. as logical consequences of an original error which has a butterfly effect.

So I analyze this as a lack of playtesting and QA, and more generally a lack of focus and effort put on the AI, rather than an intentional game-easy thing (which would make sense for lower levels, but if the devs had the skills, they would not need put such restrictions on immortal-deity levels anyway). Which ultimately produces the same result.

 

[Socrates99]

at least 2 mods for civ6.

3 combat classes stacking and 1 tile range was in my overhaul, and I've seen a mod expending stacking for more classes (which was not possible until an undocumented change in one of the patches)


it's not just numbers, it's also about combat rules, 1UPT and ranged combat on smaller maps is what truly kills the AI.

Unlimited stacks allowed the AI to convert its production bonus on one point of the map while 1UPT spread that bonus on multiple tiles which is much, much easier to handle tactically for a human exploiting choke points of cramped maps with ranged support.

Yeah, cool, I dont feel like I should have to keep repeating this but the AI in VI simply does not use that production bonus to outnumber you. It does not matter if they stacked or not because they do not build enough units to make it matter. Get it? I keep saying it over and over but they could stack units and itd probably make things easier, not harder, because their weak sauce army would be on one tile rather than 20. You could wipe it in one turn rather than 4 or five turns.

Before even talking about 1upt or stacking youd need to bump them up to what they used to build when stacked. At bare minimum they could at least produce like V's AI did. VIs AI just does not build military worth mentioning. There isnt even a way to compare stacked v 1upt gameplay because the 1upt versions do not build like the stacked versions did.

 

[Gedemon]

Yeah, cool, I dont feel like I should have to keep repeating this but the AI in VI simply does not use that production bonus to outnumber you. It does not matter if they stacked or not because they do not build enough units to make it matter. Get it? I keep saying it over and over but they could stack units and itd probably make things easier, not harder, because their weak sauce army would be on one tile rather than 20. You could wipe it in one turn rather than 4 or five turns.

Before even talking about 1upt or stacking youd need to bump them up to what they used to build when stacked. At bare minimum they could at least produce like V's AI did. VIs AI just does not build military worth mentioning. There isnt even a way to compare stacked v 1upt gameplay because the 1upt versions do not build like the stacked versions did.

people also hate carpet of doom, you are arguing about what should came first between chicken and egg, but in the end you need both if you want the AI to benefit from its production bonuses.

avoiding civ5's carpets of doom may be the reason why the AI units production is limited BTW, pretty sure they added limitations to avoid religious units spam at some point for example.

civ5 also already proved that you need a few more years of work to finally get a decent AI using 1UPT instead of unlimited stacks.

If you want a truly "smart" AI for this type of games, I believe the way to go now is by using machine-learning. Look at this old article back in 2016: https://techcrunch.com/2016/12/06/a...-human-players-at-complex-civ-strategy-games/
... sure, it may not be as complex as Civ6, but that was a massive 5 years ago, and a lot of progress has been made since in that field. If you're worried about CPU power, you could have an option to use centralized computing power, where your device would query some central server for the next move and deal with graphics and UI. A complete different way to handle the game AI and a different business model you may argue, sure. But maybe some of us would be ready to pay for a subscription (I would, if the price is reasonable).

pretty sure a lot of players would not want to play against such an AI, there is a good thread about machine learning AI and civ game here about the pro/con:
https://forums.civfanatics.com/threads/game-ai-net-based-machine-learning.627729/

 

[Socrates99]

people also hate carpet of doom, you are arguing about what should came first between chicken and egg, but in the end you need both if you want the AI to benefit from its production bonuses.

avoiding civ5's carpets of doom may be the reason why the AI units production is limited BTW, pretty sure they added limitations to avoid religious units spam at some point for example.

civ5 also already proved that you need a few more years of work to finally get a decent AI using 1UPT instead of unlimited stacks.


pretty sure a lot of players would not want to play against such an AI, there is a good thread about machine learning AI and civ game here about the pro/con:
https://forums.civfanatics.com/threads/game-ai-net-based-machine-learning.627729/

Isnt the thread about difficulty not preference? People kept pointing to 1upt as the reason its easier but that's simply not true. Carpet or stack doesn't matter. It's the numbers, I keep pointing it out. We could quibble about what people like or dont like and that generally leads nowhere. Honestly, I dont think I ever voiced a preference in any of my posts.

Let's not kid around though. There's a pretty massive gap between carpet of doom and what VI's AI does. Theres likely a better sweet spot in between that they're missing. Even just scripting them to protect each city with half a dozen units would greatly increase difficulty from where it is now just because AI love encampments and walls. If they need to reduce production, maintenance and resource costs to allow computer opponents to build and maintain deity level armies I'm all for it.

The answer to the question "why is it so easy?" isnt 1upt. It's "I meet a token force I can easily overwhelm then I can wipe them out city by city and only fend off occasional 2-3 straggler units at any given time." The problem with VI isnt 1upt itself it's that I really dont even need to use my smarty pants human brain to beat the computer. It's easy to out produce them.

 

[aieeegrunt]

at least 2 mods for civ6.

3 combat classes stacking and 1 tile range was in my overhaul, and I've seen a mod expending stacking for more classes (which was not possible until an undocumented change in one of the patches)

Ya that is what I was thinking. That plus increasing movement by one would solve a lot of issues.

it's not just numbers, it's also about combat rules, 1UPT and ranged combat on smaller maps is what truly kills the AI.

Unlimited stacks allowed the AI to convert its production bonus on one point of the map while 1UPT spread that bonus on multiple tiles which is much, much easier to handle tactically for a human exploiting choke points of cramped maps with ranged support.

Making a competent AI for Civ6 as it stands is going to be incredible difficult. You have to account for the sliding tile puzzle AND LOS rules. This eats a lot of CPU cycles for what often ends up as “AI shuffles around and gets shot to pieces by city strikes”

OH MY GOD ya if city strikes could be modded oit that would be nice as well.

Isnt the thread about difficulty not preference? People kept pointing to 1upt as the reason its easier but that's simply not true. Carpet or stack doesn't matter. It's the numbers, I keep pointing it out. We could quibble about what people like or dont like and that generally leads nowhere. Honestly, I dont think I ever voiced a preference in any of my posts.

Let's not kid around though. There's a pretty massive gap between carpet of doom and what VI's AI does. Theres likely a better sweet spot in between that they're missing. Even just scripting them to protect each city with half a dozen units would greatly increase difficulty from where it is now just because AI love encampments and walls. If they need to reduce production, maintenance and resource costs to allow computer opponents to build and maintain deity level armies I'm all for it.

The answer to the question "why is it so easy?" isnt 1upt. It's "I meet a token force I can easily overwhelm then I can wipe them out city by city and only fend off occasional 2-3 straggler units at any given time." The problem with VI isnt 1upt itself it's that I really dont even need to use my smarty pants human brain to beat the computer. It's easy to out produce them.

This problem mostly goes away if you turn all of the advanced systems off. The AI cannot handle any of it.

Playing the basic game the AI is redonkulously more effective. Unfortunatly it is still hamstrung by the Science AI Typo, at least until it is done spamming science buildings.

Personally I find all of the advanced systems total fun cancer and immersion destroying, so for me this is a win win

 

[Gedemon]

Isnt the thread about difficulty not preference? People kept pointing to 1upt as the reason its easier but that's simply not true. Carpet or stack doesn't matter. It's the numbers, I keep pointing it out. We could quibble about what people like or dont like and that generally leads nowhere. Honestly, I dont think I ever voiced a preference in any of my posts.

Let's not kid around though. There's a pretty massive gap between carpet of doom and what VI's AI does. Theres likely a better sweet spot in between that they're missing. Even just scripting them to protect each city with half a dozen units would greatly increase difficulty from where it is now just because AI love encampments and walls. If they need to reduce production, maintenance and resource costs to allow computer opponents to build and maintain deity level armies I'm all for it.

The answer to the question "why is it so easy?" isnt 1upt. It's "I meet a token force I can easily overwhelm then I can wipe them out city by city and only fend off occasional 2-3 straggler units at any given time." The problem with VI isnt 1upt itself it's that I really dont even need to use my smarty pants human brain to beat the computer. It's easy to out produce them.

My point about difficulty is that once you've solved your issue with the AI not producing enough units, you're left with the issue of it not being able to use them efficiently because of the 1UPT rules (emphasis on plural, as it's not only the fact that there is one unit per tile, but all movement/combat rules based around the concept in civ6 with less space on maps) independently of what the players preferences.

My comment on players preference is because they are not going to buy something unplayable or uninteresting and so there is a balance to maintain between interesting rules (for the human), simple enough rules (for a challenging AI) and playability (like moving your units around)

Limiting the numbers of units the AI is producing is a desired behavior if the players prefers to avoid carpets of doom at the cost of a lesser challenge, as having limitations on the number of units in a stack is a desired behavior if players prefers to avoid stack of doom at the cost of a lesser challenge.

And of course it's not the only reason why it seems simpler, it's one of many that are pointed out in this thread and others, the strategic AI is important, yes, but so is the tactical AI, you need both.

OH MY GOD ya if city strikes could be modded out that would be nice as well.

maybe not the most efficient way to do it, but I set a modifier using "MODIFIER_PLAYER_CITIES_ADJUST_RANGED_STRIKE" with a value of -999

 

[Alaindor]

pretty sure a lot of players would not want to play against such an AI, there is a good thread about machine learning AI and civ game here about the pro/con:
https://forums.civfanatics.com/threads/game-ai-net-based-machine-learning.627729/

Make it an option for immortal / deity levels, keep the basic low-CPU-cycles if-then AI for lower levels. Judging by the number of threads and posts complaining about the AI, I should think a lot of players also would like a really challenging AI (count me in)

 

[Socrates99]

My point about difficulty is that once you've solved your issue with the AI not producing enough units, you're left with the issue of it not being able to use them efficiently because of the 1UPT rules (emphasis on plural, as it's not only the fact that there is one unit per tile, but all movement/combat rules based around the concept in civ6 with less space on maps) independently of what the players preferences.

My comment on players preference is because they are not going to buy something unplayable or uninteresting and so there is a balance to maintain between interesting rules (for the human), simple enough rules (for a challenging AI) and playability (like moving your units around)

Limiting the numbers of units the AI is producing is a desired behavior if the players prefers to avoid carpets of doom at the cost of a lesser challenge, as having limitations on the number of units in a stack is a desired behavior if players prefers to avoid stack of doom at the cost of a lesser challenge.

And of course it's not the only reason why it seems simpler, it's one of many that are pointed out in this thread and others, the strategic AI is important, yes, but so is the tactical AI, you need both.

maybe not the most efficient way to do it, but I set a modifier using "MODIFIER_PLAYER_CITIES_ADJUST_RANGED_STRIKE" with a value of -999

Was there really a big outcry about "carpets of doom" though? I spent a lot of time on V's forum and I don't remember a lot of complaints. It came up sometimes but mostly it was just a retort people who wanted stacks back used when people said they didn't like SoDs. V definitely wasn't unplayable and the numbers its AI produced created stiffer competition than what we see in VI. Saying it made a game unplayable seems hyperbolic.

You're right about the rules though. Personally I think they should improve unit mobility if they want 1upt to be more viable. VI's more restrictive movement really hamstrings AI. I was kind of surprised when I got VI and found that rather than increasing the slow movement in V they made it even slower. I feel like this shows with the AI who are often more successful than others. Persia and Columbia both get mobility buffs and are a couple of the handful of Dom civs that can actually do anything. Quo's rocketboots was the first mod I downloaded and not only does it relieve my frustration but it also makes it so the AI who use horse or get a mobility buff can move in and focus fire with melee units. That definitely keeps me honest and forces me to take fewer risks.

 

[Eddie Verdde]

An article about the AI in Civ VI:

https://www.pcgamesn.com/civilization-vii/better-ai

 

[aieeegrunt]

Getting rid of the low mobility carpet of doom would do a lot to make it easier to make the Civ7 AI effective

 

[General_Sahib]

Getting rid of the low mobility carpet of doom would do a lot to make it easier to make the Civ7 AI effective

Hindsight is a great thing and I also now feel that carpet of doom is the right way to go when the only other option is stack of doom. I think we will have to tolerate suboptimal "AI" for a while yet, given the economic implications of trying to install proper AI into such games.

I'm not even sure why there was so much fuss over the concept of a "carpet of doom", as this quite reasonably reflects the nature of most episodes of attritional warfare throughout real history. Surprise attack from a near-hidden vertical stack is much less realistic.

 

[Eddie Verdde]

I'm not even sure why there was so much fuss over the concept of a "carpet of doom"

I can give you two reasons.

One is the tediousness of having to click a dozen times to move a dozen units across the map. When stacking was possible you'd just have to move the entire stack with one click until the selected destination (let's say, a city you wanted to conquer).

Another reason is the aesthetic aspect of having the map full with scattered units all over the place. When you're not at war you have to figure out where to place the units and often they will block the view of the tiles. It gets very hard to see the resources, the districts and the improvements.

 

[aieeegrunt]

Hindsight is a great thing and I also now feel that carpet of doom is the right way to go when the only other option is stack of doom. I think we will have to tolerate suboptimal "AI" for a while yet, given the economic implications of trying to install proper AI into such games.

I'm not even sure why there was so much fuss over the concept of a "carpet of doom", as this quite reasonably reflects the nature of most episodes of attritional warfare throughout real history. Surprise attack from a near-hidden vertical stack is much less realistic.

Have to disagree with you there. For most of human history outside of Western Europe in the 20th century the shulter an shulter carpet of doom contiguous front stretching hundreds of kilometres basically never happened.

The vast majority of warfare for almost all of human history consisted of stacks dancing around trying to stay fed and put the other guy out of supply

 

[TheMarshmallowBear]

I apologise if this was already stated, but for me,

I think the issue is that Civ 5 started to move away from being an Empire-Building game to a Board-Game (even though Civ 4 and previous iterations still played like Board Games)

Civ 6 was probably the most simplified of games, a lot of the depth and decision making was removed, for example:

- Taking two good systems, the Trade Route and Great Works system:
- In Civ 5, you had Resource Diversity, which meant that each city's gold output per route would differ, making certain cities better than others, in Civ 6, this was gone, as was any bonus, there's a handful of Great People that actually make a difference, and few handful of Wonders (not even Colossus offers any bonuses to Trade Routes, just an additional Trade Route slot)

- In Civ 5, a good portion of Wonders had Theming Bonuses that made reaching for them exciting and interesting (not to mention that you could constantly train Archeologist) in Civ 6, only Museums have it, and you are so limited (Because of the district restrictions, where you can't have a Museum in every city) that you often end up with several Artefacts lying around on the map.

Then you have other elements, like bulidings - they have been so simplified, there's o few of them in Civ 6 compared to Civ 5, and they don't do much of anything apart from the Government Plaza and Entertainment Complex/Water Parks. It's all just "+1 Science" per building. In Civ 5, buildings had a greater purpose, like Gardens, giving you a boost to people Production. the Granary that boosted certain resourecs. Aqueducts that retained Food when growing so you needed less food to grow. All of that, practically gone.

Civ 6 seems easy because there's just much less decisions to be making, sure you have the adjacency bonus game but they aren't "decisions", you see mountains, you know you're going to place a Campus over a Holy Site. See a River? Commercial Hub.

Sure you can choose one district over another but even then it's a fairly limited because there's only a handful of districts.

They removed any capping of expansion (other than scaling settler and worker costs), so there's no worry about settling "bad" spots, they're all viable.

Amenities rarely do any significant dent when it comes to Happiness/Stability of an empire.

A lot of the depth that the games had in previous iterations just feels removed in Civ 6 and I expect civ 7 to quite frankly continue in this direction. It's still a fun game, but I don't think it will age as well.

 

[Noble Zarkon]

OH MY GOD ya if city strikes could be modded it that would be nice as well.

That would really hamstring the AI's defences and make capturing their cities much easier!

I'm not even sure why there was so much fuss over the concept of a "carpet of doom"

Because it's tedious to move units about especially as they stop when another unit steps on their destination even when it's ten turns away from arrival!!

which meant that each city's gold output per route would differ, making certain cities better than others, in Civ 6, this was gone

Not really, in Civ VI the output from a trade route is determined by the Districts in a city so the output from a large developed city is much bigger than a newly established one.

 

[TheMarshmallowBear]

That would really hamstring the AI's defences and make capturing their cities much easier!

Because it's tedious to move units about especially as they stop when another unit steps on their destination even when it's ten turns away from arrival!!

Not really, in Civ VI the output from a trade route is determined by the Districts in a city so the output from a large developed city is much bigger than a newly established one.

Except the yield is usually negligble because it's like what.. +1 of the district's yield? Mid-late game that barely makes a difference.