Biggest problem this game still faces

[MantaRevan]

A bit surprised dumb AI has so many votes, especially when compared to shoddy, one-way diplomacy, a grotesquely out of hand warmonger penalty, and terrible social policy balance.

 

[SemperFi2382]

A bit surprised dumb AI has so many votes, especially when compared to shoddy, one-way diplomacy, a grotesquely out of hand warmonger penalty, and terrible social policy balance.

Dumb AI has a lot to do with the poor Diplomacy.

I voted Dumb AI because:

• Failure to grasp the big picture.
• Combat is still lacking.
• Raging Barbarians are still inadequate
• Poor diplomatic choices.

 

[Sturmgewher88mm]

Wow, I'm surprised that NO ONE said it's too hard

It's kind of ridiculous that increasing the difficulty just gives the AI more advantages over the player, as opposed to making the AI more of a challenging opponent. COD wouldn't be any fun if "playing at a higher difficulty" meant fighting bots with extra health and super speed. The AI gets smarter with each difficulty, and that's how Civ should be.

 

[qemist]

I would also seriously love to see a big-time ranged units nerf, though.

I'm fairly new to BNW having previously played Vanilla then GnK. My impression is ranged units have been hp buffed at least. That seems strange given that one of the most common complaints was that there was no point building land melee units (apart from a few pikes and one horseman for taking cities) before industrial.

Dumb AI has a lot to do with the poor Diplomacy.

I voted Dumb AI because:

• Failure to grasp the big picture.
• Combat is still lacking.
• Raging Barbarians are still inadequate
• Poor diplomatic choices.

Poor AI is at the core of what is wrong with the game. Unfortunately reading between the lines that Firaxis says tells me that they have given up on developing an AI that that is more than bonuses, scripting and cheating.

Wow, I'm surprised that NO ONE said it's too hard

No-one is going to say the game is too hard because it has difficulty levels which are ridiculously easy.

It's kind of ridiculous that increasing the difficulty just gives the AI more advantages over the player, as opposed to making the AI more of a challenging opponent. COD wouldn't be any fun if "playing at a higher difficulty" meant fighting bots with extra health and super speed. The AI gets smarter with each difficulty, and that's how Civ should be.

See above. The AI has, as far as I have been able to tell, no idea how to play the game. By that I mean it has no position evaluation function. Its plays are not based on an evaluation of what leads to the better position for itself.

 

[PhilBowles]

A bit surprised dumb AI has so many votes, especially when compared to shoddy, one-way diplomacy, a grotesquely out of hand warmonger penalty, and terrible social policy balance.

I've just been playing a militaristic game, and I think the warmonger penalty is fine and well-implemented, with one glaring issue:

...

It's in the wrong game.

The problem is, Civ games are not particularly complex in terms of "war goals" - take a game like one of the Paradox titles, and there are various different motivations for going to war, and different objectives the game allows you to achieve depending on your cassus bellis. By contrast, Civ is closer to a Total War style, where the only objective of war is to capture enemy territory.

Make that unattractive, and what do you have left? This came to me while defending a war against France, who had backstabbed me (probably a first in BNW). Napoleon played extremely badly - the AI now seems much too cautious about risking units, with the result that merely moving mine to threaten his attack caused him to wander around the map uncertainly, without attacking my city and loosening his formation - he only put up a fight when I went on the offensive.

But that's an aside - the point is, I defeated his attack and went on to press him back towards Orleans. I looked at the city, and at the other cities I could see - he had iron from somewhere, which I was lacking, but I suspect it was from a trade since I couldn't see it in the landscape. All cities but Orleans were in tundra (not the fault of AI city placement; he had a genuinely bad starting location, hemmed into the south by me, to the east by the coast, and to the west by Bogota and the Inca); Orleans was just flat, undeveloped grassland - fine if I wanted a food city, but I already had one.

So I just extorted the money he had for a peace deal, and went home. War over, nothing much achieved. So, on the one hand, I love the way the warmonger penalty forces you to think strategically, going to war only for particular target cities to minimise the diplo hit. It's a longstanding weakness of Civ games that most wars are wars of annihilation, but it's often more fun to have a longstanding nemesis who you clash with repeatedly, but without any intention of wiping each other out (which is the case for most wars in reality). However, without any other good reason to go to war conflict is much rarer than it ought to be. Particularly as war is a fundamental part of civ - every incarnation of the game has more military units than it has any other type of buildable item for a reason.

 

[sendos]

I still think the poor AI opinion is a bit too cynical. Every AI has some form of a plan of action when they start the game. Judging AI on just combat and diplomacy is not enough. What about their policy plans? They also beeline to the best wonder techs so you have to work even harder if you want them. Here are some examples:

Austrians choose patronage and aestethics almost always in my game to make good use of their UA and UB. Is that dumb? They often choose order because they end up with a wide empire late-in game. Venice AI always chooses tradition and they usually get commerce and patronage to bolster their influence over CS'. They'll always choose freedom unless there is a lot of pressure coming from a dominant ideology (most likely order). Religious AI's know how to use trade routes to spread their religion. It happened to me when Brazil was sending their trade routes to my crappy new cities just to spread religion faster. Cultural AI's always beeline to the internet and then aim for science victory. Zulus always build an army of impis and we always warn newcomers to plan ahead. Surely that contradicts the dumb AI argument to a moderate extent as the AI has a plan of sorts.

To keep an open mind, my main criticism of AI is when they go on an expanding spree late in game to grab every space there is. This obviously doesn't benefit their long term goal. Other issues involve CS AI, which is rather volatile. Even if they amass a small army, they still don't attack. Another issue, for any AI, is that their ranged units can't walk and chew gum at the same time, unless they're the UU's (Keshik, Camel Archer, CKN's). There are times when I see non-religious AI's choose piety or get the Hagia Sophia. When it comes to ideology, sometimes they buckle to the pressure of dominant ideologies a little too easily but it's not always the case.

As for combat, the AI at least knows how to use the flanking bonus. They send 2-3 units against your unit to ensure that they kill your units before attacking the city. Mongolian AI also know how promote their keshiks, but the AI falls short on logistics and blitz. AI also gives their melee units medic and cover promotions so they are somewhat aware of the strength of ranged units. The warmonger penalty, although rather heavily weighted, is designed to improve the perception of AI about any warmonger players/AI's. I would definitely support dumb AI vote if warmonger penalty was still weak from vanilla.

As for the big picture, it's fair to say that tall empire bias and science >>>>>everything is the bigger issue. If you play civ 5 on a high difficulty, you will most likely lose to an AI science victory. You could only lose to AI via a cultural victory if you fail to pay attention to tourism and culture. You could only lose to AI ivia a diplomatic victory if you fail to look at the number of votes that Greece or Siam has. The AI still have some beneficial traits which make this game challenging. They do need more work on AI combat, but that specific problem isn't enough to call them dumb.

 

[reddishrecue]

I kind of agree with what you guys are trying to say, so the AI is not that dumb in the harder levels at times where they sometimes wonder spam back, take the wonder and have you paid. Another issue that could be a problem that isn't mentioned could be a bad starting location for a few players. The game could still add more things to allow players to still be a challenge while having a bad start (i.e tundra start with dance of the aurora) .

 

[qemist]

I would definitely support dumb AI vote if warmonger penalty was still weak from vanilla.

The warmonger penalty is just hardcoding. What it smart about that? Why should a player (that is what the AI is supposed to be) especially care whether other players capture cities or not? Capturing cities might help them win a domination victory, but not capturing cities might be better for a cultural victory. It knows the point of the game is to win, one way or other.

 

[Bezurn]

Anyone feel that food is overpowered and that science should not be a function of population size?

Sent from my Galaxy Nexus using Tapatalk

Yes - I much prefered civ4's method of dealing with science as a means of commerce than as a means of number of people in a city. It makes a lot more sense for a wealthy nation to be more technologically advanced than it does for a over crowded nation.

 

[basy]

If the biggest problem is dumb AI (I think it is), then why dont you guys play it against each other?

 

[Talcove]

If the biggest problem is dumb AI (I think it is), then why dont you guys play it against each other?

Because multiplayer is a hassle (Finding people who won't quit, getting friends together), lacks some singleplayer features (Leaderscreens), and playing custom or modded games is ridiculously hard. With singleplayer you can just pop it up whenever you want, and leave it whenever you want. With singleplayer you ca play with however many mods you want. With singleplayer you can play with 22 civs and not have to worry about latency or lagg (In my case, anyway).

 

[PhilBowles]

I kind of agree with what you guys are trying to say, so the AI is not that dumb in the harder levels at times where they sometimes wonder spam back, take the wonder and have you paid. Another issue that could be a problem that isn't mentioned could be a bad starting location for a few players. The game could still add more things to allow players to still be a challenge while having a bad start (i.e tundra start with dance of the aurora) .

"Dumb AI" is not itself a problem, it's a symptom. All an AI code does is tell the program to respond in a certain way given a particular situation, a series of IF...THEN statements in essence. This is intrinsically difficult for a strategy game with as many variables as Civ, and an AI will come across as "dumb" if a design is overly complex or doesn't provide the game with suitable options.

Take, for instance, the World Leader vote. Past Civ games handled this in essentially the same way Civ V handles resolutions - a particular candidate was put forward (or rather, two were), and the AI simply had to decide "for" or "against" when allocating its vote. And if you look at the way Civ V AI handles resolutions, it's as good at making this kind of decision as the AI in Civ IV.

The trouble is, Civ V's leader vote works to a different system - it essentially lacks a "vote against" option. The new system requires an AI to be able to situationally analyse which of up to 22 options is the best fit for their interests, if they themselves can't obtain enough votes, while the old system just asked the AI to calculate "will I win if I vote for X instead of Y? If not, will I lose if I vote for X instead of Y?" - other situational variables are irrelevant. The non-binary Civ V system is much too complex computationally for an AI to handle, and so it only ever votes for itself.

AI coding in Civ V is at least as good as that in most strategy games, but the game itself has elements that are simply not AI-friendly. Combat, and when to declare peace and on what terms, are particular issues. Take a system like Paradox games' warscore - the AI is as dumb as it can get, and will routinely fail to make peace when out of armies and in occupied territory just because the warscore hasn't hit 100%. It won't even raise new armies during a war (a CK II patch currently in beta specifically aims to fix this). But the system works a lot better because the fewer decisions the AI has to make, the better it will perform - the AI in CK II doesn't need to weigh peace deal options, or calculate the likelihood of particular terms being accepted, or evaluate its relative performance during the war to determine whether it's winning or losing - the system automatically generates a score to tell it whether it's winning or losing, and sets the terms of peace in advance.

 

[SemperFi2382]

"Dumb AI" is not itself a problem, it's a symptom. All an AI code does is tell the program to respond in a certain way given a particular situation, a series of IF...THEN statements in essence. This is intrinsically difficult for a strategy game with as many variables as Civ, and an AI will come across as "dumb" if a design is overly complex or doesn't provide the game with suitable options.

I disagree that it's just a symptom.

For example, making a defensive pact with 2/3 of the world should signal to the AI that attacking you is ill advised. I've seen this time and time again where a weak civ will still fight a massive alliance. IF...THEN statements should clearly make the AI reconsider its strategy.

On the other hand, the AI that is already losing a war badly and is either in no position to join another war, or offer aid should simply refuse. It's not complicated, and if the AI wasn't so stupid it could recognize this.

It's the fact it can't evaluate diplomacy AT ALL on a legitimate level, it requires cheating to make it "tougher", and even simple IF...THEN statements should give the AI some level of intelligence it lacks.

For example, distance to target for pacts to join an alliance or declare war would prevent stupid agreements. Evaluating trades of gain versus loss (with relations as a modifier) would also help. Give gold per turn to have open borders, yet they don't even know where your lands are? Seriously?

Don't even get me started on the barbarians and their horrible choices that even a child would not make.

In my last game, I had a civ declare war on me when I had all but one other civ in a defensive pact with me, guess who lost? However, the other AI didn't even bother to help. Try that in game with humans, and either one side backs off, or you're gonna have a global war.

Play a game with humans, and the game is fine. Play it against AI and you realize it is predictable, and in some cases dumb as a brick. Symptom? Hardly.

 

[bc1]

Actually it's the tactical AI which is the most lacking. The move to 1 unit per hex has simply made the game too difficult for the AI to play (not that any of the previous civ series AI was particularly bright either).

 

[GhostSalsa]

Give gold per turn to have open borders, yet they don't even know where your lands are? Seriously?

They get the tourism modifier out of it regardless.

That said, I hate these new cheapskate gpt offers. The game should not draw out my turn times to nickel-and-dime me with bad offers the AI would never accept the same price for. I don't trust between-turn-trades for anything anyway since it keeps incorrectly giving away my last copy. UGHhhhh

Actually most of the stuff you listed as improvable I would regard as discardable. Defensive Pacts? No function in SP so get rid of them. Agree to go to war? the same AI that asks you will hate you if you conquer any cities so get rid of it. These systems are broken so get rid of them. The AI should be given less things to be stupid about.

 

[PhilBowles]

I disagree that it's just a symptom.

For example, making a defensive pact with 2/3 of the world should signal to the AI that attacking you is ill advised. I've seen this time and time again where a weak civ will still fight a massive alliance. IF...THEN statements should clearly make the AI reconsider its strategy.

Defensive pacts are secret agreements - the AI opponents don't know about them. This is in itself a bizarre design decision, and one I didn't know about originally, but it's not an AI fault.

On the other hand, the AI that is already losing a war badly and is either in no position to join another war, or offer aid should simply refuse. It's not complicated, and if the AI wasn't so stupid it could recognize this.

This is broadly true - other Civ games handled this a little (but not much) better from recollection. The problem, however, is more complex than you make it sound - how do we define "losing a war"? This is something no Civ AI has cracked - in some ways Civ V's is an advance in this regard, since while it offers terrible peace terms, more often than not it knows when it should offer/accept peace (as opposed to answering "You're joking, right?")

Civ V's peace-brokering system is inherently flawed in that it's based wholly on relative military power (as is its system for determining whether to go to war), the same calculation that's described by the military advisor.

Developing a better system is however not a simple task, and it's the reason games like CK II have a warscore system which takes decision-making away from the AI; a better system needs to factor in not only military strength, but numbers of losses on both sides, territory lost on both sides, and strategic positioning, and have some algorithm for assigning weights to each of those factors relative to one another.

It's the fact it can't evaluate diplomacy AT ALL on a legitimate level, it requires cheating to make it "tougher",

Every strategy game out there uses bonuses to AI players to add challenge at higher levels, and Civ has done this in every incarnation; comparisons like those made in an earlier post to Call of Duty, a far simpler game in terms of AI processing, are entirely irrelevant. Mechanically, and in terms of decision-making involved by the AI, games like Total War and Crusader Kings II are simpler than Civ V, and yet these games do exactly the same - higher difficulties = more units and more money for AI players relative to the human, not better AI play.

For example, distance to target for pacts to join an alliance or declare war would prevent stupid agreements. Evaluating trades of gain versus loss (with relations as a modifier) would also help. Give gold per turn to have open borders, yet they don't even know where your lands are? Seriously?

Recall that in BNW open borders is a major modifier for spreading tourism - as a human player I now routinely offer deals for open borders I don't need for passage if I'm wanting to spread my influence.

 

[cinattra]

I think the Civ AI is constrained by its choices just like the human players choices are constrained. The social policies should be more open and/or offer more flexibility. The social policies really need to be structured for change and should not be static through time or it could be a combination of static and changeable policies.

I liked how in games past you were allowed to change governments if you were willing to go though your period of unrest. Why can't we do the same with social policies? The more policies you want changed the longer the unrest.

I also think they should think about bringing back government types which would essentially be ideology modifiers.

I voted for Tall Empires as being the biggest problem and now I'm modifying that to the structure of social policies and ideology to being the biggest problem.

 

[PhilBowles]

I think the Civ AI is constrained by its choices just like the human players choices are constrained. The social policies should be more open and/or offer more flexibility. The social policies really need to be structured for change and should not be static through time or it could be a combination of static and changeable policies.

I liked how in games past you were allowed to change governments if you were willing to go though your period of unrest. Why can't we do the same with social policies? The more policies you want changed the longer the unrest.

Actually, in the game I just finished I changed ideology for the first time; the Inca decided - while I was at war with them - to adopt an ideology, and went for Autocracy, the ideology of their French neighbour (and also enemy at the time). Spain happened to adopt Autocracy at the same time. This sudden wave of autocrats led to immediate revolution among both the Huns and the Ottomans, leaving just Babylon (Order), me and Venice (both Freedom). My tourism was twice as high as anyone else's individually, but not enough to resist ideological pressure and I started getting dissidents. So I switched before facing more severe repercussions than losing a couple of turns' research and two ideology policies.

It's a good system but much too late in the game to be much more than flavour, or for penalties to be serious if you switch ideology early enough.

I also think they should think about bringing back government types which would essentially be ideology modifiers.

An ideology modifier is basically a social policy, and the social policies already have the names of past games' governments and/or civics.

I voted for Tall Empires as being the biggest problem and now I'm modifying that to the structure of social policies and ideology to being the biggest problem.

That's a flavour issue, but it's not one of the more significant game problems Civ V faces.

 

[cinattra]

Actually, in the game I just finished I changed ideology for the first time; the Inca decided - while I was at war with them - to adopt an ideology, and went for Autocracy, the ideology of their French neighbour (and also enemy at the time). Spain happened to adopt Autocracy at the same time. This sudden wave of autocrats led to immediate revolution among both the Huns and the Ottomans, leaving just Babylon (Order), me and Venice (both Freedom). My tourism was twice as high as anyone else's individually, but not enough to resist ideological pressure and I started getting dissidents. So I switched before facing more severe repercussions than losing a couple of turns' research and two ideology policies.

It's a good system but much too late in the game to be much more than flavour, or for penalties to be serious if you switch ideology early enough.



An ideology modifier is basically a social policy, and the social policies already have the names of past games' governments and/or civics.



That's a flavour issue, but it's not one of the more significant game problems Civ V faces.

Look... I have been playing this game since Civ I. The game as of today feels way too artificial. People are constantly playing the same path every game when in reality governments have picked and chosen their policies from among all government types. In other words no government type has a monopoly on a specific policy. Policies in the real world come about through government seeing a need and trying to fill that need. True some ideological types are more likely to pick certain types of policies than others, but again they pick and choose and are not tied into a specific path.

My argument is that Civ needs a similar flexibility. The current system attempts to create conflict. Because I believe in universal healthcare and you don't means I don't like you doesn't make sense to me. That is not why governments and countries don't like each other. They want to force Civs to finish a policy tree so they can get some additional benefit for doing so.

A social policy does not modify ideology in Civ 5 because social policies come first so it is the other way around.

As for as flavor again it is artificially constraining how I choose to play. If they are going to have such a construct as inflexible policies trees then they need many more of them to choose from.

 

[Merciorum]

Dumb AI for me. But there are very few games with remotely decent AI, so I can forgive it .