Remove production/culture/growth/etc AI's bonuses. Add combat (CS% and exp% modifier)

[Rinnero]

AI is ok in empire-building. It can develop (build cities, wonders, units, adopt SPs, tech) almost like human (70-80% of human capabilities).

But what AI lacks is Tactical capabilities! Right now AI acts at 10-20% of how human can. (that means human player can kill an AI's army that is 5-10 times bigger) Right?

AI is so bad in combat, that it is just like you abuse it. And you cannot avoid it.

So, why dont we just add some CS% modifier to AI to make him stronger in wars. Instead It will no longer build tons of units.

Why "stronger AI's units" is better than "bigger AI's armies"?

First: player is effectively able to kill first line of attackers and avoid retaliation. Numbers of enemy units just dont matter anything. That means that even if AI has 300 units you often can slow and steady safely kill 3 of them per turn without any losses.

Second: AI uses its production bonus for better non military development (building wonders/buildings). That means that it can develop faster than best human player without bonuses. It forces human to go into war, because that is the only way not to fall behind -> abuse AI's tactical stupidness.


Of course these CS% modifier will not make AI smarter and will not make SP gaming experience like in MP. But it will make it closer to it.

You will be able to hardbuild some wonders, just like in MP. You will be teching at almost same speed as AI. You will be able to win without any wars at all, but in war AI will not be an easy prey. That also fixes diplo problem: when AI has 5 times bigger army, he thinks you are weak and declares war. But actually he cannot do anything. Dont know anyone who is not familiar with such situation.

Isnt that better than always falling behind in teching and number of units, but still winning due to overwhelming advantage in tactics.


At deity all AI units will have +50% CS modifier and 2x faster exp gain (including exp by buildings). Or exp requirements for next level are twice lower (effectively its the same). At deity I think its ok to have some production/etc bonuses, but not 2x as they currently now. Maybe 30% bonus at prod/culture/growth is ok to make it competent in peaceful development.

At king for example, +20%CS, +25% faster exp gain.
emperor +30%, +50% faster exp gain. +10% prod/culture/growth
immortal +40%, +75% faster exp gain. +20% prod/culture/growth

What is good is that such things I believe are quite easy to implement, and even to mod it. Its relatively easy to check whether this suggestion is good or not.

[TW_Honorius]

that would force you to build more units since your units are weaker. The AI needs to build more troops before going to war versus the human and learn to keep units near cities in case in lose. also what does CS% mean, not city-state. plus combat 1v1 is roll of the die, not set in stone, so the human player would either build more units to counteract that or just play defensive, neither makes the game better or the AI smarter.

[Babri]

By CS he probably means 'Combat Strength'.

[M_Sparkle]

Not 100% sure I like this solution, its a good idea but it seems like putting a bandage on a broken Arm.
Better AI programming is needed bits its doubtful if they would make that investment.
Would be nice to see it in a mod and see how it goes though.
A lot of the best mod ideas have made it into vanilla so u never know, if it works we could all have it sooner than you think!

X

[Becomedeath]

Whilst gearing the AI away from growth would most likely see an increase in militaristic strategies, I don't believe that it's the AI's overall strategy that is at fault when considering it's tactical capability.

Having stronger units isn't going to teach it how to play better, it's just loosely replicating the AI's comfort zone in probability demonstrated miserably by stacks of doom.

Tactical AI needs improvement, but I don't think that civilisation wide modifers are the answer. More research needs to be conducted by the development team into the choices that human players take in a 1upt world and why they make them.

Build it smarter, not stronger.

[Strategist83]

You've said it yourself: The tactical aptitude of the AI is unacceptable and inexcusably bad for a flagship game like CiV. We need Firaxis to fix the problems rather than ask for workarounds.

[fat_tonle]

how about somewhere in between.... double promotions, but less than half the units we normally see from the AI? Right now it's ridiculous. Can't get into ten turns on Immortal without staring in the barrel of 10 warriors.

[Becomedeath]

The thing is, you have to think like the computer does. If you give it combat bonuses, that doesn’t necessarily mean that the computer will decide that combat is the way to go. Lets say the AI doesn’t have iron, and you do. So the AI knows already that it doesn’t have the capability to build units as strong as you and has two options.

A: Build units enmasse for a suicidal assault to take some of the available iron.

B: Expand and find some.

In my experience, machine logic will generally take the path of least resistance and so expanding, while not making use of it’s primary attributes, has a higher chance of success than attack does in that situation. So this puts them immediately at a disadvantage as they are now playing a defensive game against a stronger neighbour. Not an issue if you’re the human player, but if you’re the machine, you don’t think in terms of acceptable risk, you think in terms of very carefully calculated risks.

The main problem with Civ 5’s AI is that when AI becomes this complex, you’re no longer looking at a simple maths engine, like Civ 4 principally was, you’re looking at something which will exhibit behaviour, and you can’t always predict what modifications are going to do accurately. Combat modifiers might, as described above, make the computer play an even more growth orientated game than before or equally, might make the AI over confident in it’s chances and become more of a mindless unit spawning idiot than it already is.

[IONDragonX]

I'd be extremely happy to see this change.

I hate seeing thousands of gold in the AIs coffers for no good reason. Building an empire is what makes Civ fun, not wars. Giving unfair advantages to the AI so that war is your only option at higher difficulties is absurd.

[Krikkitone]

The problem is the AI is ALWAYS going to be stupid.
So the issue for difficulty levels is to match the bonuses to counter the AIs stupidity as specifically as possible.

The area where the AI is the Worst is the area where it should get the most bonuses.
The AI is worst at tactics, so its biggest bonuses should be in tactics.
(The AI is probably poor at other things too, like beelining/city planning/ properly aiming for victory, etc.)... so it should also have peacetime/economic bonuses as well.

Ideally, the Areas I would give the AI bonuses in are
1. Tactical type bonuses (as above)
2. Production/Rush buy cost bonuses
3. Tech Cost bonuses

[Becephalus]

Originally Posted by Strategist83

You've said it yourself: The tactical aptitude of the AI is unacceptable and inexcusably bad for a flagship game like CiV. We need Firaxis to fix the problems rather than ask for workarounds.

What games have you played that have better tactical AI? I don't know of any. Civ 4 is certainly worse, Gal Civ (I have played maybe 2000 hours of GalCiv2) and Wesnoth (I was a developer on Wesnoth, played probably 5000 hours, and was for a while one of the best players there was), two things I have seen suggested here are definitely worse.

The tactical AI is actually pretty decent given how hard an environment it is operating in. Making AI for a combat system like this is super super hard. The main things they could do to help the AI are make all the units melee and to change all the promotions to a flat +10%/level.

That would make the AI "better" because the problems it has to solve would be easier.

People go around trashing the AI as though there is something better out there, but that is just a joke.

[Becephalus]

Originally Posted by Krikkitone

The problem is the AI is ALWAYS going to be stupid.
So the issue for difficulty levels is to match the bonuses to counter the AIs stupidity as specifically as possible.

The area where the AI is the Worst is the area where it should get the most bonuses.
The AI is worst at tactics, so its biggest bonuses should be in tactics.
(The AI is probably poor at other things too, like beelining/city planning/ properly aiming for victory, etc.)... so it should also have peacetime/economic bonuses as well.

Ideally, the Areas I would give the AI bonuses in are
1. Tactical type bonuses (as above)
2. Production/Rush buy cost bonuses
3. Tech Cost bonuses

If you ditched great people the strategic AI would be way better. great people is the AIs weakest strategic aspect by far, the human player just abuses the crap out of the system compared to them.

[Krikkitone]

Originally Posted by Becephalus

If you ditched great people the strategic AI would be way better. great people is the AIs weakest strategic aspect by far, the human player just abuses the crap out of the system compared to them.

True, although that seems something that is reasonably fixable either
Through AI making it figure out how to a) rush Wonders+Techs and b) use specialists to get great people

OR

Through game mechanics making the Other uses of great people better (make Manufacturies/Custom Houses have a Base of +6 with +3 at Chemistry/Economics, Academies +8 with +4 at Scientific Method).. and some AI boost to have them aim for specialists.
(also cutting the University/Wat specialists back to 1.. move the extra up to research labs)

[ShahJahanII]

a lot of the higher difficulty levels give advantages (extra units) to the AIs. If they came up with an awesome tactical combat system, then they can make that the difficulty modifier, instead of just giving them 3 settlers to begin with.

[Rinnero]

The area where the AI is the Worst is the area where it should get the most bonuses.
The AI is worst at tactics, so its biggest bonuses should be in tactics.
(The AI is probably poor at other things too, like beelining/city planning/ properly aiming for victory, etc.)... so it should also have peacetime/economic bonuses as well.

Thats it. Why giving AI knuckels, when he has lame leg and needs crutch?

Through game mechanics making the Other uses of great people better (make Manufacturies/Custom Houses have a Base of +6 with +3 at Chemistry/Economics, Academies +8 with +4 at Scientific Method).. and some AI boost to have them aim for specialists.
(also cutting the University/Wat specialists back to 1.. move the extra up to research labs)

http://forums.2kgames.com/showthread...g-Great-People
Tons of ideas how to balance and to add more variety to GPs.

[Strategist83]

Originally Posted by Becephalus

What games have you played that have better tactical AI? I don't know of any. Civ 4 is certainly worse, Gal Civ (I have played maybe 2000 hours of GalCiv2) and Wesnoth (I was a developer on Wesnoth, played probably 5000 hours, and was for a while one of the best players there was), two things I have seen suggested here are definitely worse.

The tactical AI is actually pretty decent given how hard an environment it is operating in. Making AI for a combat system like this is super super hard. The main things they could do to help the AI are make all the units melee and to change all the promotions to a flat +10%/level.

That would make the AI "better" because the problems it has to solve would be easier.

People go around trashing the AI as though there is something better out there, but that is just a joke.

I know nothing about AI programming other than understanding perfectly well that it is nontrivial. However, it is also quite clear to me that it would be well within the grasp of a capable AI programmer to stop the AI from doing stuff like sending its catapults in front of its army with zero backup - or letting its units try to fortify and heal up right in front of your army where any human player can see they'll lose that unit next turn. These things - and I could mention many more - are inexcusable in a Civ game, AI programming being difficult or not.

As for a strong AI, I would point you to SupCom II's AI after Sorian came back and helped them build it. StarCraft II's AI, while perhaps not terribly challenging to skilled players, also doesn't come off as simply idiotic like CiV's does.

If they employed a guy like Sorian I guarantee you you'd soon be seeing the AI doing things we can only dream of now. Unfortunately, Firaxis seems far less interested in creating a good experience for their customers and repairing the aspects of their game that didn't work out well than GasPowered Games or Blizzard does. Which is exactly why people are justly saying that Firaxis is just after their money - it is the exact message it sends. You don't see anybody calling Blizzard money-hungry, and for a good reason.

[MadDjinn]

"You don't see anybody calling Blizzard money-hungry, and for a good reason. "

Your argument worked right up until you said that.

They really do like their money. (especially once Activision got in there and split a single game into three)

They just tend to think their reputation is worth more than toilet paper, and want to ensure they have customers for the next game.

[Becephalus]

Originally Posted by Strategist83

well within the grasp of a capable AI programmer to stop the AI from doing stuff like sending its catapults in front of its army with zero backup

Say you fixed that behavior, you could do it. But maybe it would have unintended consequences. People would be complaining that the AI never uses its siege units, that they just mill around in the rear and it never manages to successfully attacks other civs.

Originally Posted by Strategist83

- or letting its units try to fortify and heal up right in front of your army where any human player can see they'll lose that unit next turn.

Maybe when you try to fix this the AI runs constantly and never finishes a siege. This ^$%# is hard!

Right now the AI can actually attack and kill other AIs and does so with regularity. This is not the case in many many TBS games.

Obviously the tactical AI could be improved, and I think that should be the main focus of any expansion, I could care less about more content. But improving the AI will be HARDER then making 15 extra civs and 32 great scenarios. Those things just take time but are relatively straightforward, improving the AI is actual work and hard work. On top of that the gaming press and the sheeple players would scream like holy hell if the expansion was just AI improvements.

The companies are there to make money, and the press/customers mostly just want shiny things on the box, that is why games are released in the state Civ5 was in, Its been happening for all 20 years I have followed the industry

Originally Posted by Strategist83

These things - and I could mention many more - are inexcusable in a Civ game, AI programming being difficult or not.

How are they inexcusable from a Civ game? It is par for the course. This is the best AI a civ game has had yet, even Civ col's AI was pretty crappy and that was on 3 yrs ago and a much simpler combat system! All TBS ais are crappy! People loved SMAC, include me. try to play that, the AI is just atrocious. Its horrifying.

Originally Posted by Strategist83

As for a strong AI, I would point you to SupCom II's AI after Sorian came back and helped them build it. StarCraft II's AI, while perhaps not terribly challenging to skilled players, also doesn't come off as simply idiotic like CiV's does.

You do realize that RTS AI is almost entirely covered up by the fact that it can use speed and multiple attention loci to cover up its horrible play? It is using these advantages over people to make up for huge deficits. Try playing against the Starcraft AI at 1/10th speed where you have lots of time to think. The AI is just just as poor tactically as a TBS AI. Not to mention the fact that last time I played Starcraft all the single player scenarios were just as easy to stomp all over as civ scenarios and the AI get huge bonuses via the setup of the scenarios. In a straight game the AI has zero chance without huge handicaps...kind of like civ...

The AI not doing obviously dumb things does not a good AI make. Many people in the past have made AIs that are very conservative and do nothing dumb. As a result they basically turtle up and sit still and never attack. Dumb things are also easier to notice in a game where the units are moving once every minute instead of every second.

Originally Posted by Strategist83

If they employed a guy like Sorian I guarantee you you'd soon be seeing the AI doing things we can only dream of now. Unfortunately, Firaxis seems far less interested in creating a good experience for their customers and repairing the aspects of their game that didn't work out well than GasPowered Games or Blizzard does. Which is exactly why people are justly saying that Firaxis is just after their money - it is the exact message it sends. You don't see anybody calling Blizzard money-hungry, and for a good reason.

A) Re: Sorian, maybe, maybe not. More AI programmers would help, but that costs money.

B) Every game company is just after people's money. Blizzard has taken the unique position of focusing on quality. This was risky and in the late 90s many people had lost a ton of patience with them. Luckily for them and us it has worked out. Do I wish the whole industry took Blizzard's approach? ABSOLUTELY. Of course then there would be no Civ games because they don't have the audience of Blizzard type games and thus cannot have the same resources devoted to them profitably. But it is rather silly to look at an industry with a long history pick out its member most focused on quality, and demand that every single company match that standard. it just shows a complete lack of understanding of business, the real world, computer games, well it makes it seem like many of the people posting here are teenagers who haven't had real jobs.

Yes Civ could be better, yes everyone including I am sure Firaxis wishes it was better.

[black213]

This is not the solution in making the AI better. Just the intelligence has to be improved, giving them bonuses is just dumb.

[Rinnero]

Firaxis wants money, Blizzard wants money, all devs want and need money. Bliz created reputation that they polish their games to "perfection". But that doesnt mean their games are better, it means that blizz's games are technically perfect. They dont lag, crash, have great lobby, super informative replays, and so on. But gameplay itself is 10 years old. Starcraft II is even worse than WarIII. (many reasons, not part of this thread)

SCII's AI is quite good, but turn-based AI is probably harder to program.

What is important: Firaxis doesnt have enough resourses to improve AI. Its so hard that you either improve AI or fix MP, improve balance and make several DLCs. Tactical AI cannot be improved to acceptable state by changes in variables, it needs much more major changes.


But on the other hand, by adding combat bonuses to AI (its very easy), we can make things better to some degree:
AI will play with the same rules as human, and will pose threat in war.