Civ AI sucks

[Winston Hughes]

@lilnev

Fascinating read. I wish Firaxis could hire you to try out your experiments, with unlimited funds and access to all the world's leading AI specialists, if only to cut down the number of threads entitled "the AI cheats!!"

One thing, though...

Wouldn't it be more fun if each AI civ was slightly flawed in some way, causing it to make occasional mistakes which the player could exploit? I don't mean it should be too stupid to pick the best choice in any given case, but that its psychology should lead it to make bad decisions. Hence, Monty attacks even when the odds are against him, Musa trades away techs he really ought to hold on to, Izzy gets in a strop with anyone who doesn't share her religion, and Toku just digs himself a hole.

Apart from the whole flavour thing, having different personalities in the game makes for a richer strategic experience, as the player strives to exploit his rivals' psychological weaknesses.

And if the AI could identify patterns in a human player's actions from game to game, and then exploit that player's own weaknesses, it would take civ to a whole new level, where innovation and unpredictable behaviour were required just to maintain your level, let alone advance to higher difficulties.

 

[absimiliard]

What makes you say this? There's only been one post in the BetterAI forum since Oct 3 of last year. I haven't seen any other indications of activity for BetterAI either.

Bh

I wish I could recall. I thought he'd done some modifications post BtS that had tweaked the AI a bit more. Or maybe it was that I thought BtS hadn't included everything he did. It was something like that. Because wasn't city placement one of the things he changed again after BtS had come out? (I suppose it's also possible I'm confusing BtS, Blake's early better AI mod, the patch that came out w. some of Blake's stuff, and ohhhh doubtless other things too.)

You would be way more involved in such stuff than I am though. So doubtless I'm wrong.

Sorry.

-abs

 

[phurph]

I want an AI that kills of 4 other civs, that axe rushes me, that kills my unescorted settlers, etc.

Play 'raging barbs' turned on.

 

[hiddie]

Play 'raging barbs' turned on.

too easy, I usually play standard maps 12 ai's, or large maps 16 ai's, with aggressive AI and Raging Barbs

 

[aelf]

too easy, I usually play standard maps 12 ai's, or large maps 16 ai's, with aggressive AI and Raging Barbs

More AIs does not a harder game make. In fact, it has the opposite effect.

 

[Refar]

More AIs does not a harder game make. In fact, it has the opposite effect.

That depends. If you play warmongering looking for victims, you are probably right. But more AI also means messier diplomathy, faster overall tech speed, higher chances to get different AI personalities to tech along different paths - reducing trate oportunities and making beelines/wonder races tighter.

 

[Florian]

Aelf,

In my experience, more AIs makes the game harder. But I know you have way more experience than I do, so could you explain why? Might this, as Refar says, be dependent on one's play-style?

 

[toug]

I feel as though the game can be as hard as you make it. If you constantly use the same leader, map type and strategy, adjusting to the predictable AI and head starts they get can be no problem.

However, if you start using leaders and maps that make you feel uncomfortable, maybe you can start having fun again.

The regularity with which concepts are almost hammered into everyone's head on these forums is really my only complaint with them. I feel as though I've lost a tiny bit of creativity in my play by subscribing here, so I try and mix it up as much as I can.

 

[Willem]

Strange. considering chess programs, which have been around since computers started, had levels of difficulty 20 years ago that did not involve giving the computer an extra Pawn, or even the first move.

How many tiles on a chess board compared to a Civ map? How many different pieces in chess compared to a Civ game? Does a computer chess player have to contend with what buildings to build, whether to focus on culture or military units? Does it have to worry about any other neighbours other than the human? Is it concerned about trading and how to get the most out of it's resources? Does it need to set priorities as to what tile gets worked and what to put there? I could go and on but I think I've made my point. It's ludicrous to compare chess and Civ since Civ is a much more complicated game by far, with many more variables that the AI has to consider.

 

[Winston Hughes]

The regularity with which concepts are almost hammered into everyone's head on these forums is really my only complaint with them. I feel as though I've lost a tiny bit of creativity in my play by subscribing here, so I try and mix it up as much as I can.

That's the exact opposite of my experience. The amount of disagreement, often amongst very talented and experienced players, over what works best in what situations, along with the plethora of different strategies reported in great detail, leads me to adopt radically different strategies from game to game, mixing and matching from other people's ideas whilst adding my own innovations.

Still, if the stuff you find here does pull you towards the same path in every game, then go take a look at the Succession Games forum. Find a game with variant rules which will prevent you from taking your usual approach, adopt those rules yourself, and see if you can win without reading what the SG team did.

More AIs does not a harder game make. In fact, it has the opposite effect.

I'd say putting more AIs on a map makes the game harder to play, but easier to win.

This might sound contradictory, but what I mean is that those games tend to require a lot more brain work (esp. where diplomacy is concerned), even though there's considerably less chance of losing to an AI (because they generally need a lot of land to stand any chance of beating a human player).

How many tiles on a chess board compared to a Civ map? How many different pieces in chess compared to a Civ game? Does a computer chess player have to contend with what buildings to build, whether to focus on culture or military units? Does it have to worry about any other neighbours other than the human? Is it concerned about trading and how to get the most out of it's resources? Does it need to set priorities as to what tile gets worked and what to put there? I could go and on but I think I've made my point. It's ludicrous to compare chess and Civ since Civ is a much more complicated game by far, with many more variables that the AI has to consider.

I'm afraid you've shown up rather late for that discussion.

Not necromancy late, but late nevertheless.

 

[SnowlyWhite]

1. let's start with an assertion: I assume that ~97-98% of the ppl. playing a game aren't "gamers". By that, I mean they play the game 4-6h/week and, most important, don't think about strategies and such outside their gaming time, and even then probably between wife/kids/homeworks and such.

From this, comes the following: any marketing dude will tell you it's bad to present yourself to such a customer base with a game that permanently beats them. Obviously, you need to make some very hard levels in order to keep your "hardcore" base satisfied too(those, while being utterly in minority, are the ppl. which will provide you with "word of mouth", fan base sites and so on). However, the game must be winnable; and pretty easy. Or you don't sell; ok, more like you won't sell your next title, since those non gamers won't read in depth about your game before buying it.

2. comparing the budget of deep blue with the civ's is ridiculous

3. chess is easier to code; leaving aside you have an army of chess experts, while you'll never have an army of <your unreleased game> experts, at least, because... hey, it's unreleased(and well, you kinda have to provide the game with an ai when you ship it, not after 1 year )

4. the approach per se is poor; it's like the ai keep on building that(which is not last, not that'd matter anymore in bts) piece of spaceship while you're knocking at his last city door with your SoD. You don't invest(and in order to, supposing it'd be possible, you'd have to invest a ton) in something that doesn't produce you benefits. Your customers don't want it, your shareholders don't want it, so why the heck would you do it?

and trust me, after 10 years of playing a ton of games, I still dream of a good ai. However, it won't happen for sure, at least because budget makers have more brains then civ's ai(even if slightly).

 

[hiddie]

I'm not saying I'm not having fun in the game anymore. I still like playing civ.

I'm was just wondering why the AI is so poor. Those are two completely different things. But I've got my answers now.

I'm not the type of player that uses the same leader, same things. I usually play with everything random.

 

[aelf]

As Winston suggested, more AIs means the AIs have less room to expand, causing their growth to be stunted. This in turn makes them less capable of developing their economies as well as less productive, which means less advanced and fewer units and fewer total improvements on top of fewer cities. While the increased number of AIs might make up for the quantity of units, not that many AIs are generally going to be hostile to you at the same time. Instead, they are more likely to be hostile to each other and waste hammers and wealth fighting each other in the usual less-than-efficient ways. The human player, capable of focusing production and applying good economic development strategies, will more easily surpass the weakened AIs, whether one is playing builder or warmonger.

And it's not true that diplomacy will be messier. In fact, chances are the AIs will coalesce into different blocs, making it easier for the human player to identify potential allies and play divide and conquer. This compared to the greater likelihood of one super AI bloc v.s. a few isolated pariahs when there are fewer AIs.

As for the tech pace, the AIs generally follow the same research strategies until later in the game, and by then the human player could have easily surpassed the weakened AIs. And so what if the tech pace is faster? The human player is likely to benefit from that too. If techs are founded a few hundred years earlier, it doesn't mean that the game is harder.

Thus, as long as you are playing optimally, meaning you are applying good strategies and sense and are not deliberately slowing your own growth, more AIs makes for an easier game. I add more AIs when I want to play purely for fun.

Some people, and not just the OP, have the wrong conception of how stupid the AI is because they are overcrowding their maps. Some of them have even insisted on it to the point of flaming anyone who disagrees, claiming that the game is too easy and that they have mastered the game at a certain respectable difficulty level so their opinion must be right. Well, the recommended number of players for each map size is there for a reason.

 

[Bhruic]

Some people, and not just the OP, have the wrong conception of how stupid the AI is because they are overcrowding their maps.

Ok, so you spend a few paragraphs talking about how the AI does such a bad job when it's crowded, and then sum up by claiming the AI isn't really that bad? If you need to give the AI plenty of space, and don't want it to deal with lots of diplomacy, that's not really a convincing argument for the robustness of the AI.

Bh

 

[VirusMonster]

I played against 17 civs on a huge map on Immortal difficulty... I knew it should get easier over time, and it did AFTER I survived the BC era inital warmongering phase.

But the start was very difficult, definitely much more difficult than less number of opponents. Fighting vs 4AIs in the 2000-1000BC aint so easy, 2 of them being Genghis and Monty... you can check the writeup in my signature..

Today, I tried several games at deity difficulty... I again tried to make the map most populated possible by the HOF rules... I thought since it worked at Immortal, it should work at Deity as well... result?

well, the AI expanded everywhere so quickly that I could not build enough cities to produce any significant force.

Then, I reduced the number of opponents(lowest allowed by HOF rules) and was able to expand much more successfully.

so, obviously there is an optimum number of civs to play against at certain difficulties, and it varies from difficulty to another. I don't agree with cliche comments that overcrowding is always easier for the human player.

 

[vicawoo]

Well, they could always spawn incan and native american close to you and kill you. They could all be potential backstabbers, the way humans are.

One big difference between chess and civ is that actions converge more in civ. Large scale strategies don't change that much, and game play is often in the details, and that is what the computer has problems with. Also, although it is complex, it is still purposefully simplified to make it more palatable to human players, hence the converging factor.

And there's diplomacy. They could either try to win by treating computer and human players the same, and then a good computer is in the strange position of playing against itself. Is each civ trying to win at all costs, or is it trying to beat the human? It could treat them differently, which allows odd conspiring, say a there's a shaka civ that kills off civs around it and moves its units out of its cities to let mansa musa take all his land. Oh, and if the human has better odds of winning than any of the AI, all nearby AIs should dogpile deviant (human) civilizations to boost its odds of winning.

And losing AI should just start collaborating like human teammates. Point is, to make it enjoyable, good humans must always have an advantage in diplomacy (otherwise your odds of winning would be about 1/# of civs). Therefore AI need an advantage in the rest of the game.

 

[Iranon]

Are there any special restrictions on AI behaviour to ensure there are no unwinnable games for the humans? Even with Aggressive AI, they usually leave you well alone until you had more than ample opportunity to build up.

I assume most players - especially casual ones - would prefer to fight against AI bonuses at higher levels rather than to cope with a truly opportunistic AI (see the complaints about Barbarian uprisings... similarly, a close neighbour bent on rushing with copper in the capital might be impossible to defend against).
On the other hand, it would be neat if we could enable the AI to ruthlessly play to win, complete with ancient warfare.

 

[absimiliard]

2. comparing the budget of deep blue with the civ's is ridiculous

LOLZ!!!!!

Pardon me while I go put on my LOL-erskates and laugh all the way to work!

Damn true mind you, but still very funny.

For what we have, I'm fairly pleased with the AI. It's levels above what it used to be, and adding in Blakes BetterAI makes it better yet again. In previous incarnations you'd never see an AI muster up an SOD and then come trounce you with it, well, maybe in SMAC, but not in a previous Civ.

City placement is better now. I no longer feel compelled to raze EVERY non-holy non-capitol, and that's distinctly new.

All in all I think our current AI does as fine a job as we should expect from the difficulty of the task and the resources available.

But damn that reason #2 made me laugh.

-abs

 

[Winston Hughes]

so, obviously there is an optimum number of civs to play against at certain difficulties, and it varies from difficulty to another.

That's an interesting pov, and since I have no experience on Deity of anything other than getting whupped I'll have to defer to your greater experience there.

On Immortal, though, I definitely find it much easier with more civs, and much harder with less (I can usually win with more, I've always lost with less). I've never played against less than half the recommended amount, though, so it may well be that there's a lower limit beyond which the AI will be hamstrung.

We should bear in mind, however, that changing the # of civs changes the nature of the challenge faced by the player, meaning that some strategies will be more effective than usual, while others will be less so. It might be that my experiences merely show that my playing style (which is always at least partly unconscious, and thus somewhat resistant to change, regardless of the circumstances) is more effective on crowded maps.

I assume most players - especially casual ones - would prefer to fight against AI bonuses at higher levels rather than to cope with a truly opportunistic AI (see the complaints about Barbarian uprisings... similarly, a close neighbour bent on rushing with copper in the capital might be impossible to defend against).
On the other hand, it would be neat if we could enable the AI to ruthlessly play to win, complete with ancient warfare.

Both very good points.

Having Normal and Aggressive AI in BtS* is a step in the right direction, imo. Perhaps in a future version of civ we'll be able to tailor the AI settings to our preference in the same way we can with map and rule settings at the moment.

*in Vanilla and Warlords, AggAI was really just anti-human AI, and no fun at all imo.

 

[SnowlyWhite]

tbh, the reason I play this game is because I find it to have the best ai from... I think from everything I played so far. Also, the ai is still poor; I mean, equal starts(noble)... don't think I've played even the 1st game on noble nor do I think I'd want a game on noble.

and to(don't remember who, and too lazy to check) who said that raging barbarians and aggresive ai don't help; try play random personalities. Found it way more challenging then aggresive ai(raging barbarians is just more free xp...) especially if you're the type that exploits diplomacy to the maximum.