How can they make the AI smarter?

Just one extremely important thing to keep in mind when suggesting AI fixes:

Because Firaxis went with fully voice-acted leaders, they can't add new lines of dialog without a tremendous amount of difficulty (getting the VAs together, signing new contracts, paying lots of money, etc.). While it'd be great if a patch added "please stop obstructing our maneuvers or die," we likely won't see it, because that means a lot of work for everyone to fix one small problem.

Hence, why it would be better for it to simply be a huge diplomatic hit. Most sensibly, it'd be the same as if you told the AI where to shove it when they apologized for attacking your ally, or a little greater. Invisible, yes, but that's the tradeoff of having otherwise impressive leaderbodies.
 
Just one extremely important thing to keep in mind when suggesting AI fixes:

Because Firaxis went with fully voice-acted leaders, they can't add new lines of dialog without a tremendous amount of difficulty (getting the VAs together, signing new contracts, paying lots of money, etc.). While it'd be great if a patch added "please stop obstructing our maneuvers or die," we likely won't see it, because that means a lot of work for everyone to fix one small problem.

Hence, why it would be better for it to simply be a huge diplomatic hit. Most sensibly, it'd be the same as if you told the AI where to shove it when they apologized for attacking your ally, or a little greater. Invisible, yes, but that's the tradeoff of having otherwise impressive leaderbodies.

:lol: Wooow that's a funny joke right here! :lol:

So you want to tell me that you haven't noticed yet that roughly 70% of AI interaction is speechless? The animated puppets on the screen huff, puff or make some kind of gesture, accompanied by a line or two of text. And that's it. Not to mention that when they actually talk, then nearly every time they say exactly same thing, only the text can vary.

From your post it looks like some people actually think that a lot of work has been put into AI leaderheads :lol:
 
In civ3 where 2 nations couldn't share the same tile, I believe the solution to the 'blocking' situation is for the AI to increasingly get annoyed at you for blocking and declaring war on you eventually. That would work here too.
 
Why are you calling this an AI problem? I don't see a problem here. If the AI wants to declare war on you, then it can easily eliminate your units. If it's not going to declare war, then it can't, so it can't attack the city state that you are protecting. So what?

There are lots of real AI problems, but this seems like a non-issue.
 
Why are you calling this an AI problem? I don't see a problem here. If the AI wants to declare war on you, then it can easily eliminate your units. If it's not going to declare war, then it can't, so it can't attack the city state that you are protecting. So what?

The problem is that the AI doesn't see any issue with the player blocking in a city-state like that. If it happened to a human, they'd interpret it as an act of war (which is essentially is), whereas the AI sees it as you simply moving units.

In other words, the AI could declare war and wipe you - and the city-state - out, but they won't since they don't think anything is weird with your actions. They don't respond correctly to the scenario.
 
What I'd like to hear is how to actually code in these suggestions...
 
Just one extremely important thing to keep in mind when suggesting AI fixes:

Because Firaxis went with fully voice-acted leaders, they can't add new lines of dialog without a tremendous amount of difficulty (getting the VAs together, signing new contracts, paying lots of money, etc.). While it'd be great if a patch added "please stop obstructing our maneuvers or die," we likely won't see it, because that means a lot of work for everyone to fix one small problem.

Hence, why it would be better for it to simply be a huge diplomatic hit. Most sensibly, it'd be the same as if you told the AI where to shove it when they apologized for attacking your ally, or a little greater. Invisible, yes, but that's the tradeoff of having otherwise impressive leaderbodies.

This isn't very important. Most of their lines are not spoken. They only have lines for meeting up at first, being annoyed, going to war, stopping war, suggesting a trade, or something like that. Any new text line would probably just used the "Interesting offer" or "You must stop this" kind of vocal lines.
 
Yes, we all knew inevitably that 1upt would lead to potential for blocking exploits.
But so what?

It's the only answer to what's posted in the OP is all - people should just recognize it's impossible to get rid of blocking. The only possible way to get rid of a problem where players can abuse chokepoints and block other units is to have a system where units do not block each other, and if not, deal with it. This specific situation and some others would again be more "fixed" by encouraging the AI to just DoW the player though, but you'll still have things like blocking off land from enemy settlers and so on. As pointed out by several knowledgeable people here, like SG-17, none of the people in this thread saying "sure, it's so easy to fix it" have actually suggested any clue of how they could code or fix the AI - they just don't recognize what is and is not reasonable, and certainly major overhauls of the AI are not likely to happen from the devs at this point, maybe not even with an expansion pack.
 
This isn't very important. Most of their lines are not spoken. They only have lines for meeting up at first, being annoyed, going to war, stopping war, suggesting a trade, or something like that. Any new text line would probably just used the "Interesting offer" or "You must stop this" kind of vocal lines.

Gotcha. I'll admit to not paying much attention to the two leaders I understand (Washington and Elizabeth, and I've barely seen Liz), and even then, it was mostly "take this trade." Any complex diplomacy screens I"ve had were with characters like Augustus and Montezuma, and my dead languages ain't what they used to be.

It's the only answer to what's posted in the OP is all - people should just recognize it's impossible to get rid of blocking. The only possible way to get rid of a problem where players can abuse chokepoints and block other units is to have a system where units do not block each other, and if not, deal with it. This specific situation and some others would again be more "fixed" by encouraging the AI to just DoW the player though, but you'll still have things like blocking off land from enemy settlers and so on. As pointed out by several knowledgeable people here, like SG-17, none of the people in this thread saying "sure, it's so easy to fix it" have actually suggested any clue of how they could code or fix the AI - they just don't recognize what is and is not reasonable, and certainly major overhauls of the AI are not likely to happen from the devs at this point, maybe not even with an expansion pack.

The AI runs through a potential set of moves per unit.
If it decides on one, and it sees that the player is in its way, it seeks another.
If this repeats more than once within the span of two turns, or forced to give up after one iteration, the AI sends a threat.

Rough, and doesn't account for every conceivable possibility (I didn't think too hard about it), but it's approximately what would happen. Fill in your own specifics beyond this (the usual lying human modifiers, etc). If you want anything more specific, go bother Ed Beach; I'm sure he'll be thrilled to humor your every whim on top of his likely oppressive workload at the moment.
 
If it happened to a human, they'd interpret it as an act of war (which is essentially is)

No, I wouldn't, and it isn't.

E.g., in 2003, the US wanted to attack Iraq through Turkey, but it wasn't an "act of war" by Turkey to deny permission for that, and the US did not, in fact, declare war on Turkey as a result.
 
The AI runs through a potential set of moves per unit.
If it decides on one, and it sees that the player is in its way, it seeks another.
If this repeats more than once within the span of two turns, or forced to give up after one iteration, the AI sends a threat.

Rough, and doesn't account for every conceivable possibility (I didn't think too hard about it), but it's approximately what would happen. Fill in your own specifics beyond this (the usual lying human modifiers, etc). If you want anything more specific, go bother Ed Beach; I'm sure he'll be thrilled to humor your every whim on top of his likely oppressive workload at the moment.

There's a lot of edge cases to be considered. What if you're just peacefully trying to blockade the peninsula your capital is in? Or you're being blocked by someone else and an AI is being blocked by you. For all these cases and the OP's, is it better to punish the player or do nothing?
 
What I'd like to hear is how to actually code in these suggestions...

The AI has to make a decision about where to place each unit before it moves it. It looks at the available tiles, considers its options, and decides where to go.

The screenshot makes it abundantly clear that the AI is repeatedly pathing to the central square with all of those units. It knows where it wants to go. It can detect that it can't get there, or it would make moves that violate 1UPT. You just have to leverage those two things to produce the needed output - the tiles that the AI wants to reach this turn, but cannot due to third party presence in those tiles.

All you have to do is add code that flags and stores the relevant tile/unit values. If a tile/enough tiles are flagged, run the dialogue and move the flagged units en masse if the player capitulates. The best solution is probably to flag the offending units and run the script at the beginning of the AI's next turn. That lets the computer declare if refused and retain the initiative on the attack, and it also lets the computer occupy the tiles if the player capitulates so that units don't have to bounce back to the capital in order to keep the player from blocking again.

...they just don't recognize what is and is not reasonable, and certainly major overhauls of the AI are not likely to happen from the devs at this point, maybe not even with an expansion pack.

The community has shown in the past that it has a fair number of players with the dedication, will and expertise to improve on the Firaxis AI. The devs may not be where a solution comes from.

The conspiracy theorist in me figures that they'll use improved AI to sell expansion packs. Just saying.

EDIT:

What if you're just peacefully trying to blockade the peninsula your capital is in?

A blockade is an act of war, and the AI should treat it as such.
 
No, I wouldn't, and it isn't.

E.g., in 2003, the US wanted to attack Iraq through Turkey, but it wasn't an "act of war" by Turkey to deny permission for that, and the US did not, in fact, declare war on Turkey as a result.

In Civ5 terms, there's a difference between "don't use my land for troop movements" and planting Warriors at places where the AI can't attack through outside your borders.

For a real-world example of what's going on, consider if Athens declared war on Sparta, and Cornith decides to put its entire army on the Isthmus of Cornith in an effort to "peacefully" block the Athenian march. You'd better believe Athens is taking that as war.

There's a lot of edge cases to be considered. What if you're just peacefully trying to blockade the peninsula your capital is in? Or you're being blocked by someone else and an AI is being blocked by you. For all these cases and the OP's, is it better to punish the player or do nothing?

Like I said, it's not a perfect approximation, but it's more or less how the AI would pick up on interference. Maybe require that it takes two or three turns to register, in case of accidental maneuvers/marches bumping into each other - the last thing anyone wants is a war over sloppy marches.

Regardless, there needs to be some penalty for deliberate interference. Civ5 is supposed to be less micromanage-y than Civ4; nobody wants games to come down to placing Warriors to block armored columns from beating up city-states.
 
No, I wouldn't, and it isn't.

E.g., in 2003, the US wanted to attack Iraq through Turkey, but it wasn't an "act of war" by Turkey to deny permission for that, and the US did not, in fact, declare war on Turkey as a result.

Blockades in general are treated as acts of war. See "Cuban Missile Crisis."

If the AI has a diplomatic or militaristic objective to achieve involving a city-state and either an AI or a human player is blocking those objectives, there has to be diplomatic consequences. Maybe not an immediate DoW, but something along the lines of "sanctions" at very least.

Programming (and I am not claiming usage of any language personally) does not seem like it would be that difficult... here's a short example.

10 Do I have a military objective? If yes go to 20, if no go to 30
20 Am I being blocked from achieving this objective by a separate actor? If yes, go to 40 if no go to 50.
30 Proceed as normal. End.
40 Send "Diplomatic Request" for player/AI to stand down from military objective. If refused go to 50, If accepted go to 60.
50 Process Human/AI Act of War/Military Alliance and act accordingly. End.
60 (Send AI/Human units out of way of objective 2 hexes) Proceed with military objective.

Obviously you would need to get a lot more complicated to really make it a viable system with several options, and non-exploitable with teleportation, but there is a base idea of how this could be approached, and I don't claim to be even close to a professional.
 
Just because one state commits an act of war against another does not imply that the victim prefers to escalate the situation to full-scale war. The standard explanation of the outcome of the Cuban Missile Crisis is that Khrushchev did not desire full-scale war, and was unwilling to tolerate the risks of escalating to full-scale war associated with challenging the blockade. So he backed down.

If you care, Schelling (1966) is the reference, and that work directly contributed to his becoming a Nobel laureate a few years ago.
 
Just because one state commits an act of war against another does not imply that the victim prefers to escalate the situation to full-scale war. The standard explanation of the outcome of the Cuban Missile Crisis is that Khrushchev did not desire full-scale war, and was unwilling to tolerate the risks of escalating to full-scale war associated with challenging the blockade. So he backed down.

Right. People back down a lot, because war is so destructive. So why are some people here demanding that the AI must declare war if blocked from where it wants to go, even when it doesn't wish to be at war with you? It just seems bizarre.
 
Similar to the OP, you can blockade AI cities to prevent them from getting Settlers out. Also, assuming you can see, for example, Oil before they can, you can move extraneous Workers on top of their Oil resources to prevent them from ever improving them since they can't move onto those tiles.

Any solution which deals with the problem in the OP should probably try to recognize these problems as well.
 
So why are some people here demanding that the AI must declare war if blocked from where it wants to go, even when it doesn't wish to be at war with you?

If you're concerned about that, you can simply add a power check to the calculation of whether or not the AI issues the demand.

But if you're executing the strategy in question, it's because you're weak. If you were strong, you'd just kill the units. Free influence is free. You get XP for slaughtering units, too.

Any solution which deals with the problem in the OP should probably try to recognize these problems as well.

The nice thing about a demand mechanic is that any instance of blocking will be addressed by the AI. Tolerances can be set however high the devs feel is appropriate. The AI might wait until it flagged two or three hexes in a turn or closely spaced turns before issuing a demand.
 
They need to fix the AI because we paid them $50-$60 for a game that is a turn based startegy game with horrible AI strategy and a ton of easy to fix bugs taht shouldn't have been there past beta.

AI Sucks and only gets decent at higher levels where it also gets research bonuses and productions bonuses....WTH

Bugs for diplomacy
-Permanent peace treatys at random ie...I declare war, get a favorable peace treaty at turn 100. Turn 200 i am still in a ten turn peace treaty. Happens randomly and ruins games.
Un-explained Diplomacy System
-What do i get from a pact of cooperation?
-What do i get from a Pact of secrecy?
-If I don't know and it appears that it doesn't effect anything then why am I going to agree to it. This isn't a cell phone contract. As players, we need to know how these choices effect us or we are going to ignore these features. If we ignore these features are they even features?

Paid Beta testers
-Some of these guys need to be fired OR there supervisors need to be fired. Someone obviously dropped the ball. NDA or not I know for a fact that most people who beta tested do not think that the released game AI is the same AI they play tested. Most likly they mass produced the CDs. Sent out the pre-load files to steam. Got good user feed back from the New AI and never actually updated the old AI that was on steam and CDs. I would think they are now sitting around trying to figure out how they are going to completly remove and reinstall the actual game to steam easily and affordably.
Just saying this hypothetically of course. I can neither confirm or deny this information. but I will tell you this....I have made a copy of this post in case someone decides this post doesn't need to be here and deletes it I can come on and post it again.
 
Back
Top Bottom