Do you have evidence of the game "cheating"?

[pokiehl]

I tend to mostly agree with you on a lot of matters, but this statement is a bit short-sighted from my perspective.

1) The people on this forum might be split, so you have a faction that believes the AI cheats, and a faction that thinks the AI sucks.

2) The two statements (AI sucks, AI cheats) are not mutually exclusive. The AI could very well cheat and still suck. In fact, it could be that without the cheats (I'm not saying there are any, but purely for the sake of argument) the AI would suck even more.

In general, most posts discussing how the AI is breaking the game rules or cheating is by someone who is frustrated with the game and feels like they are losing. The other group complaining that the AI is ineffective never discuss cheating as far as I can tell. This is where I drew the distinction.

 

[Deadly Dog]

Two examples I can think of where possibly the AI is playing by different rules than the player come up when there are queues for Pantheon choice and when multiple Great People of one class are recruited in a single turn, then suddenly the cost for that class goes up 150%

If another AI reaches the faith threshold for a Pantheon, the player will have to wait a turn before selecting. Sometimes the wait is 2 turns or even more, presumably as other civs also reach 25 faith. If the human player is somehow always last in a tie or if the player gets bumped to the back of the queue when another AI hits the faith threshold on a later turn, that is certainly a different ruleset. What happens in multiplayer/hotseat may answer this?

The Great Person mechanic I don't really know if it's stacked against the player. Do the AI always recruit in order after the player? If so then it's not different rules as sometimes the last AI before your turn will recruit and leave you first chance at the next one, and it's another AI who misses out before the cost goes up.

 

[intellectsucks]

Here’s an example of the AI doing things the player can’t do (it’s actually pretty perfect). I have a crossbowman on the Mahabodi Temple just north of Rouen. I move the swordsman one southeast to the east side of Rouen, choose my production queues, then heal everyone else in place. The Free Cities Horseman moves onto the marsh west of the referenced crossbowman, then the Free Cities Swordsman attacks my crossbowman, then the Horseman crosses the river to attack and kill my crossbowman. It moves two tiles (one of them being marsh), then crosses a river to attack. On the next turn my Knight with identical movement can’t do the same move (move two tiles, then cross a river to attack the horseman). I’ve posted the relevant saves and screen shots. If anyone can explain this in a way that conforms to the rules, I’m open to hearing it but otherwise this is a clear example of how the AI is NOT bound by the same rules as the player.

 

[intellectsucks]

In general, most posts discussing how the AI is breaking the game rules or cheating is by someone who is frustrated with the game and feels like they are losing. The other group complaining that the AI is ineffective never discuss cheating as far as I can tell. This is where I drew the distinction.

I agree that most posts are of that nature but that is NOT the intent of this post. I'm currently playing on Immortal and doing fairly well, and I consider myself to be a very suboptimal player, so I don't believe that the "cheating" makes the AI unbeatable or even especially difficult for very skilled players. However if I was struggling with the game, and the AI did something that a player is not capable of doing, then I too would be frustrated and it would taint my opinion of the game. The intention of this post is to identify if the AI is able to things the player can't, what things they can do, and when they can do them. If we KNOW how the AI can bend or break the rules then the players can act accordingly.

 

[Deadly Dog]

Here’s an example of the AI doing things the player can’t do (it’s actually pretty perfect). I have a crossbowman on the Mahabodi Temple just north of Rouen. I move the swordsman one southeast to the east side of Rouen, choose my production queues, then heal everyone else in place. The Free Cities Horseman moves onto the marsh west of the referenced crossbowman, then the Free Cities Swordsman attacks my crossbowman, then the Horseman crosses the river to attack and kill my crossbowman. It moves two tiles (one of them being marsh), then crosses a river to attack. On the next turn my Knight with identical movement can’t do the same move (move two tiles, then cross a river to attack the horseman). I’ve posted the relevant saves and screen shots. If anyone can explain this in a way that conforms to the rules, I’m open to hearing it but otherwise this is a clear example of how the AI is NOT bound by the same rules as the player.


View attachment 592046
View attachment 592047

I've had similar situations where I was surprised by an AI being able to attack across a river where I couldn't.

What is the minimum movement needed to cross the river at that point, though, is it 3 points or just a fraction more than 2? And is there any way the horseman could be getting enough movement bonus from the road in its own territory that there is enough? And is it possible that your knight does not benefit from this bonus because its hostile territory?

I'm inclined to believe this is one of those cases though, just thinking where to look to disprove that.

 

[Kwami]

Here’s an example of the AI doing things the player can’t do (it’s actually pretty perfect). I have a crossbowman on the Mahabodi Temple just north of Rouen. I move the swordsman one southeast to the east side of Rouen, choose my production queues, then heal everyone else in place. The Free Cities Horseman moves onto the marsh west of the referenced crossbowman, then the Free Cities Swordsman attacks my crossbowman, then the Horseman crosses the river to attack and kill my crossbowman. It moves two tiles (one of them being marsh), then crosses a river to attack. On the next turn my Knight with identical movement can’t do the same move (move two tiles, then cross a river to attack the horseman). I’ve posted the relevant saves and screen shots. If anyone can explain this in a way that conforms to the rules, I’m open to hearing it but otherwise this is a clear example of how the AI is NOT bound by the same rules as the player.


View attachment 592046
View attachment 592047

Are you sure that the horseman is moving into the marsh, though? If he just moves into the grass tile northwest of Rouen, then he should still have enough movement left to cross the river and kill your crossbow.

 

[Kibikus]

Here’s an example of the AI doing things the player can’t do (it’s actually pretty perfect). I have a crossbowman on the Mahabodi Temple just north of Rouen. I move the swordsman one southeast to the east side of Rouen, choose my production queues, then heal everyone else in place. The Free Cities Horseman moves onto the marsh west of the referenced crossbowman, then the Free Cities Swordsman attacks my crossbowman, then the Horseman crosses the river to attack and kill my crossbowman. It moves two tiles (one of them being marsh), then crosses a river to attack. On the next turn my Knight with identical movement can’t do the same move (move two tiles, then cross a river to attack the horseman). I’ve posted the relevant saves and screen shots. If anyone can explain this in a way that conforms to the rules, I’m open to hearing it but otherwise this is a clear example of how the AI is NOT bound by the same rules as the player.


View attachment 592046
View attachment 592047

My guess is it's because of roads. Ancient roads allow units to ignore features (such as the marsh), and classical roads add bridges over rivers which makes crossing them cost only 1 MP.
As for the knight, he's attached to a siege tower, which only has 2 MP.

 

[Deadly Dog]

My guess is it's because of roads. Ancient roads allow units to ignore features (such as the marsh), and classical roads add bridges over rivers which makes crossing them cost only 1 MP.
As for the knight, he's attached to a siege tower, which only has 2 MP.

Yup that explains the horseman, after looking more closely I can see that there must be a road over the river there even though it's not shown

Knight should be able to cross as well though obvious need to ditch any linked units. And don't trust what the pathfinder is showing you either actually try the move I'd say.

 

[PiR]

Are you sure that the horseman is moving into the marsh, though? If he just moves into the grass tile northwest of Rouen, then he should still have enough movement left to cross the river and kill your crossbow.

In that case he would stack with his fellow swordsman to attack, but because he will win, there won't be any stack, 1UPT is respected and the game allows it.
BUT if the horseman was going to lose, the game would not allow the move, because that would mean stacking.
Does this make any sense? I'm saying that because I feel I experienced that, as a player, with my units.

 

[Deadly Dog]

I loaded the save. Knight can definitely cross the river once de-linked from the siege tower.

Wonders and districts automatically connect to all adjacent roads, including across rivers, though the graphic may not show it.

 

[intellectsucks]

I loaded the save. Knight can definitely cross the river once de-linked from the siege tower.

Wonders and districts automatically connect to all adjacent roads, including across rivers, though the graphic may not show it.

Awesome. Thanks for the analysis. I hope more people post some examples. I had one a while ago where Brazil upgraded to swordsmen with no apparent sources of iron, but I didn't keep the save. If I find any other fishy things I'll post them up.

 

[Planktonic]

I've commented on this elsewhere, but the one thing I am convinced the AI can do that we cannot is apply production that was initiated towards one thing to something else when it decides it wants to shift. That is, if it has invested 20 hammers in producing unit A, but then it decides to produce unit B, it simply applies the 20 hammers to unit B rather than having to start from scratch like we do.

I saw this in real time in a recent GOTM and wish now that I had saved the autosave and the game file at the time. I was bombing the last capital city; I had a spy in place with sufficient visibility that I could see the city's production target and the number of turns remaining. The city in question was producing a carrier (540 hammers required) for a number of turns, then suddenly switched to producing a battleship armada (>> 540 hammers required) right after I initiated my attack. I don't remember the exact number of turns remaining in either case, but I remember that I did a best estimate of the production capability of the city and whether that could account for the low number of turns I was seeing needed for the battleship armada. I thought at the time that it was pretty clear that extant production that had been achieved towards the carrier was now being applied to the armada, otherwise there was no way to explain why such a low number of turns remaining were showing for the armada.

This would also explain two other frequent observations I have made: 1. that cities under assault rapidly raise ancient walls (because I assume they apply stored production towards another target to the walls; otherwise they would have to start from scratch, or pay a fee for switching policies to Limes etc at the least). 2. that when I have (air) pillaged a spaceport with a rocket on the launch pad, showing that a Space Race project was underway at that city, somehow the project is completed in a few turns by some other spaceport city, again presumably because that city switched its production target to the Space Race project and then applied stored production towards it.

The next time I see something like this going on I will try to remember to save the autosave and game files to demonstrate it.

 

[Kwami]

I've commented on this elsewhere, but the one thing I am convinced the AI can do that we cannot is apply production that was initiated towards one thing to something else when it decides it wants to shift. That is, if it has invested 20 hammers in producing unit A, but then it decides to produce unit B, it simply applies the 20 hammers to unit B rather than having to start from scratch like we do.

I saw this in real time in a recent GOTM and wish now that I had saved the autosave and the game file at the time. I was bombing the last capital city; I had a spy in place with sufficient visibility that I could see the city's production target and the number of turns remaining. The city in question was producing a carrier (540 hammers required) for a number of turns, then suddenly switched to producing a battleship armada (>> 540 hammers required) right after I initiated my attack. I don't remember the exact number of turns remaining in either case, but I remember that I did a best estimate of the production capability of the city and whether that could account for the low number of turns I was seeing needed for the battleship armada. I thought at the time that it was pretty clear that extant production that had been achieved towards the carrier was now being applied to the armada, otherwise there was no way to explain why such a low number of turns remaining were showing for the armada.

This would also explain two other frequent observations I have made: 1. that cities under assault rapidly raise ancient walls (because I assume they apply stored production towards another target to the walls; otherwise they would have to start from scratch, or pay a fee for switching policies to Limes etc at the least). 2. that when I have (air) pillaged a spaceport with a rocket on the launch pad, showing that a Space Race project was underway at that city, somehow the project is completed in a few turns by some other spaceport city, again presumably because that city switched its production target to the Space Race project and then applied stored production towards it.

The next time I see something like this going on I will try to remember to save the autosave and game files to demonstrate it.

Again, though, a story without evidence.

One possible explanation is that the AI was using the card that gives a production bonus on all naval units except for carriers.

 

[Kibikus]

I've commented on this elsewhere, but the one thing I am convinced the AI can do that we cannot is apply production that was initiated towards one thing to something else when it decides it wants to shift. That is, if it has invested 20 hammers in producing unit A, but then it decides to produce unit B, it simply applies the 20 hammers to unit B rather than having to start from scratch like we do.

I saw this in real time in a recent GOTM and wish now that I had saved the autosave and the game file at the time. I was bombing the last capital city; I had a spy in place with sufficient visibility that I could see the city's production target and the number of turns remaining. The city in question was producing a carrier (540 hammers required) for a number of turns, then suddenly switched to producing a battleship armada (>> 540 hammers required) right after I initiated my attack. I don't remember the exact number of turns remaining in either case, but I remember that I did a best estimate of the production capability of the city and whether that could account for the low number of turns I was seeing needed for the battleship armada. I thought at the time that it was pretty clear that extant production that had been achieved towards the carrier was now being applied to the armada, otherwise there was no way to explain why such a low number of turns remaining were showing for the armada.

This would also explain two other frequent observations I have made: 1. that cities under assault rapidly raise ancient walls (because I assume they apply stored production towards another target to the walls; otherwise they would have to start from scratch, or pay a fee for switching policies to Limes etc at the least). 2. that when I have (air) pillaged a spaceport with a rocket on the launch pad, showing that a Space Race project was underway at that city, somehow the project is completed in a few turns by some other spaceport city, again presumably because that city switched its production target to the Space Race project and then applied stored production towards it.

The next time I see something like this going on I will try to remember to save the autosave and game files to demonstrate it.

On what difficulty were you playing? Remember that on higher difficulties the AI gets significant bonuses to all yields, which can be further multiplied by stuff like policy cards.

I am not saying that you are necessarily wrong, it would be curious to see if the AI cheats in that way. But what you are suggesting regarding shifting production would fall outside the rules of the game as they are presented to us. So my question would be - how, and why, would that be coded? I understand that would be hard to determine without access to the dll.

 

[PiR]

I am not saying that you are necessarily wrong, it would be curious to see if the AI cheats in that way. But what you are suggesting regarding shifting production would fall outside the rules of the game as they are presented to us. So my question would be - how, and why, would that be coded? I understand that would be hard to determine without access to the dll.

It could be coded as he says but be unintended by the devs.

 

[blackbutterfly]

Does the AI cheat in Zombie Mode?

I've noticed how zombies don't seem to pursue AI settlement in the same way the zombies attack you.
IDK I'm still figuring this out.

 

[PiR]

Does the AI cheat in Zombie Mode?

I've noticed how zombies don't seem to pursue AI settlement in the same way the zombies attack you.
IDK I'm still figuring this out.

"The more you kill the worse they get"
So... Could it be just because AI is unable to kill them as well as the player?

 

[Vandlys]

Awesome. Thanks for the analysis. I hope more people post some examples. I had one a while ago where Brazil upgraded to swordsmen with no apparent sources of iron, but I didn't keep the save. If I find any other fishy things I'll post them up.

Most likely a trade with someone or got it from being Suzerain, no?

 

[intellectsucks]

Most likely a trade with someone or got it from being Suzerain, no?

I'd prefer not to speculate, but seeing things like that, coupled with recent discussions in a couple of other posts are what made me start this thread. The things that I (and others) have noticed could have very simple explanations (like the above) that we're missing OR there could be aspects of the game that are not quite working the way we think. Either way, I'd like to know the details.

 

[Disgustipated]

This might be related to this thread. But I get irritated when my privateer is instantly discovered by the AI. I didn't take any action (no coastal raid or attack). Do zombies see privateers? It's the only explanation how it was discovered the first turn I move next to a land tile I intended to coastal raid the next turn.