I've caught them cheating again. Have you?

[johnmcd]

Originally posted by Dauphin


It is when you don't know that it's strongly defended. If you noted in my post I did say that the AI goes to the city with the least units, regardless of whether it can see what is in the cities in question or not. It is also worth noting that I said nothing about bypassing, geography is not that important.

I'm fighting a war with Mansa in my current game, his key frontier town is on a hill with walls and garrisoned like mad. My preotrians didn't even spit at on their march past. That's going to be the last town I take and it's going to be in quite a while.

My pretorians are marching into fog, I have no idea where his other cities are, but I'm damned if they aren't an easier conquest.

[Urban Ranger]

Originally posted by Dauphin
I've seen many units beeline for the least defended/undefended city rather than the well defended one, even though the quantity of defence is unknown in both, a disproportionate number of times. It can lead to useful honeypotting.

I still think your observation could be biased.

Suppose you have two cities behind the front line somewhere. One is well defended and the other isn't. You would think nothing if an enemy unit comes at your strong city but would pay attention if it comes to your weak city.

So in effect you are registering more "hits" against the weak city than against the strong one.

Alternatively the computer player may have some way of knowing your defenses, e.g. religion.

[Older than Dirt]

If you want to see the computer cheat you can - just believe it most fervently and have faith.

[Badtz Maru]

Originally posted by Dauphin


It is when you don't know that it's strongly defended. If you noted in my post I did say that the AI goes to the city with the least units, regardless of whether it can see what is in the cities in question or not. It is also worth noting that I said nothing about bypassing, geography is not that important.

The AI is aware of all information that would be available to the player, and it takes it into consideration constantly, which the player rarely does. Do you check the relative power levels of all civilizations every turn? The AI does.

Now, this leads to a way the AI could KNOW you have undefended cities somewhere. Most players are different from the AI in that we don't protect cities that we don't see as threatened. Players frequently win wars with armies half the size of the AI because of concentration of force. Now, let's say the AI next to you has 150% the military power that you do. Through spying, scouting, etc., he's aware that in your border cities, you have garrisons twice the size of anything he has protecting his cities.

The AI should be able to put two and two together, and realize that the vast majority of your troops are on the borders. If he can account for 90% of your power level in the cities he sees, he knows that the cities he DOESN'T see must have hardly any garrison. It may not know specifically which cities have one warrior defending them, but he knows they are in there, and will go looking for them.

[Prussia]

The point is, there is a difference between the AI arbitrarily placing 20 units 2 squares from your capital when the actual army is 20 squares from your capital, and when the AI gives itself an extra square of vision in order to compensate and compete somewhat with the player.

The AI HAS to cheat, IN EVERY GAME the AI cheats, otherwise you would be wooping alot of AI ass over, and over, and over again, until it isn't fun.

You want the AI not to cheat? Play any computer game that has ridiculous cheat codes for the player, and use them over, and over, and over again, and you will see that it is not fun when you have a cruel advantage over the AI.

When the AI doesn't cheat, it's like you using cheat codes.

One more thing, one of the reasons Deity is so hard is because of all the disadvantages you get (i.e. hard-to-please population, sickness, etc.).

[Badtz Maru]

Originally posted by Prussia
The point is, there is a difference between the AI arbitrarily placing 20 units 2 squares from your capital when the actual army is 20 squares from your capital, and when the AI gives itself an extra square of vision in order to compensate and compete somewhat with the player.

The AI HAS to cheat, IN EVERY GAME the AI cheats, otherwise you would be wooping alot of AI ass over, and over, and over again, until it isn't fun.

You want the AI not to cheat? Play any computer game that has ridiculous cheat codes for the player, and use them over, and over, and over again, and you will see that it is not fun when you have a cruel advantage over the AI.

When the AI doesn't cheat, it's like you using cheat codes.

One more thing, one of the reasons Deity is so hard is because of all the disadvantages you get (i.e. hard-to-please population, sickness, etc.).

The thing is, the AI in Civ4 does cheat in some ways that are not hidden from the playerbase at all - we know they get advantages over players in production, upgrades, fighting barbarians, etc. Nobody is disputing that.

But there are a lot of people who are convinced that the AI has other ways of cheating that the developers deny exist, such as spontaneously generating defending units, seeing into the fog of war, etc. I haven't seen any evidence of this kind of cheating that could not be explained in other ways. Most of it appears to be observer bias (like "barbarians never attack AI civilizations"). A lot of it is due to players forgetting how much information is available to everyone, AI and human alike, or forgetting aspects of how the AI plays it's turn (i.e. AI unit movement being done so fast that the player only sees an army spontaneously appearing on their shore).

I don't think the developers were lying when they said the AI in Civ4 cannot see into the fog of war - why would they? There have been many times where I thought for a moment that the AI was cheating, but then I thought about it and realized how that was not necessary for what I observed.

[Will9]

From the same war I complened about, I only had two troops per city when the Romans attacked me. They easily took the closest city to border, but once I got a chance to mobilize my army of fifteen units a few squares to the north they stood no chance against my cities and started pillaging. Novgorod was the city they took and later went around to take my woker.

[DrSpike]

Originally posted by Urban Ranger


I still think your observation could be biased.

Suppose you have two cities behind the front line somewhere. One is well defended and the other isn't. You would think nothing if an enemy unit comes at your strong city but would pay attention if it comes to your weak city.

So in effect you are registering more "hits" against the weak city than against the strong one.

Alternatively the computer player may have some way of knowing your defenses, e.g. religion.

It could be biased, and you could be right. But it isn't, and you aren't.

[Urban Ranger]

Spikey, you could at least pretend to add some useful information and/or observation

[Badtz Maru]

Originally posted by Urban Ranger
Spikey, you could at least pretend to add some useful information and/or observation

Cheating AI is a matter of faith, not science.

[snoopy369]

Wow ...

You guys know that they released the SDK, right?

And by "released the SDK" I mean "released the code that tells you about how the AI works", right? How it makes decisions and such?

So you can tell if it's cheating ...

[DrSpike]

Originally posted by Urban Ranger
Spikey, you could at least pretend to add some useful information and/or observation

I've given you the result. As others have posted you can look at the code, or cheat yourself to verify it. Not sure what more there is to add, since if you want neither to listen nor verify yourself there isn't much left that can convince you.

[Jaybe]

Warning: Knowing too much about game mechanics can decrease game enjoyment.

Particularly about the AI, ignorance can be bliss.

[Dauphin]

Originally posted by Urban Ranger


I still think your observation could be biased.

Suppose you have two cities behind the front line somewhere. One is well defended and the other isn't. You would think nothing if an enemy unit comes at your strong city but would pay attention if it comes to your weak city.

So in effect you are registering more "hits" against the weak city than against the strong one.

Alternatively the computer player may have some way of knowing your defenses, e.g. religion.

I'm well aware of biasing and misrepresentation of statistics in the mind. We both have good knowledge of that kind of thing from our mutual educational backgrounds.

I've certainly seen a barbarian axemen to and fro as I moved a unit in and out of each of two cities either side of him (the cities were outside of its viewing range). I suggest you try that scenario and see if you can repeat that result. Maybe it was just a quirk but its the best example I can think of that could be reproduced.

[Badtz Maru]

I now have an anecdote related to cheating to share. I just quit out of a game where I was in a lengthy war with Montezuma. I had one city with no garrison, and another with only one warrior, yet he kept sending troops to two of my cities that had longbowmen and other good defenders.

If he could see into the fog of war, why didn't he go for the cities he could have taken easily?