Can the Civ4 AI see through the FoW?

[Harovan]

Some of our beta testers said no, it does not use informations it should not have. Unfortunately there are signs, that this assumption is false.

You surely remember all the lovely games we played in Civ3, when we were at war with a Civ. Have a city undefended and see attacking units beeline for it. Move defenders in this and out of another city, and see the attacking army change direction (both cities out of sight, of course).

This trick still works. In my last game I was at war with Arabia and since I did not have a navy worth to talk of, Saladin sent ships to pillage my fishnets with a fleet of 2 frigates. The AI loves to do that, and it is good so. When he was done with my 3 resources, he anchored his fleet a bit away from my land, but still in sight of my cultural borders, so I could see it without being in danger of losing units. He stood there not moving for about 5-6 turns.

Meanwhile I built some ironclads in productive coastal cities and in the cities with the pillaged nets some workboats. Except one, these cities were far out of sight ( at least 3-4 turns range for the frigates). As soon as the first net was built in a far away city, the fleet set out in the direction of the new fishnet. I had believed in the "AI does not cheat" statement and was a bit surprised. By the way, Saladin does not hold holy cities of my cities religions - just in case anyone asks.

I had an ironclad moving in the direction to the net, but since they are so slow, the frigates won the race by 1 turn and again pillaged the net. In the next turn my ironclad arrived too and sank one frigate, but remained heavily damaged. In the next turn the other frigate sank my ironclad and remained healing for several turns.

Meanwhile, a city on the other side of the continent built another workboat. As soon as the fishnet was built, the frigate again started moving towards it, still not at full health. This time my new ironclad was faster and arrived at the resource ~2 turns before the frigate. In the next turn, the wounded frigate stopped and continued healing, as it wouldn't attack an ironclad at full health - mind you the ironclad was far out of sight!

In the same game, but earlier I had a race of my settler/defender stack to a city site with a chinese galley. The galley was 3 turns from the city site (far out of sight) when I built the city. It could not see it (I don't see AI cities founded in the FoW either), but in the next turn it did a 180 degree turn and headed home.

It could be a series of coincidences. But something (call it "Civ3 experience") makes me not believe in such coincidences. I'm far from ranting and it's not a biggie, since I can exploit it and hence countercheat, but I am still a bit disappointed.

Does anyone have similar experiences? Is the statement "The AI does not use informations it should not have" still true?

[DeepO]

SR, do you have a save?

And yes, I thought it to be true... Don't know what is happening to your fishing nets.

As to the settling party turning around: When you build a city, you make all tiles at distance 2 unsettlable. The AI detects that, yes, but so can you: the blue spots disappear, and I thought if you hold a settler over a spot too close to another city, the 'build city' icon doesn't light up.

DeepO

[vmxa1]

Remember that if you have the founding of a religion and they have adopted that religion inn a city, you can see into that city.

It may require a particular religious structure, I forget, but that is one way to know what is in a city.

[Jcg316]

After I'd played about three games I started one with my city having six crabs, plus lots of water and not too much else near tundra even.

Up to this point I'd never even seen a barbarian galley and I haven't seen one since in a lot of games even with a couple archipelago maps. When I built up the sixth crab, which was quite early since I figured with all the crabs plus bad terrain I could get the city flying just by building work boats and a defender before building workers and settlers.

At this point I finish building the sixth and a barb galley shows up, destroys the crabs, and leaves. I build a few back then start to build a galley and as I got three up and start to build the galley and its back, one tile out of the black unknown that exact turn. It destroys the crab boats again but my galley scares it off. I scout where it came from to find ocean and coast that goes past two other civs with work boats out, could have been built later or built back though.

I go back and scout the other way eventually and it comes back when I'm six to eigtht tiles the other way. I chase it off before it does much and just sentry my galley on top of the work boats and the barb galley doesn't come back.

I started again from the initial save and built a work boat then a galley and it never came. As this was one of my first games, plus I was a little frustrated and thought the galleys might be common I didn't think to look at it in world builder or anything.

In many games after I haven't seen a barb galley and it really makes me wonder if they can see ocean tiles or some of the ocean tiles as it would explain why I never find them in archipelago maps as they could avoid my scouting galley units while just wanting to pillage lots of stuff. Never have had any work boat scout destroyed either, the only do scouting while waiting for pop to be high enough that I need the work boat though.

Other times with a handful of ocean resources I haven't had problems but usually for that many it is a tiny island map or something and I have galleys out right away as well for transporting settlers and the like.

Could be just coincidence and bad luck and the AI some times just patrols the coast looking to pillage but they do come back sometimes when they shouldn't know you rebuilt a work boat. 99% certain I didn't have raging barbarians on. Certainly keep an eye out and see if I can find a case where it happens, load an autosave then save the game a turn or two before.

[Harovan]

vmxa1: I already said in the OP that the religion theory is wrong. That was the first thing I checked. You can see in the save, that the cities in question are buddhist, the holy city of which is in chinese hands, not Arab.

DeepO: Here is a save, which I reconstructed out of autosaves. It does not contain what I described, just a similar incident. I rarely save my games when I play, and what I described is not a single incident, but a series over 20-25 turns. I just found the AI behavior suspicious in hindsight.

Test it: There is a work boat near Yue-Chi (captured barbarian city). When you build the fishnets, the fleet near Düsseldorf will set out and pillage it. If you don't build the net, it stays there.

Btw, don't criticize me for what I build, when I reconstructed I often just hit the default.

[ulyanov]

I had Alexander and Catherine declare war on me the turn they landed next to 2 seperate sites with oil that they could not have known about. They headed direct to it in both cases

I've had Alexander bombarding a city of mine, then break off to destroy a crab pot that was near another city.

I've given up the game, but I still have some saves if we're doing that sort of thing...

[SpencerH]

Re: Can the Civ4 AI see through the FoW?

Originally posted by Sir Ralph
Some of our beta testers said no, it does not use informations it should not have. Unfortunately there are signs, that this assumption is false.

You surely remember all the lovely games we played in Civ3, when we were at war with a Civ. Have a city undefended and see attacking units beeline for it. Move defenders in this and out of another city, and see the attacking army change direction (both cities out of sight, of course).

I played my first game yesterday. I forgot about a city on an island miles from my opponent. Out of the blue I get a message that "Liverpool was destroyed". Yeah, the AI knows stuff it shouldnt.

BTW does anyone know if there is a page that willl let us scan through the cities for units. I like the group moving etc but I've accidentaly left a couple of cites undefended cos of it.

[DeepO]

SR,

I'll ask what the deal is, but don't expect an answer very soon. Firaxis is busy atm... but maybe one of the other testers can help.

Given the reports, I wouldn't be surprised if the same happened as in Civ3: the AI needed to see the resources or otherwise simply was not able to cope with them. The pillaging is effective in CIV (which it never was in Civ3), so the system has improved a lot... but maybe Soren needed to give the AI a 'cheat' in order to let it function.

DeepO

[Peets]

Originally posted by ulyanov
I had Alexander and Catherine declare war on me the turn they landed next to 2 seperate sites with oil that they could not have known about. They headed direct to it in both cases

I've had Alexander bombarding a city of mine, then break off to destroy a crab pot that was near another city.

I've given up the game, but I still have some saves if we're doing that sort of thing...

This isn't a bug!
Everyone can see where all the resources are on the map.
I just discovered that I could see where oil is and when I use global thingy, I can see every oil resource on the map. (I almost have the entire map).

[Arrian]

In response to the OP: hrm, sure sounds like it!

-Arrian

[Harovan]

Originally posted by DeepO
Given the reports, I wouldn't be surprised if the same happened as in Civ3: the AI needed to see the resources or otherwise simply was not able to cope with them. The pillaging is effective in CIV (which it never was in Civ3), so the system has improved a lot... but maybe Soren needed to give the AI a 'cheat' in order to let it function.

I don't have a problem with it, and did not have in Civ3 either. However in this case the statement "The Civ4 AI does not cheat. Period" can no longer be used.

I had a similar incident as the galley thing (see OP) in a subsequent game. This time I had closed a small isthmus with my very first (chop fed) settler and since I kept my borders closed I am VERY SURE, that the AI could not have seen the map on my side of it, not only by ships but also by wandering scouts. After I controlled all tiles on my side except of 3 (which however had an iron), I agreed to open borders. A few turns later an AI galley with a settler and an archer headed for exactly that spot.

I am pretty sure now, the AI knows the whole map just like in Civ3.

EDIT: Oh and btw, the AI did not need that iron, it had 2 sources on its own half.

[DeepO]

Originally posted by Sir Ralph
EDIT: Oh and btw, the AI did not need that iron, it had 2 sources on its own half.

Keeping you from getting another iron looks like a very good decision to me, even if it has 2 already

I don't know, SR. The problem with this is that you can abuse it, exploit it. AIs should not always go to the same spot, it's too easy to trap them there. At least you're not able to have them dance around your near-blocking warrior wall, but setting up such traps (like e.g. your only fishing net right next to your fleet, so that the AI will come for the fishermen and get destroyed by your subs) is still bad.

BTW, who claimed the AI didn't cheat anymore? Soren? That's in this case the only source you can trust.

DeepO

[Petek]

Originally posted by DeepO

BTW, who claimed the AI didn't cheat anymore? Soren? That's in this case the only source you can trust.

DeepO

FWIW, here's a quote from Solver:

"The AI does NOT cheat. Seriously. On higher difficulties, it gets some bonuses like cheaper research, production, etc., but it has no rule cheats. It doesn't know where resources are in advance, etc."

For full context, the quote is from this thread:

http://apolyton.net/forums/showthread.php?s=&postid=4064509&highlight=cheat#post4064509

[DeepO]

Yeah, well, I trust Solver in most of his information, and he knows more about certain areas of CIV than I do. However, Soren is the only one knowing the AI... it could very well have been changed in between Solver hearing it and final release.

DeepO

[vee4473]

Originally posted by Sir Ralph
vmxa1: I already said in the OP that the religion theory is wrong. That was the first thing I checked. You can see in the save, that the cities in question are buddhist, the holy city of which is in chinese hands, not Arab.

This is a BIG guess, but do allied civs share this type of info? Were they allied?

In other words, if the Arabs were allied with China, could it be that that allows the Arabs to see whatever info China has as a result of having the Buddhist holy city?

just a theory.