Does the AI have preknowledge of the map?

[The Condor]

Uh, I just tried the test and I got no results. The Ai didn't leave the island at all (also I have a few graphic errors that are mine, not yours for the funny screenshots).

The destroyer is insane:

 

[PekkaM]

Should this be taken so that using World Builder is fair play from now on?

Really lame in my opinion

 

[The Condor]

What is really lame? This thread or that firaxis lied to us?

 

[PekkaM]

What is really lame? This thread or that firaxis lied to us?

The AI seeing what it shouldn't.

 

[The Condor]

Oh, I agree. Congrats on your one hundredth post.

 

[eg577]

About the civ declaring war after you disbanded units: The AI bases this decision off the RNG and your power rating. The power rating info is available to you too via the info screen chart, although you have to use a ruler/guesstimate to get an actual number.

The findings here really bug me. Even if the AI does get free info (which I think is unacceptable) why of all things does it still get resource locations which, you know, was only the number one complaint about the AI from civ3? To top it off we were flat out lied to about the AI not doing it anymore.

 

[onedreamer]

To above poster (Noble).

Hi, I did both (as good as I could do) and neither are true. The first one I did as Psyringe instructed with Tokogawa (whatever sp) and had one city (the right) be the weakest (two warriors where as the others had a warrior and a archer) and he stayed holed up in his city (with his three axemen and spearman I gave him) till his borders expanded and then he cruised around up to the borders of every city (first capital then second city then doubled back to weak city) and when he found what was the weakest city he attacked.

For the second I used Roosevelt and plopped him down in a barren desert with four land points that shot out in four direction (right best but all were grassplains [right had horses and another resource]) with a spy I created (my settler and warrior I put in a remote island so they would not interfere) and against all previous guesses he plopped his city in the desert. He didn't explore (if he would of moved in any direction just a tad he would of been able to see much better land) or anything. He just created a nine square desert working city of stupidity. That is all and if you feel I am not correct feel free to do it yourself and post your results.

heh, sorry but this does not prove anything of the points raised by the thread. The problem here is that the AI has pre-knowledge of UNREVEALED resources, not of the map. That's quite different. Also, your test would have been 100% better if you gave the AI two settlers, because I have never ever ever seen the AI move and look for a good spot for their first city, in any of the Civ series.

 

[Ice]

The AI recommendation of where to place your cities is definitely influenced by the location of resources even if those resources have not been revealed to you. I've seen evidence of this in every Civ4 game I've played.

Probably just a bug, that will hopefully be fixed sometime. TBF you get the same advantage the AI does out of it. I assume the AI doesn't know *what* resource is there, they just know that tile A is better than tile B even though with their current knowledge B should look better than A.

 

[onedreamer]

yeah as I said before, almost every time I ignored those advices I later had to regret.

 

[Harald]

yeah as I said before, almost every time I ignored those advices I later had to regret.

I thought this as well but if you make a a similar island like in savegame in the attached zip-file it is true that you are given recommendations. But it recommends the corresponding squares on every arm. So the human player doesn't get any info that tells you where the 'hidden' resource is while the AI player allways goes for the arm with the hidden resource. If you move the resource while the settler is in transit he will change direction towards the arm with the resource.

 

[Psyringe]

heh, sorry but this does not prove anything of the points raised by the thread. The problem here is that the AI has pre-knowledge of UNREVEALED resources, not of the map. That's quite different. Also, your test would have been 100% better if you gave the AI two settlers, because I have never ever ever seen the AI move and look for a good spot for their first city, in any of the Civ series.

This has been cleared up yesterday already. The test has been improved in several ways and now shows that the AI does have pre-knowledge of resources. Just read on.

 

[onedreamer]

yeah I had read on after
@Harald I must try that this evening.

 

[Renata]

I don't believe the AI has knowledge of unrevealed resources under most circumstances. The days of AIs sending settlers halfway around the map to claim unseen resources are over. In one of my first games I wound up with a whole swath of tundra and ice to my south that nobody bothered to settle the whole game. When I finally got to see oil, lo and behold, there were two sources down there. The first AI to show up by the oil with a settler was my nearest neighbor, about five turns after I'd sold him the tech to see it.

As for the city-finding algorithm, I suspect it's a mistake arising from the re-use of the map-design code. It is known that resources are taken into account in determining the worth of potential starting sites -- the map designer has said this. So he must have prepared an algorithm that incorporates all of that map knowledge, including things that should be hidden.

So far, so good, but it appears that the *same* algorithm was incorporated into the mechanics of the game itself as a way for the AIs (and players who want hints) to figure out where they should put subsequent cities. So here all of a sudden we have full map knowledge where it does not belong.

But can you jump from that to say "Firaxis is a bunch of liars"? I don't think so. It's too easy to ascribe the error to pure carelessness. Someone writes a piece of code that determines good initial city placements. Someone else needs a piece of code that does the same thing within the game itself, so they just borrow it, without considering the implications. And with Firaxis having been *so* vehement that the map foreknowledge was gone, I can't imagine they would have done it deliberately.

Anyway, I don't think this is a huge issue in practice. Given that the AIs do not any longer send settlers willy-nilly all over in the map in pursuit of resources they can't see, the only advantage they're going to get is by putting cities they were going to found in the area anyway into slightly better spots. And human players can do the same, now that the flaw is known.

 

[dimebolt]

So far, so good, but it appears that the *same* algorithm was incorporated into the mechanics of the game itself as a way for the AIs (and players who want hints) to figure out where they should put subsequent cities. So here all of a sudden we have full map knowledge where it does not belong.

Amen. I was thinking this very same thing while reading the thread. I'm convinced the same algorithm is also used for determining the blue circles (which are probably peaks above a certain threshold in the tile value function). Very often I see that a 'stupid' blue circle recommendation (which I then choose to ignore) actually would have given me an extra resource I had yet to discover.

However, since this function takes into account undiscovered resources, using it is in fact almost the same as knowing the exact locations of all resources. The only real difference is that the AI won't know it doesn't have oil and thus won't found a crap city to get it until the discovery of oil (as in the game Renata played).

 

[Psyringe]

Hmm, so your hypothesis is that the AI does not see any resources *unless* it's trying to build a city, because in that case it makes use of a piece of code that was meant to evaluate starting positions (so that civs get equally fair starting points), and not meant to be accessed by the AI. Even then te AI does not see the resource, it just knows that "something" is there which makes this spot more valuable. (Please correct me if I misunderstood you.)

I can see that this is possible. However, I have two problems with this:

1. If this is a bug (and not planned behavior), then - and I do this for the first time, I have always defended the testers in many previous posts - I wonder how it could slip through. The claim that the AI doesn't have pre-knowledge of resources is and was an important issue for many people, and Firaxis knew that. As you said, they have been quite vehement about map pre-knowledge being gone, and I'd expect an issue with so much importance to many fans to be tested thoroughly. I mean, it's not difficult to test it the way we did here. I devised the basic testing procedure within five minutes, sitting in an internet cafe, where I had something to eat while waiting fo my train home. The Condor and Harald only needed a couple of minutes to perform the test. I think that is something that *could* (and should) easily have been tested before.

That doesn't mean I'm slamming the game, the testers, or the testing process as a whole. I actually love the game. But I do think that for this specific issue, something went wrong in a way it should have been prevented under the circumstances which were known to Firaxis. I also must admit that this test leaves me wondering whether other claims about the AI are true. If a bug slipped through in this (important and easily to test) algorithm slipped through, then there may be other bugs in equally important (and more difficult to test) algorithms as well.

2. It is not true that the player has the same knowledge because of the "blue circles". Harald's second test proves that, and I have modified a previous test and got the same result. The player gets blue circles on all outbound corners of the cross. He has no way to know where the best spot (with the resource) is. The AI knows, it always settles right beneath it. Furthermore, the human player doesn't even see the spots where the blue circles will appear unless he moves his settler towards them. The AI however will head straight to the resource, no matter whether the tile in question is visible to it, or has been explored yet.

So, even though I would *very much* want to be in error here, I don't think our findings can be dismissed so easily.

 

[MrCynical]

Does the AI follow exactly the same blue circle advice that the hints give? If so they have relatively little info which isn't available to the human player.

Calling Firaxis a bunch of liars is just silly. The vast majority of the AI's map knowledge has been removed, as is very clear from a couple of games. The undiscovered resources being left in is a fairly minor issue which seems to have returned by mistake. This issue has been vastly improved since Civ 3, as the AI can no longer beeline to the best sites on the map from turn 1.

As for whether this make worldbuilder "fair play", of course it isn't. The AI has a trivial advantage which is unlikely to survive beyond the next patch. Given the AI has about the 'intelligence' of the average ant you ought to be able to beat it despite this. Cheating in a single player game is always a rather futile exercise.

Still it is good to see this was actually tested to confirm whether an issue exists. It makes a pleasant change from the neverending stream of "OMG the AI cheats!!11!!!1" which are mostly because someone hasn't bothered to check how something is supposed to work. Complaints that the test should have been done by Firaxis seem a little odd however. The AI was programmed not to have knowledge of the map, and if they forgot about undiscovered resources being in the city placement evaluation there would be no reason to assume there was a problem. I doubt the beta testers would have noticed because, simple as the tests in this thread are, there would be no real reason to run them. I doubt they set up a special map to test each rule and element of the game individually. It is far from obvious that the AI still has this ability in a conventional game, as is evident from the fact that it has taken more than two months for it to be proved the issue exists.

 

[Xenocrates]

There could be a difference in the way the AI behaves between the computer generated maps and the test maps.

On the other hand when a civ walks straight past two gems to place a city in an apparently eccentric spot something is amiss. It happens frequently. Dumb AI that sometimes gets lucky with resources or AI that can see under the bonnet? Either way doesn't cut the mustard.

There was something unsaid in Renee's post. Maybe the AI's didn't prioritise getting the tundra oil in that game because they already had a source? Maybe after X cities the AI changes to a different algorithm?

Maybe the AI doesn't feed the resource into the equation only the value of the hammers/gold/food that the tile provides. They probably list good spots in terms of productivity and not acquisition of resources and settle them in order. That's what I would do if I was a programmer without a better idea and believe me; I don't have a better idea!

Maybe it was to much to ask for Civ 4 to break from too-simple mechanical decision making, but ask we did and get we didn't. It appears!

If the civ series wasn't so damn good we wouldn't have expected so much.

The game should include two more modern-era wonders of the World:

1) Civilization and 2) Civfanatics forum.

The first halves worker productivity and the second takes it all the way to zero!

 

[Carewolf]

I think the AI also gets a few infos outside of FoW. It might just be the barbarian code that also affects AI, but Firaxis is not kidding when the in the tips notes that an unescorted settler attracts barbarians. The AI can also tell un-escorted transports far outside their view but inside the range of a destroyer.

Should be tested though.

 

[The Condor]

I could test that. What do you want me to do?

 

[Carewolf]

That brings up one more gripe. Why place two other civs 13 tiles away, and 3 more just the other side of that, on a large continent world. Just how big of a map do you have to create to have a little breating room early on for expansion? (madface). But that is a topic for another thread.

This is certainly not a cheat. I almost accused the computer of doing the opposite after in two new games to have put me on my own continent (with no opponents within galley reach). This is fun to begin with, but less fun when the AI finds you in galleons and attacks your archers with grenadiers (you research faster the more opponents you know).