AI cheat?

[alvan]

Yea, now when I see back on it, it prolly was a chop. Only availble explination (or the ai cheated!)

 

[Soneji]

A chop before BW?

I don't know of any civ that starts with that, and so if it is a chop then thats clearly not right.

 

[Spearthrower]

The AI doesn't cheat.

If the AI was given a free warrior on declaration.... can you show me the code for that please?

I've never seen it!

 

[Soneji]

I know he does, but when you've just watched that one walk westwards past your city.. it seems a coincidence that the AI manages to pull off a warrior build everytime I go to warrior rush..

I had a save a while back, I'll need to go trawl it out (and hope its from the recent patch)

 

[Spearthrower]

It's easy to check, you can reload the last turn and open the WB to see where the city in question is in its building of a warrior.

Coincidence and anecdotes are interesting.... but if someone really wants to prove this, I think code is needed. If this really is a function in the game, that an early (only?) declaration of war against an AI will run a piece of code that inserts a warrior into the threatened city.... then I want to see it.... especially as the few times I have been in this position, it hasn't happened in my game.

 

[Soneji]

It does seem to be that to me, however I would not know where to start looking for the coding.

It make sense for the devs to put that in though, as I have never seen the AI opportunistically do the same to me when its had plenty chances too

 

[Sycraft]

Yep, as mentioned the real solution is to check on world builder. Remember you can pop in and out as much as you like. It may spoil some of the game for you, but if you are checking for cheats then I'd think that would be better than wondering. If you suspect something is up, hop in to world builder and see what the situation actually is. You can then exit out and continue playing.

 

[Psyringe]

I don't think having the actual code is necessary, although it would certainly by a proof, and AFAIK the AI code is available in the SDK. However, not everyone is a programmer and able to read it.

A savegame that reproduces the alleged "cheat" would work just as well as a proof, and wouldn't need programming skills to produce. However, as I said, in all the time since Civ4 came out, no one has been able to produce a save that proves such cheating. The closest thing we had was that we found a bug that would allow the AI to "see" were future resources would pop up due to the tile bonus leaking through when it calculated the best position for a new city. This bug was recognized and fixed by Soren very quickly.

And since no one so far has produced a save that shows the AI cheating, and I haven't seen any evidence of cheating in all my games, I'm very reluctant to believe in such claims without any proof.

 

[azzaman333]

It's easy to check, you can reload the last turn and open the WB to see where the city in question is in its building of a warrior.

Coincidence and anecdotes are interesting.... but if someone really wants to prove this, I think code is needed. If this really is a function in the game, that an early (only?) declaration of war against an AI will run a piece of code that inserts a warrior into the threatened city.... then I want to see it.... especially as the few times I have been in this position, it hasn't happened in my game.

Yep, as mentioned the real solution is to check on world builder. Remember you can pop in and out as much as you like. It may spoil some of the game for you, but if you are checking for cheats then I'd think that would be better than wondering. If you suspect something is up, hop in to world builder and see what the situation actually is. You can then exit out and continue playing.

The only time I have noticed it is when playing a game to submit to the HoF which requires Lock Modified Assests on. And I'd rather play than do WB tests.

 

[Spearthrower]

Anecdotes that can't be proven!

You have to admit Azzaman333, it's hard to accept such things as true without any facts to support them, especially when we are expressly told that there is no such cheating in the game.

If someone can dig up something in the SDK, I'd be persuaded.

 

[azzaman333]

Anecdotes that can't be proven!

You have to admit Azzaman333, it's hard to accept such things as true without any facts to support them, especially when we are expressly told that there is no such cheating in the game.

If someone can dig up something in the SDK, I'd be persuaded.

I know, I just don't like the AI getting a warrior just before I steal his capital when he's clearly been working a 3 tile the whole time.

 

[Cvntpumeler]

I once played a game on Noble. I was a small crowded map and Huayna Capac was right next to me. So my warrior declared war and entered his territory. This was Turn 4. He had a quecha outside of his city exploring and another quecha already in his city.

Obviously this was cheating? Unless if he had a goody hut in BFC and it popped a quecha.

Um, im pretty sure the AI just gets "free units" at the start of the game. On Diety the AI starts with 3 settlers. It's not cheating, its just the way the difficulty level is. Winning on diety is hard, but it wouldnt be if the AI didnt get any free archers...

 

[Ermelinho]

i know, that ai knows about resources before they are revealed. that´s cheating , ain´t it?

 

[Cvntpumeler]

i know, that ai knows about resources before they are revealed. that´s cheating , ain´t it?

Sometimes I think it's weird that the Military Advisor will recommend Bronze Working as the first tech I research seemingly only if there is a hidden copper resource within my fat cross... anyone notice that - or just a coincidence? I never checked cause im lazy.

 

[Elandal]

AI no longer knows about hidden resources. That bug was fixed long ago.

 

[Psyringe]

i know, that ai knows about resources before they are revealed. that´s cheating , ain´t it?

Now please, not that unfounded rumor again. Where's your proof?

The AI *did* know about hidden resources in Civ3. This was one of my main gripes with it, so I tested specifically this question extensively when Civ4 came out.

The testing method was easy yet effective: Make a cross-shaped island with four identical branches. In one of these branches, hide a resource. Place an AI city in the middle and of the island and give the AI an additional settler. Then play a few turns and see where the AI founds its second city. If it can see the hidden resource, then it will settle in the branch where it's hidden. If the AI can't see the resource, then it will either randomly select one branch, or always head in the same direction no matter whether the resource is there or not.

Much to our surprise, the settler *always* preferred the branch with the hiddne resource, so we were very very alarmed. Firaxis specifically claimed that the AI would *not* be able to see hidden resources, yet our test showed clearly that it did. This called up Soren, who took a look at the code and figured out that we discovered a bug: The AI didn't see the resource itself, *but* when it calculated the best spot for settling, it could see the bonus yield of the resource, hence it always went in the direction of the resource. Needless to say, the problem was fixed in the next patch. Afterwards, no one has ever been able to demonstrate that the AI knows about hidden resources - because it doesn't.

Of course, this doesn't stop people from still claiming that the AI cheats and knows about resources, with no proof whatsoever, in a thread where I stated six posts above that this rumor has been *tested* and debunked.

Makes me wonder why programmers even bother programming a non-cheating AI when their customers simply refuse to believe them against all evidence anyways.

 

[Cvntpumeler]

Now please, not that unfounded rumor again. Where's your proof?

I have an anecdote with no proof, so bear that in mind... (I also don't think the AI cheats, merely questioning if this is a bug).

I was playing a game the other day in which I settled my capital city and went to explore the map, but ended up spacing out watching baseball and neglected to defend my capital against barbs (if playing diety, this will make you dead very quickly) at an early enough stage. So, a barb came and killed my city. I really liked my starting location, so I decided to load the 4000 BC autosave and actually pay attention this time.

I totally cheated (hey, its on Diety!) by deciding to place my capital in the original location my settler started at instead of the location I had moved him to in my first game -- mostly because the place I moved him to took the hidden Copper resource outside of my capital fat cross and I obviously would prefer the copper to another sheep resource... So, I settle the city there... and a funny thing happens. The (Recommended Military) technology is Bronze Working when in the previous game BW had not been one of my recommendations. I loaded the game again and settled away from the copper and it recommended 2 things neither of which were BW. I loaded the game again, and again BW was the (Recommended Military) tech after I settled next to the hidden copper.

This could have been a coincidence, but I didn't know about the bug from civ3 that had been fixed... Maybe there is a bug with advisors knowing about resources too? I didn't check it in WB, I'm just too lazy .

 

[Spearthrower]

Maybe the alternate position had more trees that the city placement AI wanted you to cut down?

 

[Cvntpumeler]

Maybe the alternate position had more trees that the city placement AI wanted you to cut down?

No there were actually less... thats the reason I moved, I wanted some trees to chop Stonehenge. It was a flood plain.

Edit: Thinking about it, the "city placement" AI is so stupid that it probably couldnt deduce that BW would be helpful to research if there was copper nearby... I've seen where it puts those freakin blue circles.

 

[Munro]

Now please, not that unfounded rumor again. Where's your proof?

The AI *did* know about hidden resources in Civ3. This was one of my main gripes with it, so I tested specifically this question extensively when Civ4 came out.

The testing method was easy yet effective: Make a cross-shaped island with four identical branches. In one of these branches, hide a resource. Place an AI city in the middle and of the island and give the AI an additional settler. Then play a few turns and see where the AI founds its second city. If it can see the hidden resource, then it will settle in the branch where it's hidden. If the AI can't see the resource, then it will either randomly select one branch, or always head in the same direction no matter whether the resource is there or not.

Much to our surprise, the settler *always* preferred the branch with the hiddne resource, so we were very very alarmed. Firaxis specifically claimed that the AI would *not* be able to see hidden resources, yet our test showed clearly that it did. This called up Soren, who took a look at the code and figured out that we discovered a bug: The AI didn't see the resource itself, *but* when it calculated the best spot for settling, it could see the bonus yield of the resource, hence it always went in the direction of the resource. Needless to say, the problem was fixed in the next patch. Afterwards, no one has ever been able to demonstrate that the AI knows about hidden resources - because it doesn't.

Ingenious! I admire your dedication and curiosity here. Seems inevitable that occasional bugs like this will creep into the logic but I find it hard to believe they are there intentionally.

Should be easy enough to test out the BW advisor behaviour the same way - just edit the save in worldbuilder to add or delete the bronze and see if the advice changes (without moving the settler or introduce any other changes).