Report Questionable Behavior

Barbarian pathing exploit:

If you create a line of units that completely cut off barbarians from your cities, the barbarians stop trying to go to your cities. They have a chance to attack your line if they randomly wander up to it, but otherwise they stop being a problem.

Basically, they don't see units as passable, and path around them.

I don't know if this will work against a full AI empire.

The AI is not always using the cannons etc. in the proper way. I had a huge pile of units attacking a city. Inside the city it had many units, including cannons. The thing is it didn't make use of the cannon, which could destroy my pile, and defend its city becasue it would greatly weaken all my units. I would have no chances of destroying the city. The AI stayed in the city with all the units, thus couldn't protect itself. My point is the AI should distinguish between cannons and other types of units, also those to defend the cities. I hope and wish it could be improved in the future.

Yeah. Something is wrong with AI_COLLATERAL tactic. They act like it is AI_CITY_DEFENCE in most cases. Really, it is not that hard to script: If an AI city has a large stack next to it, then attack with AI_COLLATERAL units . Something strange here...

shouldn't the blue circle be on the mounting to catch 2x fish, corn and clam?

AI BANKRUPT
I recently played a map (160x100) on emperor, and two AI-nations had conquered large quantities of cities (doubling their initial amount of cities). As they declared peace they went bankrupt and couldnt support any of their units. All their cities were left undefended, trying to produce units for defence. A never-ending circle of stupidity. Another thing to consider is that none of the surrounding AI's attacked the undefended cities...

If the AI didnt have the income to sustain that amount of cities, it should have razed some and maybe emphasize a bit more on economy. I think its cool how the game evolved, but it just seemed like taking candy from a child at the end, when there wasnt any defence...

I apologize if someone else has reported this already. I'm getting this error message regularly in Better AI games:



I'm getting it whenever I load a Better AI save; it usually pops up during the AI's "think cycle" on one of the first few turn changes after loading the save. Everything in the game hangs until I close the dialog box (using "Ignore Always"). With that option, I don't see it again until the next time I load a Better AI save.

I should probably mention that I'm also loading the Civ4Alerts mod from Custom Assets, so that could have something to do with it. That is the only other mod I'm using when I see this.

It's not a huge issue, but it is a minor annoyance, obviously.

Apologies if this is in the wrong place, but here are a couple of things that might need addressing in the Better AI mod.

1) Ref: http://forums.civfanatics.com/showthread.php?p=7124523#post7124523
There is yet another questionable trade denial in CvPlayerAI::AI_civicTrade(). AI nations will always redline a change into Emancipation because "we don't want to deal with the unhappiness." This is a misunderstanding of the way that civic works since the unhappiness applies to running anything other than Emancipation rather than running Emancipation itself. There is a (minimally-tested) proposed fix in that post along with a bit of discussion.

2) Ref: http://forums.civfanatics.com/showthread.php?p=7127772#post7127772
As of 3.17 the Spread Culture Espionage mission (implemented in CvPlayer::doEspionageMission()) appears broken. According to the release notes it was changed from 3.13 to fix a bug; no details were provided but I'm guessing it was related to how much culture accumulates in the tiles around the target city. However, part of the change involved switching from changeCulture() to changeCultureTimes100() without modifying the base change amount. As a result, instead of spreading 5% of your culture to the target city, you wind up spreading 0.05%. This pretty well renders the mission useless. I expect the fix is to use changeCulture() instead but haven't tested that yet myself.

This thread works fine, particularly since you've already got discussion threads for these effects.

Thanks for digging in so deep and uncovering these well hidden bugs.

After Astronomy I was racing several AI's to settle an empty island continent. It was big enough for 5 cities. The AI's settler+escort stack stalled when a barb city popped up at or neaar its detination. Then a different AI's settler plus escort stack stalled when I beat it by founding my city first. When I say 'stalled', I mean just sat there and did not retarget a new planned city site when there were several others available.

This is in BTS 3.17, tectonics monarch

oops wrong thread

Grames said:

After Astronomy I was racing several AI's to settle an empty island continent. It was big enough for 5 cities. The AI's settler+escort stack stalled when a barb city popped up at or neaar its detination. Then a different AI's settler plus escort stack stalled when I beat it by founding my city first. When I say 'stalled', I mean just sat there and did not retarget a new planned city site when there were several others available.

This is in BTS 3.17, tectonics monarch

Click to expand...

Have a saved game?

Sorry, if that issue has already been brought up, as I haven't read the whole thread...I think the voting behaviour of the AI is often very questionable.

Especially the AP votings are usually about declaring peace between warring nations and usually every AI agrees with signing peace, including the one which started the war (even if they are about to win the war). Same with allies - the AI always seems to think that it can "help" a friend by ending his war, even if the friend is clearly winning this war.To an extend this could be justified (they perhaps just want to stop the war to get not drawn in), but at least no AI should vote for ending a war it has started and it is about to win - that seriously damages the effectivity of the strategic AI.

If needed, I can probably provide a save for this (with v0.35 of BetterAI), one turn before such a voting takes place.

Please do post a save if you already have one handy, it's nice to have sort of standard tests to compare behavior against. I would hope the AIs in the war would use the same logic as if some was asking them to end the war in diplomacy to decide yes/no ... declaring "Never!" would need to be a more extreme decision since it carries extra consequences, but the simple yes/no for players in the war seems to me like it should use the standard diplomacy logic.

We should start up a discussion topic for this ...

Just had a 17 hour game from start to finish on normal speed with 2 human players and 5 AI, no tech brokering, no vassals, no permanent alliances. First time trying out Better AI.

They certainly seemed to know how to do things better all over. But still there were a few weird things that humans wouldn't do:

1. Throwing everything they have at just one impossible-to-take city, over and over for hundreds of years. With the most enormous losses (tons of big stacks). They could just as easily gone around it, During this moment too, it did not try to shift more towards economical warfare such as espionage missions, pillaging, blockading etc. Just went for that one city.

2. AI without nukes, attacking human player armed with a very large nuclear arsenal. Got nuked to the stone age. This happened two times. One time was conventional with a reasonable/good land gain in mind. The second was completely ******ed:
* The AI without nukes declared war
* The goal was to take a very very horsehockey nuked to the stone age city which had been captured by the nuclear arsenal owning human previously. It had no workable tiles, had fallout around it and was swimming in someone elses culture and was going to flip for sure at a later date.
* The city was undefended, and I suppose the AI couldn't resist such an opportunity, not considering the nuclear angle. It got nuked to the stone age too.

3. The most powerful AI had a 'worst enemy' which was a seemingly much weaker AI, without any friends, sharing borders. Yet never did the logical thing: Annexing this weaker AI. They were frequently 'worst enemy' throughout the game.

4. The AI refused to trade with long time friend humans who never did any hostile thing to each other, when pleased at the end after being nuked. They are still so concerned of giving away too much, no matter how helpful and loyal their friend and trading partner has been, that they don't want to trade. Not even for fission.

5. The AI, after being nuked, had very slow teching. It was researching fission and had 11 more turns to go on it. It refused to let me even GIVE it fission to catch up, as it was 'near' getting it itself.

6. Only one person (An AI) was eligible for the apostolic palace chairman election. Yet voting for him would make him like you more, with a cap of +2 attitude modifier.

7. After over 65 nukes had been used by one of the human players, the two remaining AIs not long after declared war on each other and used their newly acquired nuclear arsenal against each other. Not too many nukes (Maybe 5?) but the global warming was turning the whole planet quickly into desert.

8. AI refusing to change religion: seems blind to diplomatic benefits of such acts.

9. AI never ever donated anything to it's friendly civs. It was like that in normal BtS as well most of the time. No gifts as a reward or to help one out in the time of dire need. Also, it does not try to use gifts to make you like them more if relations are bad already (but it wanting to make it better, if it even can want that?).

to 1.: yeh, AIs should build/use units with collateral damage more often

3. The most powerful AI had a 'worst enemy' which was a seemingly much weaker AI, without any friends, sharing borders. Yet never did the logical thing: Annexing this weaker AI. They were frequently 'worst enemy' throughout the game.

Click to expand...

*nod*, you'd expect them to clean the weaker AI out.

4. The AI refused to trade with long time friend humans who never did any hostile thing to each other, when pleased at the end after being nuked. They are still so concerned of giving away too much, no matter how helpful and loyal their friend and trading partner has been, that they don't want to trade. Not even for fission.

Click to expand...

What message? It is possibly you hit the anti-trade-exploit cap in the AI (ie, I'm afraid you have traded for too many technologies).

5. The AI, after being nuked, had very slow teching. It was researching fission and had 11 more turns to go on it. It refused to let me even GIVE it fission to catch up, as it was 'near' getting it itself.

Click to expand...

Tech brokering was off. If they research fission, they can trade it away. Being given fission means you cannot resell it. So accepting a technology when you are almost done is generally a bad idea, and something that players could exploit easily.

6. Only one person (An AI) was eligible for the apostolic palace chairman election. Yet voting for him would make him like you more, with a cap of +2 attitude modifier.

Click to expand...

Heh. But you like him you really really like him.

If enough people abstain, does it remain/become empty?

7. After over 65 nukes had been used by one of the human players, the two remaining AIs not long after declared war on each other and used their newly acquired nuclear arsenal against each other. Not too many nukes (Maybe 5?) but the global warming was turning the whole planet quickly into desert.

Click to expand...

A human would quite possibly say ". .. .. .. . the world" and shoot off nukes in that case. Practically, the world is already going to hell, why not have more of hell when it happens?

8. AI refusing to change religion: seems blind to diplomatic benefits of such acts.

Click to expand...

The AI has to both roleplay, and play the game.

9. AI never ever donated anything to it's friendly civs. It was like that in normal BtS as well most of the time. No gifts as a reward or to help one out in the time of dire need. Also, it does not try to use gifts to make you like them more if relations are bad already (but it wanting to make it better, if it even can want that?).

Click to expand...

The AI only generally gifts things to civilizations that are much weaker than it.

It also maintains a "total amount you gifted/over-traded it", and that value is used for when you ask the AI to spare stuff for a friend.

What message? It is possibly you hit the anti-trade-exploit cap in the AI (ie, I'm afraid you have traded for too many technologies).

Click to expand...

"we don't want to start trading with this technology just yet" I believe. No matter how desperate the situation is for the AI in question to get 'something' to help it out.

Tech brokering was off. If they research fission, they can trade it away. Being given fission means you cannot resell it. So accepting a technology when you are almost done is generally a bad idea, and something that players could exploit easily.

Click to expand...

Good point, but 11 turns was not 'almost done'. There was only 4 civs left, 2 were AI, and he was not (If I recall) on good terms with the other AI which was the only one without fission left. Furthermore, this AI was in dire dire need of fission to defend itself/keep up with the stone-age nuking human player. So it offered no trading possibility (not really) and the situation was extremely gloomy and desperate for the AI in question.

Heh. But you like him you really really like him.

If enough people abstain, does it remain/become empty?

Click to expand...

I think the main guy will vote for himself and that'll be enough to get the seat.

A human would quite possibly say ". .. .. .. . the world" and shoot off nukes in that case. Practically, the world is already going to hell, why not have more of hell when it happens?

Click to expand...

Is there some kind of threshold where the AI thinks that it's no longer useful trying to prevent causing their own destruction (through global warming)?

The AI has to both roleplay, and play the game.

Click to expand...

But it doesn't really roleplay about religion anyway - it switches to another religion if it thinks it can benefit from it from religious bonuses (i.e. more spread/buildings), or if you pay it handsomely. If that is it's real considerations, and I think it is, then it should also be aware of diplomatic implications of different religions.

The AI only generally gifts things to civilizations that are much weaker than it.

It also maintains a "total amount you gifted/over-traded it", and that value is used for when you ask the AI to spare stuff for a friend.

Click to expand...

Yep, it can be made to give things, quite generously sometimes. But it never uses this to try to better it's standing with others, never takes initiative to suck up to someone or further the relationship.

jdog5000 said:

Please do post a save if you already have one handy, it's nice to have sort of standard tests to compare behavior against. I would hope the AIs in the war would use the same logic as if some was asking them to end the war in diplomacy to decide yes/no ... declaring "Never!" would need to be a more extreme decision since it carries extra consequences, but the simple yes/no for players in the war seems to me like it should use the standard diplomacy logic.

We should start up a discussion topic for this ...

Click to expand...

Here is the save (Better AI 0.35)...it's the turn before the voting comes up; so to see Catherine voting for making peace with her enemy Bismarck (she had declared on him 1790) press "turn end" twice. According to the graph, Catherine was more sucessful in that war (though she had not captured a city)

The are more AI voting oddities and I will post saves of them when I stumble across them, but thats probably the most clear and annoying one.

This save is from over a year ago so maybe it's no longer relevant because of new patches or current AI changes.
Korea declares war on France and France is losing. I join in because I was planning on taking France and they've forced me to act before they take everything. After a few turns the Apostolic Palace gives me the opportunity to end the war against Korea, actually saving France so I'll have France for myself. I vote yes, the results come back and Korea votes "Never!" as one would expect, but France also votes "Never!" which considering their situation is suicidal. France did only have 4 votes so it wouldn't have made a difference but still, what up with that?
I thought maybe they were deciding based on their relations rather than their strategic situation, so they would never end a war against someone they hated. As much as I'd hate to reinforce a stereotype, France really should have run from that fight.
End the turn and the resolution option will come up.

Sometimes you can see the AI settling 1 plot away from the ocean (so, no lighthouse can be built, no water units trained, ...). This should be fixed.