Report Questionable Behavior

[TheLazyHase]

A "true" questionable behavior about naval invasion : I see the already pissed and WHEOOH ottoman sneak 4 galleon not far from my main cities. I dispatche the bunch of frigate and east indiamer I have to make sur they can't declare war and land on the same turn. They declare war, then understand they cannot go throught, and begin to flee. I love this behavior because it has allowed me to destroy the galleon without declaring war first, but it's badly played.

 

[PieceOfMind]

I don't know how BBAI is supposed to treat world-buildered settlers but I have to say the actions taken by Mansa Musa in the following save game are questionable.

I put him on 4 small islands in the middle of the ocean with 1 settler on each. After he builds a first warrior, he then proceeds to build a settler which will take him 50 turns. Meanwhile he is surrounded by fish with no fishing nets and already has 3 settlers which he doesn't use to settle.

In particular, I was testing whether the AI knows that it can settle cities within 2 tiles of another city if they're on different landmasses. I'm not convinced the AI does this properly at the moment.

Savegame is attached, version is BBAI v0.82, installed as a standalone mod.

EDIT Image attached

Spoiler :

 

[Afforess]

I believe that those spots are all too close to the original city. Isn't there a spacing issue?

 

[PieceOfMind]

I believe that those spots are all too close to the original city. Isn't there a spacing issue?

In particular, I was testing whether the AI knows that it can settle cities within 2 tiles of another city if they're on different landmasses. I'm not convinced the AI does this properly at the moment.

Even if there were a spacing issue, why build another settler then? The 3 tile space rule is only for cities on the same landmass, by the way.

He is researching fishing though, which is obviously the best tech choice.

 

[Afforess]

Even if there were a spacing issue, why build another settler then? The 3 tile space rule is only for cities on the same landmass, by the way.

Really? I learned something new today then.

He is researching fishing though, which is obviously the best tech choice.

 

[jdog5000]

A "true" questionable behavior about naval invasion : I see the already pissed and WHEOOH ottoman sneak 4 galleon not far from my main cities. I dispatche the bunch of frigate and east indiamer I have to make sur they can't declare war and land on the same turn. They declare war, then understand they cannot go throught, and begin to flee. I love this behavior because it has allowed me to destroy the galleon without declaring war first, but it's badly played.

If you have a save or autosave that would be very helpful.

 

[Mattygerst]

I'm sure this has been talked about before, but I have only started using & watching Automated Workers with the 0.81 patch release.

But (and I THINK I have a save game available, but I'm not positive). The scenario: I have a capital, and inside my cultural borders but outside my BFC are horses - my ONLY horse resource. The resource is in such a position that I can't really put another city where the horse is inside a BFC. Now, I create the required pasture for the horse. Then a few turns later I automate my workers. And as I'm war building - catapults, sword, horse archer...my worker starts and completes building a farm over the horse - even though it is my only horse resource and the farm is not a useable tile. Then, after that, I watch them rebuild the pasture. Then...some time goes by, and they rebuild the farm.

This is "normal" activity right now? Seems like that could be pretty bad in terms of war building for the AI...

 

[bestbrian]

That's why you never automate workers.

And yes this is a huge handicap for the AI, in that it can never figure out what to build in the first place, and once it does, it changes it to something else.

 

[Mattygerst]

That's why you never automate workers

Obv you never automate workers. But I was doing it with the sole purpose of watching their behavior. I knew that the cottage replacement had been fixed 2 patches ago, along with the mine/windmill this patch...so I wanted to watch their behavior.

Then to see a plot un-usable in the BFC...but the only resource the civ has of horse - an mega important military resource get turned from: Pasture, to farm, to fort, to farm, etc...

Just felt like it should be reported.

 

[MartinHarper]

Automated workers also mess up forts used for canals.

 

[Guest01]

I wonder why AI never falls in some kind of desperate behavior in face of imminent defeat with taking any possible measures to hurt winning side or at least join it (depending on dipomatic attitude and leader properties)? It seems totally appropriate for obviously losing side to fall in desperation and thus making desperate things. Instead now AI would simply look that that you launched your spaceship or approaching cultural victory and would be doing nothing about it.

 

[TheLazyHase]

I wonder why AI never falls in some kind of desperate behavior in face of imminent defeat with taking any possible measures to hurt winning side or at least join it (depending on dipomatic attitude and leader properties)? It seems totally appropriate for obviously losing side to fall in desperation and thus making desperate things. Instead now AI would simply look that that you launched your spaceship or approaching cultural victory and would be doing nothing about it.

Let's suppsoe that every civ DoW on you when you are near cultural victory. Believe me, to survive such an onslaught, you need quite an army. And if you have that kind of army, why not use it to conquer the world instead ?

In the same token, how can somebody with that kind of comportemebnt vote for someone in UN ?

Making that kind of thing can be like removing all non-domination victory condition. So, it look like a good idea, but it's a bad one.

 

[Guest01]

Let's suppsoe that every civ DoW on you when you are near cultural victory. Believe me, to survive such an onslaught, you need quite an army. And if you have that kind of army, why not use it to conquer the world instead ?

No need to force every civ make DoW. Let's suppose that would be only agressive/imperialistic leaders or only civs that already annoyed or furious toward winning side (human or another AI). Leaders with good relations could even try to make permanent alliance (if available).

Or let's suppose AI would implement such a plan only if option "Agressive AI" has been chosen on start. What do you think about that last suggestion?

And if you have that kind of army, why not use it to conquer the world instead ?

Well, maybe I just don't like such kind of victory? After all, defend myself against massive onslaught is not the same thing as go around the world while exterminating or vassalising every single breathing thing.

For now it simply sad to look at obviously losing AI which didn't even try to make something about such a situation.

 

[TheLazyHase]

No need to force every civ make DoW. Let's suppose that would be only agressive/imperialistic leaders or only civs that already annoyed or furious toward winning side (human or another AI). Leaders with good relations could even try to make permanent alliance (if available).

Or let's suppose AI would implement such a plan only if option "Agressive AI" has been chosen on start. What do you think about that last suggestion?

First, I'm not somebody involved in the project, my opinion have about the same value as yours for thoses who maintain this project

Then, I still don't like that in a general sense. The furious and such civ already have pretty high chance to DoW on you ; if they can do something against you, are angry at you and still does not engage, then the problem is whithin war check, not victory.

Also, it's a problem of planning how to win. If my opponent is going cultural and suddenly engage in war because I'm near victory, then it's poorly played ; he'd better stay in his track, hoping to beat and/or that some warmongering civ will destroy the other's civ victory hope.

Well, maybe I just don't like such kind of victory? After all, defend myself against massive onslaught is not the same thing as go around the world while exterminating or vassalising every single breathing thing.

For now it simply sad to look at obviously losing AI which didn't even try to make something about such a situation.

Yes, but 99% of the time, I use thoses paths because I don't want to be the first military force in the world. So, an AI that specifically declare war when I near victory is not something I would welcome.

All in all, I think that effort should be put to make sure the AI can have a strong economy and choose his battle wisely. Doing so, spaceship and cultural victory will feel a lot tougher, since A) I would have to make sure I have the warmonger in check B) the non warmongering AI will be a lot more near victory when I choose.

 

[jdog5000]

The AI is learning how to go after victory conditions other than Cultural, and part of that will bring an ability to guess what victory condition another player is going for and how close they are. With this knowledge, then the AI can do something ... what exactly makes sense is still being decided, but the end game should become much more exciting.

Having all AIs dogpile a player close to winning goes much too far to the game side of the roleplay vs game debate in my view. Instead, AIs who might be considering war with a player near winning anyway (due to attitude and the other normal factors) won't wait ... they'll pick the near winning player over other targets and speed up their planning and preparation phase. AIs close to their own victory conditions may try to just ****** the other player's progress enough so they can win first, through espionage and diplomacy.

It should be really a great improvement when it comes together, but it will take a while to get right and have all the features in.

 

[Mattygerst]

Repost for JDog because I think maybe it got lost in the above posts:

I'm sure this has been talked about before, but I have only started using & watching Automated Workers with the 0.81 patch release.

But (and I THINK I have a save game available, but I'm not positive). The scenario: I have a capital, and inside my cultural borders but outside my BFC are horses - my ONLY horse resource. The resource is in such a position that I can't really put another city where the horse is inside a BFC. Now, I create the required pasture for the horse. Then a few turns later I automate my workers. And as I'm war building - catapults, sword, horse archer...my worker starts and completes building a farm over the horse - even though it is my only horse resource and the farm is not a useable tile. Then, after that, I watch them rebuild the pasture. Then...some time goes by, and they rebuild the farm.

This is "normal" activity right now? Seems like that could be pretty bad in terms of war building for the AI...

 

[Guest01]

TheLazyHase

First, I'm not somebody involved in the project, my opinion have about the same value as yours for thoses who maintain this project

I understand that

Then, I still don't like that in a general sense. The furious and such civ already have pretty high chance to DoW on you ; if they can do something against you, are angry at you and still does not engage, then the problem is whithin war check, not victory.

Desperate times call for desperate measures. There's no point in making normal war check, continue research and production as usual if looming defeat is obvious and ultimate. If I've lauched my spaceship while AI just begun to construct it and there's no possibility to win culturally/domination and such for him then AI is obviously doomed. In such case there's no point to care about anything except to hurt me as hard as possible no matter what the cost. Currently AI would continue behave as usual after spaceship launch as if nothing happend. Some kind of "dead hand" policy would be appropriate for such cases.

Also, it's a problem of planning how to win. If my opponent is going cultural and suddenly engage in war because I'm near victory, then it's poorly played ; he'd better stay in his track, hoping to beat and/or that some warmongering civ will destroy the other's civ victory hope.

I agree here: it would be wise for AI to divert available resources to speed up cultural growth if there is threat to lose while using spies and diplo to slow down its opponent. And only if after that it becomes obvious that AI is too late it should going mad.

Yes, but 99% of the time, I use thoses paths because I don't want to be the first military force in the world. So, an AI that specifically declare war when I near victory is not something I would welcome.

But does it intelligent behavior for AI to no try to stop you by military means if there's no other way?

jdog5000

Having all AIs dogpile a player close to winning goes much too far to the game side of the roleplay vs game debate in my view.

Not necessarily dogpile a player but any winning civ, human or AI. And that policy could be activated only if "Agressive AI" option chosen. BTW if you mention roleplay vs game debate it's very human-like to feel desperation and going mad about that.

 

[JujuLautre]

BTW if you mention roleplay vs game debate it's very human-like to feel desperation and going mad about that.

That's the whole point of the debate: do we want that much human in the AI or not ?

 

[TheLazyHase]

BTW if you mention roleplay vs game debate it's very human-like to feel desperation and going mad about that.

It's not roleplaying a nation. Nations usually don't suicide themselve on something else to avoid him to win. They usually surrender or do something along thoses line, the self preservation for entire nation is usuallymuch stronger than that.

In other world, they should play as if victory condition were a significant step, not the end of their world. And suiciding on the first is generally not a good way to get going.

 

[r_rolo1]

Desperate times call for desperate measures. There's no point in making normal war check, continue research and production as usual if looming defeat is obvious and ultimate. If I've lauched my spaceship while AI just begun to construct it and there's no possibility to win culturally/domination and such for him then AI is obviously doomed. In such case there's no point to care about anything except to hurt me as hard as possible no matter what the cost. Currently AI would continue behave as usual after spaceship launch as if nothing happend. Some kind of "dead hand" policy would be appropriate for such cases.

This is pretty much the only case where a AI should think on dogpiling a player , but....

I have noticed that the discussions about inteligent behaviour of the AI commonly fall in 2 fallacies:

- A game with a AI more chalenging for the human is a game with a more inteligent AI

- The AI should use the strategies human use beating the AI

The first is not completely wrong ... if you have only one AI oponent, a thing that is rare in Civ IV. Normally you are facing more than one AI player and all of them ( ideally ) are trying to win most likely at the expense of the other AI in game. And to say the truth all this discussion of "AI should go burn" is filled with this "one AI" fallacy .... think on it: if a AI player is coded to go burn the cap of a player ( human or not... ) that is close of winning without thinking twice, who profits with it? Most likely other AI that stayed in home preparing it's army to invade the guy that decided to go "bomb Iran" losing a good part of it's army, right? What I'm saying is not that the AI player should not go burn period, but that the AI player should not go burn if it will not win anyway or if it will be in a perilous situation even if it sucessfully thwarts the win and that it should first try to find someone to do it's dirty job if it can ( that suposing that we teach the AI to bribe with a minimum of efficiency ) ... if a AI player wants a city burned, it doesn't matter who burns it, right? Other thing linked with that is that you can have more than one player closing to a win at the same time.... what should the AI do if there are 2 SS close of being finished, one guy with two legendary cities and close of the third, one guy close of winning via UN and other close of winning via AP ( nothing forbids this scenario .... ) ? Nothing? Try to burn them all? Who to prioritize in case of trying to stop more than one?

The second fallacy is pretty widespread unfortunately People consistently forget that a lot of techniques or habits the humans have in game are the direct result of how the AI plays.... the extreme example is the liberalism beeline: it only looks so shiny in SP because the AI ( sensibly, I must add ) chooses normally to also tech the lower side of the tech tree, and because of that, the human can buy/steal/extort the techs of the lower part of the tech tree with ease.... It also applies in here: saying to the AI to burn something because a certain player is close of a win implies that you have some kind of alarm linked to some kind of game event, like getting the 2nd legendary city, a certain SS part, a certain % of the votes in UN/AP ( or something more rebuscate ) ... Humans can do it well with stock AI ( and even with Better AI ) because the AI players normally play in a gradual way, stockpiling it's culture in a predictable way, building the SS parts as soon as they get the tech, fighting one war at the time, buinding up it's diplo links slowly... Because of that, humans can learn easily to time their game in a fashion that allows them to go burn something and still have their chances of winning intact. The issue is that, as any competent player knows, nothing forces the aproach to the victory conditions to be gradual: you can pass from 40% of land mass to 60% in one turn, you can only secure the necessary votes for a diplo win in the turn where the vote is cast, you can build all the SS parts in one turn, you can even ( theoretically atleast ... never saw it done , and in fact I'm trying to do that in a SG as we speak ) raise 3 legendary cities from 0 culture in one turn. No matter how sophisticated the trigger mechanism we put in the AI to go burn ( and even suposing that we somehow resolve the issues I talked before ) , due to nature of the victory conditions in civ IV, there will be always the risk of a player ( AI or human ) being caught by a sudden aproach to a VC by other player.

To add , as any player knows, a military operation needs time to be prepared, especially if the player is a rather inflexible computer algorithym That is why the Civ IV AI has the "hands full" period, right? Saying to the AI , like you said in the part I quoted, to forget about normal war preparations , is pretty much condemn the AI to do a uncoordinated ( thus, with little chances of suceeeding ) attack , especially if the target player adopts a non-gradual aproach to the win, even by acident ( a unexpected event spread the AI religion to a recalcitrant theo civ, one AI changes religion and sudently will vote of one of the candiadates for the UN win, one civ vassals other and grabs it's votes .... this things can even happen to the AI today ).... even humans, that are normally far more inteligent than the AI, can be easily surprised by issues like this and be unable to mount a sucessful offensive in due time ( I've seen that happening more than once ), so the danger of the AI suffering from that is quite high.

To resume, I'm not saying to tell the AI players to not go burn something to stop a enemy win ... I'm just saying that it might be hard to pull a mechanism that works better than the current "do nothing" aproach, atleast in terms of increasing the odds of a certain AI player winning the game by using that mechanism.