Report Questionable Behavior

I think the attack tactics have been improved since this rather old post. What's still a problem, however, are rivers. The AI doesn't seem to have any logic to avoid the "-25% for crossing river" penalty. I can provide a screenshot later, but it's a nonspecific problem that I've seen over and over again across many games.

I can't think of many reasons why you wouldn't want to move the extra tile to be on the same side of the river as your target city before laying your siege. I suppose it provides some protection from the city defenders attacking you, but that's rarely the case since the defenders have enough movement points in their own territory to cross the river first and then attack.

This is at least better in svn (version p). The issue was that with the old attack city logic, once the AI was next to the target city it wouldn't move again ... it would attack from which plot it first approached the city from. The AI is now able to move around the target city, though it may need to do it under more circumstances.

(The attack city stack logic has been almost entirely re-written in the last week ... )
 
[BBAI 0.90o developer] I've found out that the AI is still not tough enough to conquer a city in a short time. Maybe the broken river war AI is the reason. The following happened while using AI autoplay:

The AI has an overwhelming stack in front of the enemy city. Step1: Bombardment (which is acceptable). Step2: Waiting for other troops with THAT stack? :lol:


Next round the city gets conquered. Too late imo.

There were a number of issues with the attack stack code which have been ironed out in version 0.90p, so in terms of attacking cities it now (finally) appears to be doing what I want it to. There were all kinds of interesting hidden ways the AI would decide to commit suicide or do other stupid things which stopped it from attacking.

The AI is set up to attack with bombarding using the following formula:

Code:
stackStrengthPerc = (100 * my offensive stack strength) / (their defensive stack strength)

if( stackStrengthPerc >= 350 )
    ->  attack without bombarding

if( iFirstAttackOdds >= 10 )
    if( stackStrengthPerc >= (350 - 3 * iFirstAttackOdds - 3 * iBombardTurns) )
         ->  attack without bombarding

where iFirstAttackOdds are the odds of the AI's first attack winning (a sense of how much better attacking units are), and iBombardTurns is the number of turns it would take to bombard the enemy city defenses.

So, the AI will go in guns blazing for sure if its stack is 3.5 times stronger. If it has non-suicidal odds with its first attacker, then it will consider reducing that based on how good its attack odds are and how long it will take to remove the city defenses.

Some of these settings can be controlled in XML (your variables may be a little different), so you can tweak things. It's set up to be conservative as there were still too many AI suicides ... turns out those AI suicides were unintended effects of other parts of the code (and that the BTS defensive stack strength calculation always ignored walls/castles ...). Now that the stack math and attack decision making is working right, it might be okay to make things a bit more aggressive.
 
Has anyone else noticed that AIs with universal suffrage will collect much gold (and lower research to 10%) but won't spend it to upgrade units or buy buildings/units? In my game, the AI has now over 10.000 gold.
 
Has anyone else noticed that AIs with universal suffrage will collect much gold (and lower research to 10%) but won't spend it to upgrade units or buy buildings/units? In my game, the AI has now over 10.000 gold.
How late in the game is it? I know that in BTS, the AI stores gold proportional to the number of game turns, or something like that.
 
The AI seems to be able to enter Conquest4 without declaring war 3 times (I checked the logs, only 1 DoW and even that was after the player was already in Conquest4). But that really the issue here, more importantly than that it was able to start a warplan against a civ without having any means of travelling there.

This is a test game* where I had AI Autoplay running for 200 turns. 6 Civs, everyone alone on his continent, no barbs, huts or random events. Alexander and Montezuma are only 1 ocean tile apart, and Greek Culture covers some of that wich makes it possible for the Greek to travel to Aztek lands, it opens trade routes too. But even with open borders, Montezuma can't get to Greek lands with galleys, still he entered Conquest4, planned total war, declared on turn 205 - only to make peace 10 turns later... and starts the same war plan again (screenshot).

( * I wanted to watch the AI settle their continents because I am delusional enough to think that I might be able to figure out how to make the AI choose a bit better locations for city founding. )
 

Attachments

  • Civ4ScreenShot0074.JPG
    Civ4ScreenShot0074.JPG
    313.8 KB · Views: 131
If Monty actually declared war in that situation .... it is already a big improvement over vanilla BtS behaviour :lol: In vanilla BtS monty would stay in war prep state until it collected enough ocean going ships to carry his SoD ( seen that a lot in the LHC )

Not that the issue is not serious ... but it might be hard to solve. The fact is that Monty needs Ocean going vessels ( note, not Astronomy ), and this not considering some tricks that work in other maps, like armies being expelled to other continents by closure of borders of third parties ... and above all, needs to recognize it needs Ocean going vessels before all that ;) . Anyway, making the war preps against some civs depend on the actual presence of some special units might not be good for the AI performance ....
 
In my game (1.00) Monty declared war on me w/o being able to capture at least 1 city.
So he just wandered around them with his stacks pillaging.
 
If Monty actually declared war in that situation .... it is already a big improvement over vanilla BtS behaviour :lol: In vanilla BtS monty would stay in war prep state until it collected enough ocean going ships to carry his SoD ( seen that a lot in the LHC )
It is what I would expect the AI to do, how is that old behaviour not as it is supposed to be? Note that the AI will still build enough cargo ships to carry his SoD, only here it seems it settled for galleys - which is just wrong when you can't get to the other continent with those.

Not that the issue is not serious ... but it might be hard to solve. The fact is that Monty needs Ocean going vessels ( note, not Astronomy ), and this not considering some tricks that work in other maps, like armies being expelled to other continents by closure of borders of third parties ... and above all, needs to recognize it needs Ocean going vessels before all that ;) . Anyway, making the war preps against some civs depend on the actual presence of some special units might not be good for the AI performance ....
War preparations, or at least which target to pick and when to declare war should depend on whether or not the armies can actually get there. But what really bugged me was that he entered Conquest4 when having only met 1 of 5 opponents and couldn't even touch that one.

Isolations isn't easy to play, even for humans. Massing armies before anyone gets ocean going vessels is still strange in my eyes.
 
It is what I would expect the AI to do, how is that old behaviour not as it is supposed to be? Note that the AI will still build enough cargo ships to carry his SoD, only here it seems it settled for galleys - which is just wrong when you can't get to the other continent with those.
Notice that there can be a huge gap between meeting a overseas civ and being actually capable of launching a attack to it. And remember that we are talking of Monty: agressive, builds a bunch of units, but normally techs poorly ... this means that the old behavior can create situations where it takes 200/300 turns ( in standart speed ) from Monty targetting a overseas civ ( sometimes at turn of contact ) and being actually able to send a attack via sea. This means that Monty AI is stuck in looking at the sea waiting for galeons while maybe it could be using their army in more productive ways in his landmass ( not what happened here OFC ) ... not exactly a inteligent decision :p Atleast DOWing and make peace allows it to actually target a better and more reachable target ;)

The situation that you posted is a little bit special because monty theoretically could have galley acess to Greece if he has settled 1 S that southermost city. A thing that only adds shame to injury ...
War preparations, or at least which target to pick and when to declare war should depend on whether or not the armies can actually get there. But what really bugged me was that he entered Conquest4 when having only met 1 of 5 opponents and couldn't even touch that one.

Isolations isn't easy to play, even for humans. Massing armies before anyone gets ocean going vessels is still strange in my eyes.
Well, I definitely do agree with you that isolation is not easy ( for anyone, even more to the AI ) and that ammassing a army in those situations is stupid. i was just pointing that the definition of "the armies can actually get there" is not as streamlined as that. Think on what would happen in the 1S scenario in your map for one quick example: it could happen easily that either Monty or Pericles could be able to ferry troops via galley in one turn and unable to do that in the next turn , or vice versa , due to changes in the cultural borders ...

In fact it is worse than that: the only way of actually checking if you "can get there" is to pathfind with one unit ( even if you have ocean going ships , it is not warranted that you can get there, due to OB issues/landlocked target civs ) ... this brings a good bunch of issues: you, even if you have ocean going ships, might not be able to ferry the whole SoD ( imagine that the AI is playing a scenario where it has a couple of galleons , but has no Astro ... or a modded version what gives ocean going ships in a quest/event ... or even one modded version that has a promo that makes ships ocean going ), you can have the units to do that but be unable to get to the target temporarily ( OB restrictions ) ... it is a very complex issue. That was my point: this is not ok, but fixing it is not easy ... and has the potential of making things worse if not thought right, simply because you have no way of actually warranting that you can reach someone with a SoD just because you have some ocean going ships or that you can reach someone with a SoD just because you don't have them.
 
AIs picking wars against overseas targets long before galleons has been a problem from the beginning ... it's gotten better recently, they will now abandon their unachievable warplans after some time and plan something new. The problem is really that detecting this situation to help the AI avert it is not easy, especially when you consider that if Monty starts planning war in a situation like this and then starts working towards Astronomy it's actually a good decision. You also don't want to block the AI from engaging in galley wars early on archipelago maps.

I have just given a big boost to the value of galleons when choosing techs when the AI has naval assault war plans ... this should help the AI beeline Astronomy when in this situation.

I agree with you that entering Conq4 without having met most of the other civs or having taken any aggressive action is bad behavior. Monty entered Conq4 because he has by far the most powerful army in the world (both top power and a huge edge over the average). I'll tweak this so it works better for continents.
 
AIs picking wars against overseas targets long before galleons has been a problem from the beginning ... it's gotten better recently, they will now abandon their unachievable warplans after some time and plan something new. The problem is really that detecting this situation to help the AI avert it is not easy, especially when you consider that if Monty starts planning war in a situation like this and then starts working towards Astronomy it's actually a good decision. You also don't want to block the AI from engaging in galley wars early on archipelago maps.

I have just given a big boost to the value of galleons when choosing techs when the AI has naval assault war plans ... this should help the AI beeline Astronomy when in this situation.

I agree with you that entering Conq4 without having met most of the other civs or having taken any aggressive action is bad behavior. Monty entered Conq4 because he has by far the most powerful army in the world (both top power and a huge edge over the average). I'll tweak this so it works better for continents.
That is a sensible aproach to the general issue, but I think that it still leaves 2 holes:

- The more obvious one : the AI might take too long to get Astro. I , in my LHC days, had routinely seen that isolated vanilla BtS AI might be found by the humans in almost stone age ( sometimes not even compass was researched by them ) even in levels where the human is competitive, especially the moronic dudes that build units like crazy . Those guys might easily have more power than the human due to their love by the military, but being 100 turns away of being able to tech astro, even if they putted full steam on that...

-There is a stock civ in BtS that has a ocean going unit before Astro :p
 
hello there..I have only played one game but i've noticed some behavioral changes of note since .83.

Monarch/Standard sized custom map/Archipelago-ish/11 civs

1) The Ai isnt exploring early with fishing boats any more
2) I'm seeing a lot of stacks roaming around near cities as if they are waiting for transport but transport is not built
3) Some AIs seem to build lots of boats and Some AIs dont build any at all. I've had a gally blockading the spannish capital for 3000+ years and while the Ai walks land stalks in front of it not a single boat has been built
4) The ais that do build boats build and attack with gallys
 
Top Bottom