Report Questionable Behavior

Maniac said:
Granted, this doesn't happen in vanilla Civ (but it can happen in Planetfall for instance), but I see a problem when the only GPP modifiers a faction has are the city's getGreatPeopleRateModifier(), and almost all cities have the building giving the GPP modifier.
Click to expand...

Example: Roman Forum (+25% great people).
 
On SS launches.... CvPlayerAI::AI_launch

Change this
Code: 
	for (int iTeam = 0; iTeam < MAX_CIV_TEAMS; ++iTeam)
	{
		if (iTeam != getTeam())
		{
			CvTeam& kTeam = GET_TEAM((TeamTypes)iTeam);
			if (kTeam.isAlive())
			{
				int iCountdown = kTeam.getVictoryCountdown(eVictory);
				if (iCountdown > 0)
				{
					if (iCountdown < iBestArrival)
					{
						iBestArrival = iCountdown;
						eBestTeam = (TeamTypes)iTeam;
					}

					if (iCountdown < GET_TEAM(getTeam()).getVictoryDelay(eVictory) && kTeam.getLaunchSuccessRate(eVictory) == 100)
					{
						bLaunch = false;
						break;
					}
				}
			}
		}
	}

to this:
Code: 
	for (int iTeam = 0; iTeam < MAX_CIV_TEAMS; ++iTeam)
	{
		if (iTeam != getTeam())
		{
			CvTeam& kTeam = GET_TEAM((TeamTypes)iTeam);
			if (kTeam.isAlive())
			{
				int iCountdown = kTeam.getVictoryCountdown(eVictory);
				if (iCountdown > 0)
				{
					if (iCountdown < iBestArrival)
					{
						iBestArrival = iCountdown;
						eBestTeam = (TeamTypes)iTeam;
					}
/********************************************************************************************************************************************************/
/* BETTER_BTS_AI_MOD                      		07/12/09                                Ricardo Rolo ( aka r_rolo1 )     			*/
/*                                                                                              							*/
/* Removal of clause that blocked AI SS launch if there is a a 100% safe SS on air faster than theirs                                                   */
/********************************************************************************************************************************************************/
					//if (iCountdown < GET_TEAM(getTeam()).getVictoryDelay(eVictory) && kTeam.getLaunchSuccessRate(eVictory) == 100)
					//{
					//	bLaunch = false;
					//	break;
					//}
/********************************************************************************************************************************************************/
/* BETTER_BTS_AI_MOD                       		END                                                  						*/
/********************************************************************************************************************************************************/
				}
			}
		}
	}
Given that there is always the possibility of a ( normally human ) raze to the capital of the civ that has the ship already in the air, that safeguard is actually conterproductive to the AI if it has a SS to launch that would not get there before the already launched one in normal circumstances....
 
I played BBAI for the first time after a while last weekend and here are my observations. Not everything I list is necessarily "questionable" but I had a hard time deciding where to post so forgive me.
This was a Prince game, normal speed, a standard size-pangea-like custom map, standard settings.
1. AI built WAY TOO MANY ships. When Zara declared on me, he had about 3x advantage in the power graph but I quickly found out that about 80% of his power was destroyers. He eventually sent them against me piecemeal to kill fishing boats but didnt matter much in a pangea map. I ended up winning the land war suprisingly easily. This is doubly questionable beacuse he declared on me counting on the power of his destroyers.
2. Caesar desperately needed oil, which lay just on the other side of the border on my side. We were same religion but at "cautious" almost all game and he had many opportunities to backstab but... you guessed it, he sat on his huge army WHOLE GAME.
3. The AI doesn't take advantage of its unique units. Some of the earlier UUs are pretty powerful. It would also add to the roleplay/flavor aspect if AI was more aggressive when its UU is in play.
4. AI still did a poor job going for victory towards the endgame. At least two AIs were in good shape tech-wise but were suprising slow to start constructing the spaceship. They just emphasize units, which is wise if you eventually use them but wasted if you just sit on them.
5. At Prince, AI builds a tad too many cities. At higher difficulties this is probably good for AI with all the bonuses it gets, I dont know for which difficulty better AI is supposed to be optmized for.
Just my two cents....
 
solanacea said:
I played BBAI for the first time after a while last weekend and here are my observations. Not everything I list is necessarily "questionable" but I had a hard time deciding where to post so forgive me.
This was a Prince game, normal speed, a standard size-pangea-like custom map, standard settings.
1. AI built WAY TOO MANY ships. When Zara declared on me, he had about 3x advantage in the power graph but I quickly found out that about 80% of his power was destroyers. He eventually sent them against me piecemeal to kill fishing boats but didnt matter much in a pangea map. I ended up winning the land war suprisingly easily. This is doubly questionable beacuse he declared on me counting on the power of his destroyers.
2. Caesar desperately needed oil, which lay just on the other side of the border on my side. We were same religion but at "cautious" almost all game and he had many opportunities to backstab but... you guessed it, he sat on his huge army WHOLE GAME.
3. The AI doesn't take advantage of its unique units. Some of the earlier UUs are pretty powerful. It would also add to the roleplay/flavor aspect if AI was more aggressive when its UU is in play.
4. AI still did a poor job going for victory towards the endgame. At least two AIs were in good shape tech-wise but were suprising slow to start constructing the spaceship. They just emphasize units, which is wise if you eventually use them but wasted if you just sit on them.
5. At Prince, AI builds a tad too many cities. At higher difficulties this is probably good for AI with all the bonuses it gets, I dont know for which difficulty better AI is supposed to be optmized for.
Just my two cents....
Click to expand...

I think this highlights the need for the differentiation of the 4 types of power. If the AI considered land, air, naval and nuclear power separately, I think it could focus on unit production much better, and would be able to evenly match and attack enemies better. What do you think Jdog?
 
solanacea said:
1. AI built WAY TOO MANY ships. When Zara declared on me, he had about 3x advantage in the power graph but I quickly found out that about 80% of his power was destroyers. He eventually sent them against me piecemeal to kill fishing boats but didnt matter much in a pangea map. I ended up winning the land war suprisingly easily. This is doubly questionable beacuse he declared on me counting on the power of his destroyers.
Click to expand...
Whereever this problem is coming from, and I don't think it's the inability to consider sea power seperately - I actally think the AI can do that - that problem is not really old but it definitely exists now.
If my suspicion is correct and it was only added rather recently, reverting should be easy enough. Or so I hope.
 
karadoc said:
I'm not Jdog, but I agree with that. I think it is particularly important to distinguish between land and naval power; not just to help choose what to build, but to choose how to fight the war.
Click to expand...

I'm also not jdog, and I also agree. :D

Power considerations are sometimes going to be worse than none at all if they don't separate the different types of power, as demonstrated in solanacea's example.
 
In my current game (immortal, team game), the AI (Shaka) has built around 35 ships (galleons, caravels), and almost no land unit (compared to the 'usual' Shaka...) It was really easy to steamroll his land (big continent) - and he never used the ships, not even for pillaging.
I have never seen this extreme behaviour on big landmass maps before.
 
PieceOfMind said:
I'm also not jdog, and I also agree. :D

Power considerations are sometimes going to be worse than none at all if they don't separate the different types of power, as demonstrated in solanacea's example.
Click to expand...

I'm also not jdog, and I approve this message. :)
 
In my last game (Immortal Pangea Epic Huge 18civ betterAI!) I declared war on my neighbour Mansa Musa in lategame, and he made serious mistakes with guided missiles and bombers. Sadly I have no appropriate screenshot or save now.

Anyway, he had a fort right outside my border at the egde of his territory, protected with 3 infantries. This fort was the only way his missiles could harm my closest city, that might be important.
On his first turn he moved 1 bomber and 3 missiles there. On my turn I killed the infantries and got his missiles and bomber for free. So far so bad.

After that I moved my SoD further into his territory, and Mansa kept moving missiles into the now completely unprotected fort! Within the next 5 turns I got 8 missiles for free, simply by moving units onto the fort-tile. Whenever none of my units were in that tile for 1 turn (but right next to it) he would immediately move even more missiles.

Please look into this, I know that leaving threatened cities has been an issue before, but somethings wrong with forts now.



Might add that Mansa did a great job with spies and artillery, and the war was very tough despite his slipups.
 
In my previous game, Shaka who is half the globe away, thought it was cost effective to wage a war with me (Dutch). He took me completely by surprise because his direct neighbor (Louis, with whom he already had war) was an obvious attack choice. A human player would kill/capitulate Louis first and take into consideration several factors for selecting his prey (distance, possible countermeasure moves, unit supply costs, possible gain, technological lead, power difference). Even if I had half his power, it would not be cost effective because he would have nothing to gain from cities so far away. He didnt take into consideration my counter moves, which were nationhood>draft like crazy rifles against macemen.
But the worst part is that he could take my city (the one in the screenshot), had he pressed on the attack. It would be a nice explanation if he thought that he wouldn't be able to hold it.
Last but not least his stack pillaged a matured town. You certainly dont do this when you plan to hold the city (which is what the AI do) and not raze it.
Had he been a little more careful choosing his opponent I would be quite impressed with the capabilities of the AI. Especially the 2 fronts move was quite good.

 
While that's not a great move, I would say that there should still be a chance of it happening, especially from nuts like Shaka... so long as the odds that he chooses to hit France are at least 10x the odds that he chooses you (assuming similar hatreds) I think it's ok that there's a low prob outcome that is poorer, just to keep the player on his toes (otherwise you could too easily say "I know he'll never attack me, I don't have to worry"). Now, I don't know if this is the case here at all, so it might still use some looking into, but if this is a one time thing and not a recurring pattern I would say it isn't necessarily problematic. But that's just my opinion on it..
 
I'm tired of reading about human players pwning Dariuses and Willems with Electricity who still think that Muskets are acceptable city defenders and that there's no need to tech these brutish, ungentlemanly Rifles.

Human player beelining Cavs/Rifles usually can exploit the way peaceful AI's ignore Rifling for a long time.
 
Catching up a little bit:

I am jdog and I agree, the AI was building way too many boats. I also just found the source of this bug and fixed it, so when 0.83 is officially released (a couple days?) you'll see much better behavior ... there are several new fixes I found in the last couple days which are small things with significant effects.

I'll add checking out missile movements to my list.

For war target choices, there are a lot of potential causes for why Shaka would pick you over the "logical" target ... we do want the AI to be unpredictable within some constraints. One issue which I've been thinking of changing is that if the AI has just ended a war with another player, it will not consider starting up war plans against that player until the peace treaty is over. This can cause exactly what you saw, the AI picking a lower value target to start planning war against. Since the AI usually spends more than 10 turns building up its preparations, it should still consider players it has a peace treaty with in my opinion.

As for some AIs delaying rifling too long, I'll add that to the list of specifics to check out when I redo the way the AI chooses techs.
 
jdog5000 said:
...Since the AI usually spends more than 10 turns building up its preparations, it should still consider players it has a peace treaty with in my opinion.
Click to expand...

Certainly. In the later game (from the classical age and onwards), wars can last throughout several cease fires (but that may be a part of another AI mechanic altogether). I say that it should have a higher possibility of continuing the war with the civ it already had it, even higher if the civ is a direct neighbor. It would maximize the benefit, as at least one stack has already made its pass. More stacks, mean concentrated attack against a weakened civ->more cities/land to grab->profit. This of course cuts both ways as the attacked civ is more likely to build up defences in those precious peace turns and thus caching of guard another civ would be good.

Also in my current game Joao attacked me with carracks which have a cargo space of 2, and thus is completely counter productive to wage intercontinental war, especially when I had a fully developed fleet of galleons/caravels and frigates just of the corner. They could be used however for pre Astronomy for free land grabbing (though I'm not sure how this could look for the economy, perhaps affordable when colonizing). Especially in this case there was practically a free continent.
 
phoinix said:
Last but not least his stack pillaged a matured town. You certainly dont do this when you plan to hold the city (which is what the AI do) and not raze it.
Click to expand...

All things being equal, there's a 50/50 chance that the AI will take and hold the city, and then the town will eventually be useful. By contrast, pillaging gets the AI certain gold now. That's normally going to be the right choice.
 
I'm not sure if the this is the right way to watch the AI in action, but in some AI autoplay games, I noticed that when the AI went into anarchy for civics and stuff, they adjusted their commerce, usually increasing espionage by 20% or some other stupid action. They corrected it the turn after anarchy, but still, this is pretty dumb of the AI, and one turn of wasted research is still important. I don't think the AI should be doing the commerce adjustment in Anarchy, since most of the income/expenses calculations will be funny in anarchy. What do you guys think?
 
You must log in or register to reply here.

Similar threads

Anarcomu
  • Locked
Bug discussion before report (on Github)
Replies
8
Views
590
Recursive
Recursive
rahorvat
DOS Civilization 1991 Bug(s) discussion - What should be fixed in OpenCiv1?
2
Replies
30
Views
1K
rahorvat
rahorvat
Eyswein
  • Sticky
6otM180 AAR
Replies
7
Views
756
Monthar
Monthar
Tiramisu
[BNW] Bug? AI receiving thousand gold pert turn
Replies
3
Views
362
Tiramisu
Tiramisu
Eyswein
  • Sticky
6otM179 AAR
Replies
9
Views
1K
Soger
Soger
Top Bottom