AND2 and SVN Bug Reports - A New Dawn 2 ONLY

[Rezca]

That's it. Go and steal their advanced techs; change their civics/religion (when they change back, that likely results in anarch for them); aid rebel factions (with a constant effort you may brake an empire in halves or even smaller parts).

Except technology stealing loses it's value after the Medieval and until the early Industrial, unless you can really afford to push 40 ~ 60 % on the Esp slider for a dozen or two terns per tech against one AI. High Level espionage like that isn't really worth it until Jails and the like begin appearing. If you can get Fascism early then that's great.

I think forced Religion change also had its espionage count like, quadrupled or more so I've never used it. If I can pay them to switch I will. What's 700g or an out of date tech compared to losing forty turns of Espionage saving?


I think city revolts (In certain cities, though it only lasts a single turn) and if you can afford it, Anarchy missions - are the best way to go.

The problem I've seen with the costs for Stealing Techs isn't exactly their cost in themselves, but of how rapidly they spike upwards in that cost while the rate you can get Esp points remains low as before. You're going through the Classic with maybe 2,500 ~ 6,000 for a tech, then they magically spike up to 16,000 ~ 45,000 or more once you send them into someone's territory during the early Medieval, and it just goes up from there, while your empire still produces a meager amount of espionage. The science rate will go up eventually, but unless you focus your entire empire on producing espionage - which will end up further crippling your research and position on the tech tree - you aren't going to catch up to that new massive leap in esp costs
.
There's not a lot of espionage boosting buildings before Jails and the like hit the scene, so it's pretty disheartening to see half the world with a tech you won't get for 30 ~ 40 turns and no one - even your best of friends - will consider trading it with you, and you can't steal it nor buy it. Well at least it is for someone like me, who used and abused spies almost all the time in BTS. Almost always had my Espionage slider at 10 ~ 25 % at all times when I found out how much fun Advanced Espionage was

Takes some getting used to I know, but I've yet to find a reliable way to get a good or decent espionage income after the research cost spike that doesn't involve crippling me elsewhere.

 

[45°38'N-13°47'E]

Next changes I've planned are reducing tech cost for Middle Age and cutting down a bit esp costs.

 

[gay_Aleks]

I use primarily spread culture mission, really. Do revolts from time to time, but they're not exactly of use, since I use them on already useless border cities, and I am not on Rev.

 

[ciadude2]

Thanks for the advice guys... ye know in all the time I've been playing this mod(since the original RoM) I've never actually bothered with spies, but I'm going to try that now.

I do have +150~ espionage points per turn though so I have a bunch to work with.

 

[Afforess]

Thanks for the advice guys... ye know in all the time I've been playing this mod(since the original RoM) I've never actually bothered with spies, but I'm going to try that now.

I do have +150~ espionage points per turn though so I have a bunch to work with.

RAND adds a lot more extra spy missions, it makes it a bit more interesting.

 

[Rezca]

RAND adds a lot more extra spy missions, it makes it a bit more interesting.

My favorites are bribing a city (If only to force the AI into war with me) and planting nukes - though part of me kinda wishes they could do the same to me. Seems almost like a subtle 'cheat' for the player since the AI tends to always focus on sabotauging buildings and improvements (And often the buildings they pick are easily replaceable or not important)

One of my favorite things to do is bribe an enemy city, then promptly sell it to a rival for a few techs or a truckload of cash before the former owner can move their units back into the city. Often they'll take me up on that offer and then the city revolts due to culture or they gift it back to the original AI. I haven't done this in a while, but maaan it sure feels pleasing when it works.

 

[RyoHazuki]

My favorites are bribing a city (If only to force the AI into war with me) and planting nukes - though part of me kinda wishes they could do the same to me. Seems almost like a subtle 'cheat' for the player since the AI tends to always focus on sabotauging buildings and improvements (And often the buildings they pick are easily replaceable or not important)

Thats actually a "problem" for me, when the AI dont use something i can use, or when the AI cant do something i can do, i feel im cheating and i simply dont do/use it. Spy missions are an example, i would love to make revolts all around the world, but i feel it unfair if Ai dont use it, making the game easier.
Same happens with barbarian workers, which are too easy to capture (they dont always scape when you are close), and extremely valuable if you sell them to another civ.

 

[Rezca]

Thats actually a "problem" for me, when the AI dont use something i can use, or when the AI cant do something i can do, i feel im cheating and i simply dont do/use it. Spy missions are an example, i would love to make revolts all around the world, but i feel it unfair if Ai dont use it, making the game easier.
Same happens with barbarian workers, which are too easy to capture (they dont always scape when you are close), and extremely valuable if you sell them to another civ.

That's why I haven't used the Anarchy missions in a while: the AI has difficulty recovering from it unless they've got a nice solid empire, and they never use it against me - if they can, they haven't yet.

As for workers, I still see AI fooling around with them. I'll move some military next to them, and the worker will end its turn there. If I end my next turn without capturing it, THEN and only then will they move it to safety. Yes, sometimes I see workers ducking for cover when I have an approaching army, but if I manage to get a military unit beside a worker before it can run, he just sits there for a whole turn before trying to run more often than not.

 

[Afforess]

Thats actually a "problem" for me, when the AI dont use something i can use, or when the AI cant do something i can do, i feel im cheating and i simply dont do/use it. Spy missions are an example, i would love to make revolts all around the world, but i feel it unfair if Ai dont use it, making the game easier.
Same happens with barbarian workers, which are too easy to capture (they dont always scape when you are close), and extremely valuable if you sell them to another civ.

That's why I haven't used the Anarchy missions in a while: the AI has difficulty recovering from it unless they've got a nice solid empire, and they never use it against me - if they can, they haven't yet.

So what you are both really saying is that the AI for spies needs some work?

I can look into it.

 

[Rezca]

So what you are both really saying is that the AI for spies needs some work?

I can look into it.

Pretty much

I think I mentioned this before when Espionage got brought up some time ago, but I have *never* once - in my entire time playing AND or pre-AND - seen the AI ever use the following on any difficulty:

*Bribe City
*Assassinate missions (When enabled)
*Planting Nukes
*Disable Power Grid
*Spread Anarchy
*Forced Civic/Religion change
*Steal Technology (In standard BTS I saw this often. AND? Once and only once, and it was before the massive Medieval Era tech cost increase)
*Sabotage Research
*Remove Corporation/Religion (Why can't we force a corporation out of our *own* cities?)


What I see most often:
*Poison Water (Not so often anymore for some reason)
*Sabotage Improvement (Ever notice how they pick something that isn't exactly useful most of the time or a resource you have like, ten of?)
*Sabotage Building/Production
*Counter-Espionage Mission
*War Rallies


I'm not so sure about Spread Culture, City Revolt, or Aid Rebel Factions. If they have used them, I probably missed it or passed it off as something else.

Bribe City *seems* powerful, but you can just move a unit back into it on the very next turn, so I don't see how an AI could use it to ruin you like an Anarchy mission could. Just step your units back into it on the very next turn

 

[45°38'N-13°47'E]

Try tweaking the XML constants. I'd lower the welfare threshold to 80 (from 90) and try increasing the modifier to 1000.

As I said it may need tuning.

Well, I forgot that the modifier gets divided by 100, hence that constant is more in the 10000-15000 range. I need one more day, I think.

 

[Afforess]

Well, I forgot that the modifier gets divided by 100, hence that constant is more in the 10000-15000 range. I need one more day, I think.

It doesn't get divided by 100... 10000 will likely result in some absurd results.

If you're frustrated with some AI falling behind, I had a brilliant inspiration for a quick fix for this. Flexible Difficulty, but for the AI as well. Basically an additional option to toggle on Flexible Difficulty for non-humans. The result should be that the bottom AI players all get lower difficulties, and a boost to their economies as a result.

By Default, all AI's are locked at Noble difficulty. So this would change that.

 

[45°38'N-13°47'E]

It doesn't get divided by 100... 10000 will likely result in some absurd results.

If you're frustrated with some AI falling behind, I had a brilliant inspiration for a quick fix for this. Flexible Difficulty, but for the AI as well. Basically an additional option to toggle on Flexible Difficulty for non-humans. The result should be that the bottom AI players all get lower difficulties, and a boost to their economies as a result.

By Default, all AI's are locked at Noble difficulty. So this would change that.

int CvPlayer::calculateBaseNetResearch(TechTypes eTech) const
{
	TechTypes eResearchTech;

	if (eTech != NO_TECH)
	{
		eResearchTech = eTech;
	}
	else
	{
		eResearchTech = getCurrentResearch();
	}

[COLOR="Red"]	return (((GC.getDefineINT("BASE_RESEARCH_RATE") + getCommerceRate(COMMERCE_RESEARCH)) * calculateResearchModifier(eResearchTech)) / 100[/COLOR]);
}

This is the research that's displayed in the main interface upper left corner. Well, actually it's

int CvPlayer::calculateResearchRate(TechTypes eTech) const
{
	int iRate = 0;

	if (isCommerceFlexible(COMMERCE_RESEARCH))
	{
		iRate = calculateBaseNetResearch(eTech);
	}
	else
	{
		iRate = std::max(1, (calculateBaseNetResearch(eTech) + calculateBaseNetGold()));
	}

	return iRate;
}

But it looks to me like it's getting divided by 100. Or am I missing something? I'm trying now and maybe 10000 is too much but 1000 is definetely to little.
Flexible difficulty for AI could be an idea but doesn't it make things a bit too complicated to balance?

 

[Afforess]

int CvPlayer::calculateBaseNetResearch(TechTypes eTech) const
{
	TechTypes eResearchTech;

	if (eTech != NO_TECH)
	{
		eResearchTech = eTech;
	}
	else
	{
		eResearchTech = getCurrentResearch();
	}

[COLOR="Red"]	return (((GC.getDefineINT("BASE_RESEARCH_RATE") + getCommerceRate(COMMERCE_RESEARCH)) * calculateResearchModifier(eResearchTech)) / 100[/COLOR]);
}

This is the research that's displayed in the main interface upper left corner. Well, actually it's

int CvPlayer::calculateResearchRate(TechTypes eTech) const
{
	int iRate = 0;

	if (isCommerceFlexible(COMMERCE_RESEARCH))
	{
		iRate = calculateBaseNetResearch(eTech);
	}
	else
	{
		iRate = std::max(1, (calculateBaseNetResearch(eTech) + calculateBaseNetGold()));
	}

	return iRate;
}

But it looks to me like it's getting divided by 100. Or am I missing something? I'm trying now and maybe 10000 is too much but 1000 is definetely to little.
Flexible difficulty for AI could be an idea but doesn't it make things a bit too complicated to balance?

Ok yes, at the end 100 is divided, but 10000 will lead to tech diffusion giving 100X more free beakers. That is too much IMO.

I don't think Tech Diffusion alone can help out civs that fall behind. Instead of focusing on just tech diffusion, we should try and determine what exactly leads to run-away civs taking over.

It might be worthwhile to play a game, let some players fall behind to where they are unable to catch up, and upload the save. Then we can switch into the AI's player and look at what they are doing.

 

[45°38'N-13°47'E]

Ok yes, at the end 100 is divided, but 10000 will lead to tech diffusion giving 100X more free beakers. That is too much IMO.

I don't think Tech Diffusion alone can help out civs that fall behind. Instead of focusing on just tech diffusion, we should try and determine what exactly leads to run-away civs taking over.

Well, from my experience in past months, I've seen that once some AI gets ahead, other civs will fall behind at an increased rate. Mainly I think because tech leader is usually getting free great people and can build wonders which makes it even stronger or more advanced. With my old clunky code for tech diffusion, civs falling behind were getting a help to keep up the pace of more advanced civs; I'm trying now with 9000 and lowering threshold to 80-85; probably modifier can be lowered even more; 100x is too much, right now it's going to 80x in my current test and it could be still too much. But think about it: if there's a civ in middle age while I'm at the end of Renaissance (as it's happening now), if we want that civ to get closer, it will need a multiplier definetely above 10x or more. Using 1000 as multiplier, you get 8x max. That's definetely not enough if a backward civ is doing +100 while the leading civ is doing +3000 It happens that when a backward civ meets a tech leader it will greatly increase science output for 1-2 turns, then it will slow down again because of the scaling of your formula.
From my experience, AI wasn't falling behind too much anymore even with my old code. Yours is 1000% better because it's more gradual.
It's ok to me if we want to explore other solutions, but I think we can solve the problem easily with current code. If you can find a better solution, that's more than ok with me, but I can't help you there, I can only test your solution.

 

[Afforess]

Well, from my experience in past months, I've seen that once some AI gets ahead, other civs will fall behind at an increased rate. Mainly I think because tech leader is usually getting free great people and can build wonders which makes it even stronger or more advanced. With my old clunky code for tech diffusion, civs falling behind were getting a help to keep up the pace of more advanced civs; I'm trying now with 9000 and lowering threshold to 80-85; probably modifier can be lowered even more; 100x is too much, right now it's going to 80x in my current test and it could be still too much. But think about it: if there's a civ in middle age while I'm at the end of Renaissance (as it's happening now), if we want that civ to get closer, it will need a multiplier definetely above 10x or more. Using 1000 as multiplier, you get 8x max. That's definetely not enough if a backward civ is doing +100 while the leading civ is doing +3000 It happens that when a backward civ meets a tech leader it will greatly increase science output for 1-2 turns, then it will slow down again because of the scaling of your formula.
From my experience, AI wasn't falling behind too much anymore even with my old code. Yours is 1000% better because it's more gradual.
It's ok to me if we want to explore other solutions, but I think we can solve the problem easily with current code. If you can find a better solution, that's more than ok with me, but I can't help you there, I can only test your solution.

I think your solution is very pragmatic. I am just hoping we can find a more holistic approach.

 

[ciadude2]

Spent the better part of the day playing a new game in rev689 as Mehmed II
Flexible difficulty - 25 turn intervals with the max set at the difficulty right under deity.
I've been leading the score for most of the game by a little bit(less than 100 points over 2nd place) so I would presume that I am at the difficulty level ceiling.
Map settings: Large fractal, starting ancient, speed normal.
Tech diffusion off

So far a very interesting game. I made straight for the wonder that gives all of the religion civics, switched the atheism while expanding my economy and number of cities as quickly as possible. Stopped at 9 cities when I ran out of room to expand. Have been pretty much just trying to crank up my science and economy as much as possible while keeping my military at such a level as to make me not look like an easy target. When I saved I was almost finished researching flintlock(the others are just barely getting gunpowder, but they have a few techs I don't have so we're fairly even, I just got lucky hitting liberalism and physics first to give me a boost into that line of techs).

As of present, there are around 8-9 civs that are all fairly close together at the top in terms of score and power. I have the top score by about 150 points, and have started cranking my military production up bringing me to .88 the power of the strongest civ. I am 1.2X stronger than my eastern neighbor Babylon who has 7 cities, and they are warring Spain(the strongest civ in the game currently). Although I have quite a few bombards, elephant bombards, and arquebuisers all at decent experience(garrison + volunteer army ftw) I'm waiting to flintlock to upgrade them all to Janissaries before I invade.

My plan is to split the army into 3 groups. The largest will head straight to Babylon, which is 2 tiles from my nearest border city. The second largest will be invading from the north to tackle their border city and their main army(since that is also where Spain is fighting them). The smallest group will head south of Babylon(since it is dead center of the border area) and intercept any troops they send to either reinforce the city or attack my city, and possibly to take cities to the south if they're weak enough.

I don't expect the entire invasion to take that long assuming all goes well. One that's taking care of, I'll wheel the entire army around northwestward to invade England. The English are quite a bit stronger but I'd really like to kill them off before they get Redcoats. Following that, I plan on sweeping northward to take out the Americans and then the Incas(who are both fighting the English). Once I subdue all of these countries, then I plan on coming back to the south east and invading Spain, and then Rome. Once that is done, the continent is essentially mine(there are 2 major continents).

After that I'm going to try and further press my technological advantage and launch an invasion of the weakest country on the second continent, at which point I will hopefully have flight so reinforcing my armies there will be much simpler.

Oh also, the mod seems to be loading much faster and playing much more stable than it did in previous versions, both in loading up and in going from turn to turn. In the last version prior to SVN by this point I would usually have been waiting around 45-60 seconds in between each turn. At present I am waiting around 5-10(prior to the Renaissance it was almost instantaneous).
So keep up the good work. I'm having loads of fun here.

 

[45°38'N-13°47'E]

I think your solution is very pragmatic. I am just hoping we can find a more holistic approach.

Well, since I can't take your pace in modding , I've just uploaded your tech diffusion code with the new display in the main interface. I'll need some time to tweak it, but in the meantime I'll try your solution.

 

[ciadude2]

So in that game I mentioned above, I'm getting a crash on turn 561, not sure why. It's a shame because I'm 1-2 turns from taking the last major English city lol.

Edit: updated to rev 692 and when loading up the mod, I get a weird pop-up about the Mosque of Djenne(no error message just it says "The Mosque of Djenne"). The mod loads fine other than that so I can't see any problems caused by it.

Edit2: rev 693 no longer has the Mosque of Djenne pop-up

 

[Hyzoran]

I'm on turn 701, and when I end my turn, game endlessly processes the AI turns and never ends the turn. This is reproducible every turn and im worried that my save might be destroyed. I enabled debugging in BUG and spams 'BugPath - Initializing...' in the debug messages. Although I am running on a Giant-sized map full of civs, I wouldnt think this would cause this to just endlessly calculate the turns like this.

Here is the link to my save:

https://dl.dropboxusercontent.com/u/93583028/Jesse BC-4750.CivBeyondSwordSave

Is it a bug with RAND(2)? Does anyone have any suggestions to how I can fix this?