Buffed AI for BNW

So I have played halfway through the reinnasance era now, 200 turns, emperor level with small map, 6 civs.

Poland have conquered Hiwatha and have 10 cities. Atilla have wiped out Bismarck and have four cities. And they did it very quickly. The AI is very aggressive. And strong. Me, on the other hand, had to fight with bits and nails to conquer a Venetian city state! I think I used 6 trebuchets, 4 composite bowman, one pikeman and one musketman. And he had like only two, three units, and of course, his deadly city bombardment. They are very tough to kill, and they do serious damage. I find it nevertheless very much a new fun experience. It took me a fair ammount of time to get that city.

One city bombardment and one ranged composite bowman attack from the AI; and a unit of my own is killed. The other way around, and I need maybe 4-5 units to kill one of his. So, maybe a very tough ratio. Because my units were as well, a lot stronger. Maybe tune it down just a notch. But at the same time, does this 25% buffed AI, snowball when he get stronger units?

But it has been still very entertaining, I have to be up to war whole the time. Because if they surprise me, I'm dead. So I have, for the first time, buildt forts around my cities. And buildt a lot of units, to be prepared for a backstab. I never felt the need for them before, but now, it's a war zone out there At the same time, the AI is a lot "smarter", and make things it has never done before, which is great. Because I can't exploit longer. So that's certainly very good!

Tech wise it's very close. Science output very close. Wonder building very close. So it's a close game. It's like the AI maintain a good development, instead of having a strong start and then losing it.

The main period to watch out is the industrial era onwards for science. If the AI pulls ahead in science post industrial, let it do that to see whether it can pull of a competitive victory that you would expect on emperor.

As for the buffs and nerfs to combat, they can be tuned later, because they are independent of the economic game from the AI's point of view, but merely cause us humans to spend more on military, dipping deep into the bottom half of the tech tree (I doubt that skipping it is wise), thus equalising the economic/cultural/scientific game. The idea with the combat buffs is that the AI should roughly be loosing as many units in warfare as you (on average) on whatever difficulty level you think is fair (which mimics equal skills)

Capturing cities in this mod needs a lot more thought than vanilla. Why? Because vanilla was supposed to be as complicated as this mod is tactically, but the AI spoils it, making the entire game unbalanced....

As for taking down a city, I find that the extra cover promotion is sufficient to limit the amount of siege needed to 2-3 at most. The trick with taking a city in this mod is to study the surrounding terrain, to ensure your army is sufficiently experienced, and to pick your cities carefully (and be prepared to pull out from strategically suicidal campaigns - yes you can get away with it on vanilla but not here). Do we have to build roads to the war front? Do we need workers to clear forests/jungle so that our siege have clear sight of the city walls? Forget about using composite bows to take down the city as per the vanilla game. You need melee, letting them take the punches and cycling them in and out of the front line combined with focused siege fire. One properly experienced swordsman will do as much damage or more damage to a city than a catapult hit will. Ranged bows are there to clean up and to soften units up, they have not the strength to take down the city and often not enough to kill units, unless they concentrate fire (but they also should last longer and thus get more experience). You also need a general most of the time behind the front line so factor that in. Remember that the AI will fire at the damaged units as it's priority, which gives your other units time. Use cavalry to distract the AI, to flank it and to try as best as possible to clean up the AI's undefended ranged units, being willing to loose cavalry for the sake of a clean attack on the city by your melee/siege. You must pillage at every opportunity. In this mod, AI workers are very efficient at improving the terrain, which also means that humans have more pillaging opportunities.

Still, you may be correct that the AI is too tough for emperor, we'll see over time. The point is that tactical skill should be as complicated as the other skills this game demands against competent opponents, and that emperor is already supposed to be a skilled level....

EDIT: Dekraa, are you playing at standard or slower speed? I have no idea how balanced this mod is at quick speed....

TIPS:

• Keep in mind that human ranged-3 units only get random chance at damage at range-3
• Take serious notice of what your military advisor is telling you, unlike in vanilla, what he says means something. If he is saying the enemy has overwhelming forces he is probably right. If he says that military is similar, it does not mean that you can walk over the AI any time you like, it means that if you take on the AI at similar strength, stagnation is possible.
• In this mod you have to take seriously neighbours that have overwhelming forces and play the arms race game to some degree and judge whether they are preparing to attack another AI or you. This is not always a simple judgement.
• Keep in mind that counter-attacking in this mod matters. If an AI is really strong, retreat and let it take your city, stay at war, beeline a military tech and counter-attack especially if you have a science lead. Often the AI will become scattered and disorganised once it's primary objective has been satisfied. Once the AI's military force is spent, human nerfs and AI buffs matter less.
• Try not to attack a city without first having trained units (makes sense yes?) It is ok to pillage the AI with inexperienced units, but against cities the units need to have promotions.

- Emperor
- Standard speed
- Small map

Reaching the modern time. The AI has already defeated three other civs. So there's only me and two civs left. The AI is very efficient of wipe out other civilizations. I never seen something like this.

I'm still ahead in science race by a pair of techs. But the biggest civ, 9 cities, (against mine of 4), has an output of 250, I have only 200. It would be interesting to see if he will pull ahead now when going into late game.

Hi Dekraa well done for reaching the latter eras! The AI get's exponential science bonuses per era like this (the numbers are a preliminary guess):

0% Pre-industrial
10% Industrial
30% Modern
60% Atomic
100% Information

All the best with the game, enjoy Hey, if the AI is efficient at wiping out other civs, it might even win a domination game?

EDIT: I'm thinking about changing it in the future to nerf the human per era rather than buffing the AI, but that would slow the game down so it would take more turns to finish (which could be a good thing).

How is the nerf calculated?

Nerf Human Strength modifier = Work both way, when you attack, and when you defend?

Nerf Human Ranged modifier = Siege units and range units only? Do this modifier only work when you attack with those units? And when you defend from attacks it is calculated from the Nerf Human Strength unless it's a ranged attack?

I'm a little confused, shine some lights on the matter are you kind

Why do you have a modifier for the AI vs AI ? So they can win a domination? Well, I would say that the rate in my game, Poland would have got domination around turn 350.

I have a feeling that the AI is on equal foots with tech, but I must say that I would want the AI to have faster science eventhough. But more cost to build things. Because it feels like the AI can construct a wonder in no time.

Hi Dekraa:

• The nerf to human strength simply shaves x% off the units strength before all the other modifiers are applied. It applies to all human units and cities.
• The nerf to human ranged is simply to shave off y% off ranged attack before all the other modifiers are applied. It applies to all human ranged units and cities.
• The AI offensive attack modifies the calculation of the attack based on the units strength or ranged attack so that the AI's melee and ranged attack are z% stronger than they would otherwise be. It applies to all AI melee and ranged units but not cities. It does not discriminate between a human or AI target.

The AI offensive attack is an experiment to see if it helps compensate wide playing AI's that have a lower science rate than a tall AI. This is because tall AI's get to the next era first, and thus are first to get to the next era's science bonus. It might also help the AI to finish an attack and thus suffer less if is should fail. It should help AI's capture city states as well.

From your game, I gather you think this?

• The science bonus is ok
• The AI's building construction bonus is too high
• The AI's combat bonuses are a bit too high

Is that correct?

In the next version I will post shortly, the AI's combat bonus is reduced because it no longer get's free promotions, it must earn them like humans. In the next version, the AI's social policy bonus is eliminated. In the next version, the AI's science rate prior-the-industrial will be lower, as the slight buff it was getting against humans is eliminated.

I think that should help your concerns. AI combat will be a bit less stinging. Because the AI's science rate and policy rate are brought back to level, that should help building wonders before the AI.

For a readable explanation on combat mechanics, see here:
http://forums.civfanatics.com/showthread.php?t=432238

EDIT:
See the screenshots below for a comparison of before and after this mod is applied. *Note* that the numbers are not exactly the same as the V11 version, because the screen shots are from a development version of this mod. If enough people show interest in this mod, I'll show the exact changes in the user interface with text (not just numbers), but for now I'm too lazy.

Before:


After:

I had a long session rapport that I accidentally deleted. Here is the abridged version. Played Persia at immortal/fractal/standard size/standard speed. The start is harder, especially surviving barbarians and stealing workers. Did good diplomacy and built 4 cities and conquered 3 more. The siege for the Brazilian wonder spam capital with great wall was tough, lost 5 units. After that it was easy sailing and I wont a lazy science victory on turn 266 without much effort (I usually win science around 240 if I aim for it). The problem with the science modifiers coming in the later eras is that a decent player can win before the AIs even get to modern. In my game Korea was the first AI to reach modern at turn ~240 and he stole a crazy amount of techs from me. The science situation has not changed noticeably. You still get a large tech lead after universities, first ideology, choice of wonders and military dominance from the renaissance onwards. The AIs need at least 30% more science throughout the game at immortal, especially if you remove the starting technologies. To sum up: the start is harder/better but the endgame needs more attention.

Thanks chumchu. How many AI neighbours did you have? The difficulty of this mod is threat based. For this mod to work, it needs humans to experience at least one serious major attack on them during the game (not just humans attacking but being attacked), and for that attack to have slowed them down. This mod does not work if there is no serious military problems that the human has to deal with.

EDIT:
In the next version, there is going to be a major but simple change to the way the AI evaluates human military power. In V11, the AI doesn't take the humans nerfs to strength into consideration when it is considering it's DOW probability (it just has a bit more likelihood to DOW humans generally). In V12, ALL THE AI's are going to assess their likelihood to DOW humans based on a real assessment of your actual nerfed power. That should make this mod work better because the AI will more likely slow the human down with its combined military force.

I had one close neighbor (Korea) who started 12 tiles to the east. I managed to steal his worker and redirect his first settler away from me in an early war. I had Brazilians start ~20 tiles north who settled close towards me. And last there was Germany who started ~ 30 tiles northeast who also settled close to me. Germany had the largest army so I bribed him to attack Korea when his troops were hanging around a bit too close to my borders. I then denounced him and got Brazil and Korea to join in that effort as I chewed up his wounded caught between me and Korea. With his army weakened I bribed Brazil to attack him and then I attacked Brazil when our DoF and research agreement ran out. Korea and Germany was a bit pissed of after that so I bribed Korea to attack Germany. The Koreans and Germans coveted my lands throughout the game but after turn 180 I was ages ahead of them. They probably did not dare to attack after that point.

To quote Feynman: "there is room at the the bottom". Most players use the higher difficulty levels and many can consistently beat deity. If the game becomes more difficult with your mod this is a good thing as the best players get more challenge while worse players can move down a difficulty level. I would aim for every level to be roughly one level harder than the original game.

Thanks chumchu. Suggest with this mod to have at least two other neighbours next to you. That is the equivalent of increasing a difficulty level, just by doing that.

EDIT: Next version, humans will not nearly as easily be able to trade for a war on a third party. It's too much of an exploit, see Chumchu's strategy above. The code for that looks like it was rushed out the door. It needs to be fixed.

See OP and change log below it.

The AI will be less tolerant of human weakness now that it properly calculates human weakness into it's war plan. It also cannot be bribed into a war nearly as cheaply. Science ramps up sooner as well and the human nerf is slightly greater. This mod shows it's best dress when there are at least two civs on the same continent as the player.

EDIT: The up side is that some of the other AI bonuses are gone.

Enjoy

You work incredibly fast = ) Thanks for your effort!

glider1 said:

Next version, humans will not nearly as easily be able to trade for a war on a third party. It's too much of an exploit, see Chumchu's strategy above. The code for that looks like it was rushed out the door. It needs to be fixed.

Click to expand...

I've always used one house-rule: I can't try to bribe another civ to attack another unless I'm already at war with the civ I want attacked. I feel like it's already too easy to manipulate the AI as it is. and I can't really imagine Britain (during WWII) desperately trying to get the US to declare war on Germany if they hadn't already been at war with Germany themselves.

Hi Chumchu and Marshall. The bribe exploit in the AI was kinda closed in V12 if you want to download that. If you are happy with the new behaviour of the AI, I'll report the change to the community patch people.

EDIT:
ON second thought, I think I didn't ratchet up enough the AI's asking price for a third party war? I changed it so that only the warring civs (WAR APPROACH > 5) will consider it and their asking price is higher and their asking price escalates more with each era:
WAR_PROJECTION_GOOD: Willing to trade 600 (was 400)
WAR_PROJECTION_UNKNOWN: 1000 (was 600)

Reason I am having second thoughts is that I still managed to bribe Monte into a war on Genghis. He accepted one luxury and 3 horses. The war was a stalemate with neither side benefiting, except me! I took a city off Genghis and have become the diplomatic good guy (even though I was a deceptive @#$).

Perhaps I should up the asking price more and only allow the AI to consider it if they think they will do well out of it (WAR_PROJECTION_GOOD).

Another potential issue is that now the AI factors in human strength properly, I think it might be a bit too keen to go to war on humans because I haven't turned down the AIDeclareWarProbability to compensate.

Only guessing time will tell.

Lately I've been "stuck" playing as Assyria every game because, with other civs, I don't think that early conquering (in most situations) is worth the opportunity costs; better to invest in science and then consider aggression later.

One possible solution I was thinking of would be to greatly increase the gold for conquering a city (maybe double -or even triple it). This could then possibly be used to purchase a library somewhere. Even with Assyria, I tend to fall behind in tech when I'm investing in replacing destroyed units and building military experience buildings (both are essential with this mod). If I was playing as any other civ, I'd be several more techs behind.

I think it should be a difficult decision and one should suffer the opportunity costs of investing in early war; but as it is currently, early conquests may not be quite rewarding enough. It does seem perfectly balanced when playing as Assyria though.

I also have a new idea to further discourage over-reliance on my old strategy of waiting for artillery (dynamite) to be discovered before becoming aggressive. I'll write about it tomorrow in my next post.

I also just want to tell you that this mod is now better than ever and it's giving me the best games of CiV that I've ever had. Thanks again!

Early conquering in this mod is unlikely unless one of these happen:

• You have catapults
• You have enough melee and some ranged and another civ is preoccupied with their own war (then you don't need cats)
• You have enough melee and are capturing back a city state that someone else conquered (because the city state is weakened)
• You are counter-attacking after the AI's attack fails (because they AI cannot defend)
• You have some specific civ based bonus for combat that exceeds the nerf (so you are fighting as if you were not nerfed).

However, early warfare can be good. I did capture a city off Mongolia with the Iroquois using one catapult, a general, two archers, three Mohawks and one spearman without too much trouble during the classical age. Is that early?

I think you are hinting that the strategy range may have narrowed because humans are now always better investing in science first (which they can more successfully do because the AI science rate is lower). To that I would say yes, except that this mod is supposed to cause investment in defence in the bottom of the tech tree which should be slowing the science rate down?

More than happy to hear ideas out because they may be very good!

EDIT:
Keep in mind that for this mod, it is all about changes that buff the AI and nerf humans. If increasing the gold for capturing a city helps the AI more than us humans, perhaps you have point? I worry that the AI will waste the money from conquest and us humans will use it excessively well in relative terms.

PS) Man has this mod has improved the one-more-turn factor of this game that is one thing I can say about it. Even when I am doing quite well, the fear of an AI military buildup keeps me playing. Another thing I like about is that because the AI is not spamming cities as much, there is actually more room on the map to have battles with the AI in open territory without cities getting in the way. I actually had a battle with Mongolia on a large expanse of plains on a standard sized continent map today with three AI's and me on the one continent. We're into the renaissance and three other unknown civs on the other continent are ahead of me on score....exactly what I would hope for on emperor, how it should be.

See this hilarious thread for more about the AI and worker stealing. He breaks the AI by settling on top of their capital. What this points to in my opinion is the problems with zone defense as opposed to point defense regarding the early warriors. The warriors go off on barb hunts and other stuff that leave the civilian units vulnerable. We need an algorithm that tells it to have at least one more military unit than civilian unit early game and to point defend the civilian units.

The thing that the siege units gets cover promotions. I read somewhere that siege units doesn't have that ability to start with, and can get promoted to it naturally, and will thus not knowing how to use it, and it will have no effect gameplay. Is that so?