K-Mod: Far Beyond the Sword

Karadoc - you're right that the drill promotion is very good in Kmod, and thanks for that change, but I still wouldn't say the PRO trait matches the AGG trait for the human player.
The main problem is the units affected. Classic age war? Horse archers, if you can't use them, use axes/spears/swords/catapults. No one is going to do anything with archers.
Medieval war? Macemen and pikes are great on the offense, crossbows and longbows are less useful.
Protective becomes equal to aggressive by the time you reach gunpowder, but for the first half of the game I'd say aggressive wins. Another thing - cheap barracks is actually a useful bonus especially in the early game. An aggressive leader can launch an axe or sword rush earlier. Cheap walls/castles is mostly useless. Those buildings are great for the AI and not used so much by the human. Walls are rarely built, castles are almost never built.
The only time I think the PRO trait makes any sense is with the Chinese, since their UU is a crossbow. In that case I'm glad they are protective as I would find it annoying if they were aggressive since it would have no synergy.

Karadoc, I have a question. You once mentioned that you have made AI averse to building Manhattan Project. How exactly was it done? Does AI dislike nukes or only the project itself? I want to mod a game in such away that nukes don't need the project and are available with a National Wonder instead. Will AI want to build this National Wonder and then nukes?

I have not made the AI less likely to build the Manhattan Project, but I have significantly changed how the decision is made.

The original bts AI's decision to build the Manhattan Project was completely random (and quite rare). A peaceful civ with no access to uranium and no intention to build nukes was just as likely to build the Manhattan Project as a warmonger civ with all the prerequisites and motivations.

I've changed it so that the AI will only build the Manhatten Project if they think they will benefit more from it than their opponents, and even then they will usually only consider it when they are in the late stages of a conquest victory.

So although the AI isn't less likely to build the Manhattan Project than before, it is deliberately only built rarely.

--

Once nukes are enabled, the some AI's will be more likely to build nukes than others - based on personality and situation. But the more nukes that are built and used in the game, the less reluctant other civs will be to use nukes themselves.


--

The AI I've described for the Manhattan Project applies to any 'project' which has the 'allows nukes' property. But that property enables nukes for the entire world - not just for the civ building the project.

As far as I know, there is no existing mechanism in the game for you to have a national wonder which enables nukes only for civs who have built the wonder. So if you wanted that functionality, you'd have to edit the DLL source to add new xml data types, and the current AI would not know anything about these new xml data types.

You could make a building that allows local construction of nukes - like monasteries allow missionaries. That would use a very basic AI analysis which doesn't really take anything much into account at all.

Keindorfer said:

Thx for the answer karadoc. we are all using windows 7 64 bit versions, and we are running out of ideas to fix this issue, so we continued playing on version 1.40. i think it is not a big deal if nobody else has this problems, it has to be a problem from our third pc. if we start a game without our 'problem child' we don't get OOS errors.

Click to expand...

I'd like to fix this, but I still have no idea what could be causing the problem. I've again looked through all the changes since v1.40, and I still can't see anything that could cause directly cause OOS problems. And I've played another multiplayer game on v1.41 using a different computer, and I still didn't see any OOS errors.

The only hints are that one of your three computer is behaving differently to the other two, and that it wasn't a problem in v1.40. Maybe it's something to do with differences in the CPU, or differences in OS settings, or differences in Civ4 settings, or even just differences in how the player on that computer plays the game. I really don't know. And whatever it is, I reckon it must be something that's only indirectly caused by a change in v1.41. ie. there probably is no new OOS bug in v1.41 but rather something is different to make an existing bug manifest as a common OOS error.

Either that, or the mod still isn't installed properly on that computer... but it sounds like you've checked that pretty thoroughly. Besides, usually when the mod files are different on one computer that computer can't see the multiplayer game on the LAN games list. (The game can still be joined using "Direct IP Connection", but the game doesn't appear on the games list in the "LAN Games" section when the mod files are different.)

So I still haven't been able to reproduce the problem myself, and I'm no closer to finding a solution.

-
One other hint for the problem is that you say the OOS happens between turns 15 and 40. I wonder what kinds of things are common in those turns.. maybe the first religion is founded, or the first random event, or some aggressive civ considers planning a war... Maybe it's not related to any of those things.

@Keindorfer

Did you check, if that computer had compatiblity mode on? I completely forgot this on my own computer and was cursed with random OOS because of that.

@Karadoc
I'm still wondering, if the peace conditions are a subject to change as I described a few weeks sgo. The answer of another person had valid points, but i still think, that loosing cities should mean something to the AI

karadoc said:

You could make a building that allows local construction of nukes - like monasteries allow missionaries. That would use a very basic AI analysis which doesn't really take anything much into account at all.

Click to expand...

That was basically my idea. I wanted to make a National Wonder (which would require a Great Scientist to build) which gives a free building in all cities which would allow nukes. What kind of AI analysis I may expect in this setup?

hey karadoc,

how are you?

i wonder, i started to notice something in the auto plays,

did you change the way conquest victory works? cause i can see the victory is achieved when there are still live and well civs on the map - it seems to work like domination.
in my mod - i added 2 victories , not at the end of the file, maybe indeed to move them to the bottom of the file? i dunno im guessing here , its weird, i wanna see the auti ai plays kill all other civs, but its not happening.

any ideas?

and ofc, im using your mod with some of my xml works.

TheOnlyDJCat said:

I'm still wondering, if the peace conditions are a subject to change as I described a few weeks sgo. The answer of another person had valid points, but i still think, that loosing cities should mean something to the AI

Click to expand...

Losing a city already does have a large affect on whether or not the AI thinks it is winning a war, but that affect is often less than the effect that Ondskan correctly described earlier. When the capture of a city is imminent, the AI tries to factor that into the peace value calculation such that peace is somewhat urgent if it is their city that will be captured; and that peace should be avoided if the AI is the one that's about to capture the city.

I think it's sensible for the AI to work in this way, but I certainly don't think it's flawless. I agree that the peace deals offered (and asked) by the AI are not good. Sometimes they are far too generous, and sometimes they are too stubborn. I've been half-intending to rewrite AI_endWarVal, and to make some new special rules for how the AI handles peace trades so that they are more reasonable, and flexible in what they'll accept if they don't actually want the war anyway. But I haven't gotten around to it because I know it's a massive can-of-worms and that I probably won't be able to come up with a good system without a massive amount of work on it.

The AI's 'war success' score that it keeps internally is very simplistic, and so it's not easy to turn that into a proper cost-benefit analysis - but that score is also so embedded in the code that it would be difficult to make changes to it without breaking something else. And evaluating cities in terms of other resources is a situational calculation. Sometimes losing a particular city would essentially mean losing the game - whereas other times the city just isn't important, even if it has the same kinds of yields and buildings. It depends on all sorts of things which the AI would need to be taught if it were to evaluate it properly.

...

So the short answer is that that stuff is subject to change, but that it might not be soon.

Gavagay said:

That was basically my idea. I wanted to make a National Wonder (which would require a Great Scientist to build) which gives a free building in all cities which would allow nukes. What kind of AI analysis I may expect in this setup?

Click to expand...

Essentially no analysis, and I don't think the current AI would value such a building very highly. You may need to use the iAIWeight value in the XML to encourage the AI to think the building is worthwhile. iAIWeight is roughly in units of 1/4 gold per turn. (ie. if you set iAIWeight == 4, the AI will think it's worth around 1 gold per turn more than it would for iAIWeight == 0.)

keldath said:

hey karadoc,

how are you?

i wonder, i started to notice something in the auto plays,

did you change the way conquest victory works? cause i can see the victory is achieved when there are still live and well civs on the map - it seems to work like domination.
in my mod - i added 2 victories , not at the end of the file, maybe indeed to move them to the bottom of the file? i dunno im guessing here , its weird, i wanna see the auti ai plays kill all other civs, but its not happening.

any ideas?

and ofc, im using your mod with some of my xml works.

Click to expand...

I'm awful.

The victory conditions have not been changed. In your case it's probably just that the AI is accepting its enemy as a vassal right at the end of the game. When a civ is a vassal, they are counted as being conquered.

awful??

no! you want something to drink ? eat? maybe a hug?

ok, i guess its vassals, ill check it out.

thanks dude.

Karadoc - congrats, man, you have programmed the AI so well that it is virtually impossible to beat on higher difficulty levels in the late game. Before playing Kmod I found that the A.I. was strong right up to the industrial era, and then dropped off. I could die to hordes of cavalry, but once the game got to tanks/bombers, etc, I was good. The A.I. just didn't seem to grasp industrialization, spamming power plants, using its air force, etc.

Now, it's the opposite. Starting at monarch and up the A.I. gets increasing bonuses with each era. In the past I didn't really notice this as the A.I. played the late game so poorly, but now I surely notice it. Once it gets to the industrial age I find the A.I. becomes a beast. It's not uncommon to face opponents with double my MFG and GNP, even though they aren't much bigger than I am.

For all you Kmod fans, is anyone beating this on emperor? I have never been able to consistently win on emperor and even lose quite a few games on monarch but with each new version it gets harder and harder, I win perhaps only 15% of my emperor games now, then again I play with continents which makes it harder. Once the A.I. gets to the industrial age it seems unstoppable on emperor. It seems no matter how well I fight a war I get completely slaughtered by endless hordes of units. I'll take a city defended by 15 cavalry, my stack being 40 infantry and 10 artillery, and then I lose the city to 50 cavalry. So I hit it again with 20 artillery and 50 infantry and take it, and then lose it to a stack of 80 cav and 80 infantry. The numbers are just astounding. I'll put in factories and coal plants and go to war while my opponent doesn't have a single factory or power plant and yet they manage to outproduce me somehow..... it's crazy!!!

That depends really on the map type (and number of opponents)
I win on immortal if I do not share my land mass with too much AIs.
It is stil relatively easy to roll over 1-3 AIs, defends against naval attacks, and kill the other continents.
BUT...
I still wait for my first win for months on huge landmasses with 17 opponents. I could do it on emperor with some leaders regularly, but on immortal I'm killed by neverending SODs.

If anyone can enlight me, how to win with KMod on huge landmasses on immortal, I would be really thankful

Problems:
- AIs have too much space too settle for peaceful play. Winning liberalism race with 7 cities does not matter much, if 4 AIs roll over at 1000BC with stacks>60 units...
- It's hard to play the diplo game with 18 players to keep enough of them calm
- Early rushes with Inca or Romans are not as strong with huge landmasses

This was not a problem before KMod, because having rifling or cannons 400 years before the AIs will win no matter what size the AI stacks have

Yeah, Noto, the main way I lose is playing a continents map or not too big not too small map and theres a civ that vassals someone mega early and then they become massively ahead in tech. I usually can win Pangea maps.

Why I lose,

1. One super AI is too far away for me to do anything until I get tech enough to sail there. And that ai has ended up vassaling one or two people really early, which directly leads to massive tech lead through tech trading Im guessing.
2. I usually can never vassal people early and its a very painful process.
3. I actually get alot of times stuck on a continent on my own (which is still hard to come back from, on its own), or my one neaby neighbor (nearby but not on the same continent) will NOT tech trade with me, which I find werid.
4. If I do finally manage to vassal someone, I usually get naval invaded right after by someone else.

I find in Kmod the AI is very good at emergency military production. If a war drags on I'd better really have a significant productiong/military advantage or it will grind to a stalemate. On the lower difficulty levels I suppose this isn't a problem because one can fairly easily have a MFG advantage, but even on emperor in Kmod it's quite difficult to stay ahead of the AI in MFG. I find I start a war and win battles with something like a 1/5 casualty ratio and think that I'm doing very well, and take a couple of cities, only to then face truly enormous stacks of AI units and I end up getting bogged down in a massive war of attrition. This only happens late game. Early game I can still win wars as well as I used to, but something just seems to explode with Kmod emperor AI in the industrial age. I know that the AI bonuses get higher with each age, perhaps by the industrial age the AI bonuses are just crazy. The unfortunate thing about that is that it just encourages people to win the game in the rifle era... which is so overdone. I actually enjoy the industrial era and wish that it was more practical to take the game that far, but most of the time it seems you need to seal the deal in the renaissance.

That is exaclty why I play huge highland maps with 18 AIs. The game can't be sealed with rifling alone, because the landmass is too big. You have to win in industrial age to conquer the big AIs on the other side of the map.

noto2 said:

Karadoc - congrats, man, you have programmed the AI so well that it is virtually impossible to beat on higher difficulty levels in the late game. Before playing Kmod I found that the A.I. was strong right up to the industrial era, and then dropped off

Click to expand...

^^ This

Far beyond the sword indeed.

Holy Cripes!! I haven't been able to win a game in the latest version, and even in the last version I was on a crazy losing streak...something like losing 14 out of 16 games. I'm really noticing a difference compared to Kmod, say, a year ago. Karadoc whatever you've done in the last year has really buffed the AI. I mean, I played monarch for years and finally made it up to emperor before playing Kmod and of course it was not comparable at all. The AI always got economic cheats but I suppose it just didn't understand how to exploit them. Now, in Kmod, a 20% production/science cheat is very noticeable and seems impossible to beat!

Danwood said:

FINALLY a mod which cares a bout the global warming.

FINALLY an author who understood that GW can be an interesting features is well developed.

Karadoc, you made me very happy

Click to expand...

A good GW mod would include increased crop yields due to additional CO2 in the atmosphere. Available CO2 is the greatest constraint on plant growth in our environment. Science experiments have shown yield boosts of 50-100% at 400 ppm.

Also, there is a great deal of land in the northern hemisphere that is not currently suited to agriculture because it is too cold. VAST stretches of northern Canada and Siberia would greatly benefit from 5-10 degrees of warming in those regions. The American plains could produce 2 crops per year (like China does, to feed 1.2 billion people using only 1/3 the arable land that the US uses) instead of just one.

Current CO2 level in the atmosphere is around 290 ppm (parts per million). 65 million years ago it is estimated in excess of 2500 ppm. And, the world was at its greatest fertility during that time, supporting far more vegetation and greater wildlife than it does today.

Oh, and it didn't destroy the earth.

Common sense should inform us that all the oil and coal in the ground that we are burning was once CO2 in the atmosphere (and biosphere), before it got locked up by algae falling to the bottom of great inland seas and getting covered over, most of which happened in the last 300 million years. before that, all that CO2 was in our atmosphere, and the earth functioned just fine.

One last point: the earth's greatest carbon sink isn't coal, oil, or our current biosphere, it is limestone, aka calcium carbonate (CaCO3). I'd estimate that more than 90% of the earth's available carbon is sequestered in the earth's vast limestone deposits, many miles thick in places. This is carbon that is no longer available to the earth's biosphere, limiting how much life the earth can sustain.

Danwood said:

@karadoc


Turn melted polar caps into water tiles, like the "rising seas" feature from GW mod (which is optional): for every (insert number here) polar cap tile melted, there is a coast (non hill-mountain) tile which turns to water (this would not include cities if possible). Indeed the new ocean tile would lose any improvement-resources on it.

This would make GW more severe, but also more interesting, and most important would be totally optional.

Click to expand...

It is also scientifically inaccurate. When floating ice melts, it has zero effect on sea level. Since the ice is floating, it is already displacing an amount of water exactly equal to its mass. Ssince ice is less dense than water, when it melts all the water locked up in the ice will exactly fill the "hole" in the water that the floating ice was making.

Charles555nc said:

I just saw a russian worker (belonging to russia) building a mine in an American city plot, I thought you could make roads for someone, but you couldnt change their plots (washington currently has a windmill and the russian worker is making a mine). They arent even vassal or master to each other.

As far as slowing down, and this might be my computer but the entire time of all the ais turns are probably only between 3-6 seconds total (longer if you have show ai moves on ofc) at least up until the last 75 turns of the game. But I realize other people may have older machines...

Click to expand...

This can happen if the plot was culture flipped, and the worker hasn't had a chance to move yet. I play a lot of culture-conquest games (always peaceful/domination victory only), and I see this a lot.

@Marguerite Ming, have you actually played this mod? The land in the icy areas does become more productive when global warming strikes - and the mod is biased in a way such that the cold regions are more likely to be hit than the hot regions. Your criticism seems to be about some vague notion or realism rather than on accuracy - and the reason I say "vague" is that while you keep invoking 'science' and 'accuracy', you yourself are making mistakes.

One key thing about climate change that everyone needs to understand is that the world is no in any danger. The planet earth itself doesn't care in any way about what is going on on its surface. It doesn't care about the temperature or the weather or about what the air is made out of. But the current life on earth cares because we rely on the current climate for our lives and livelihood. It's possible that some life might be better of with a different climate, but that's not much consolation for species that will be wiped out.

The earth's climate has been different in the past, and life as evolved to adapt to the changing climate - but that might not work so well this time. Maybe in the long term, us humans can move our major cities closer towards the north and south poles, but that's not an easy thing to do. And besides, there are more immediate problems arising from climate change facing us already, such as the increase in the frequency of extreme weather events.

In any case, this stuff is obviously beyond the scope of this mod. It's clear that you've come here to argue about climate change rather than discuss the mod. If you want to discuss that stuff seriously, take it to this thread.