A Better AI.

I reiterate something I've said a number of times before:
When the AI is failing to compete or adapt, PLEASE send me, or post a, save or screenshot (saves are best) exhibiting the problematic behavior, even if the AI isn't doing anything particular wrong but is just... failing in general. There might be something about the settings or the landscape which the AI needs to be tweaked for. I can generally fix these but I can't find all these problematic cases on my own.

edit: I've created a gmail account:

betterai@gmail.com

If you wish to send me a save by email then send it there and include "Save" in the title and a brief description of the problem would certainly help.
 
This saved game seems to crash every time Christianity is founded. FWIW, I'm running on an AMD X2 3800+, nVidia 6800GS, and with all the latest drivers.
 
Hey.. I'm still waiting for an answer to my last post about starting points... I posted 2-3 screens of very weird starting points with the latest build.. is this normal?
 
I'm forced to assume you are using freakish settings that or high sea level archi, or both. I might have to look into high sea level archi since that map kind can present unique challenges to the start point selector/normalizer.

edit: Also the settlers are not placed on the food resource, the food resource is placed under the settler. Strange, but that's what it does in 2.08 and I have not changed that.
 
I'm forced to assume you are using freakish settings that or high sea level archi, or both. I might have to look into high sea level archi since that map kind can present unique challenges to the start point selector/normalizer.

No.. I'm using custom continents, 3 continents, sea level low.
Also.. did you.. by any chance.. touch the tribal villages spawn frequency? I'm finding many more than what I used to do.

edit: Also the settlers are not placed on the food resource, the food resource is placed under the settler. Strange, but that's what it does in 2.08 and I have not changed that.

But it never happened to me to see a food resource under a settler in the starting point... nor in 2.08 or in old build of the mod.
 
I dont know what is happening in this game(Highlands(so pangea), 8 CIVS, standard size, the rest is default I believe), but the AI is declaring war on me ALL the time! ANd only on me!
Alex declared war 3 times on me, Julious and Hanibal 2 times each and Cyrus 1 time! They have weaker CIVs to pick and with less diplomacy points than me, but htey keep declaring war on me!! ALl those dagger stacks attacking me at the same time all the time! Oh my, that was hard to handle! For 2 occasions I had 3 civs at war with me at the same time(and a lot ofr times I had 2 CIVs at war with me) and all with dagger stacks!
I dont know what is it, but I guess my starting location in the map doesent help at all! Im in the middle of everybody :S

And even then, I could handle my own and only lost 1 city(twice), but in the end got back as a gift in a peace agreement! That city was barbarian and probably changed hands lik 5 times heh..And also I got 2 cities already! 1 from Alex and other from Julious. I found that both empires were weakly de fended, so I sent some sneak attacks in border cities when the situation was a little better..

Now FINALLY after something like 150 turns in constant war I got completely peace! At least fo a while..:rolleyes: First time I usaed Nationhood ever, what a civic >.< I survived basicky whiping and drafting!

Here are some screenshots: And Blake, later I send you a email with some saves if you want?


Here Julious Caesar almost giving my death sentence(note that I was already in war with 2 more CIVs)



Here a look at 1 of his 3 stacks:




there are some more images, but no time now, later I edit!
 

Attachments

  • Civ4ScreenShot0024.JPG
    Civ4ScreenShot0024.JPG
    148.4 KB · Views: 80
  • Civ4ScreenShot0026.JPG
    Civ4ScreenShot0026.JPG
    191.7 KB · Views: 74
Now the default governor is a really smart guy - he is really good! This is largely in thanks to Iustus and in the latest version I removed some things which were holding the governor back from his full potential.

How do I keep Mr. Governor from allocating specialists that pollute my GP building? I've noticed that the governor seems to be getting creative that way, but not creative in a way that I always appreciate. It would be a good start if it didn't allocate specialists that don't correspond to either a) specialists that are set (forced) by the user, and/or GP points resulting from wonders or special buildings. Also, the obnoxious behavior of allocating an engineer whenever a forge shows up is just ... obnoxious.

I'm not at all sure under what circumstances the above happen, which is probably as disturbing as the happenings themselves.

The governor does seem pretty good. I do wish I could request some tweaks to it as game-wide options though.

It's not the governor's fault, but I wish there were a way to dump production when a city shouldn't be producing anything, for example, in an early pacifist government when money is tight, and there's nothing to produce (before you can build beakers/culture/gold) except military units that will have to be deleted to save cash. I wish I could build Nothing.

How does the governor hold cities at "growing in 1 turn" in "don't grow" mode when it seems that the city must grow next turn?

Also ... in re the start code, I got a bunch of less than exciting starts in a row on a Continents map a couple days ago, including one on an isthmus that probably had only 6-8 land tiles in the fat cross and no particular seafood, and jungle in both directions. I like having coastal starts (it's great if the HE and so forth can be built on the coast) but they've been kind of blah, it seems, and not in the "I don't have 5 crabs anymore oh bummer" kind of way. Maybe it's just a bad run with the random number generator.
 
whats the latest with this mod? i use the the visa mod so can this be used alongside it?

the readme says put the dll file in the warlords custom assets folder which i have done but find it a bit strange that there aint already a dll file in there. is that the correct place to put it so that it works with all mods?
 
How do I keep Mr. Governor from allocating specialists that pollute my GP building? I've noticed that the governor seems to be getting creative that way, but not creative in a way that I always appreciate. It would be a good start if it didn't allocate specialists that don't correspond to either a) specialists that are set (forced) by the user, and/or GP points resulting from wonders or special buildings. Also, the obnoxious behavior of allocating an engineer whenever a forge shows up is just ... obnoxious.

I'm not at all sure under what circumstances the above happen, which is probably as disturbing as the happenings themselves.

The governor does seem pretty good. I do wish I could request some tweaks to it as game-wide options though.

It's not the governor's fault, but I wish there were a way to dump production when a city shouldn't be producing anything, for example, in an early pacifist government when money is tight, and there's nothing to produce (before you can build beakers/culture/gold) except military units that will have to be deleted to save cash. I wish I could build Nothing.

How does the governor hold cities at "growing in 1 turn" in "don't grow" mode when it seems that the city must grow next turn?

Also ... in re the start code, I got a bunch of less than exciting starts in a row on a Continents map a couple days ago, including one on an isthmus that probably had only 6-8 land tiles in the fat cross and no particular seafood, and jungle in both directions. I like having coastal starts (it's great if the HE and so forth can be built on the coast) but they've been kind of blah, it seems, and not in the "I don't have 5 crabs anymore oh bummer" kind of way. Maybe it's just a bad run with the random number generator.

Well that partially depends on whether the specialists are to be used for their 'normal' production or GP points. In Most Cities, Specialists are used for their normal production (hammers, gold, flasks).

Perhaps when the 'Emphasize GPP' button is on, the Governor could attempt to maintain GPP Purity (ie deemphasizing all specialists that don't match the 'Forced' Specialists)

As for building Nothing, just delete the units.

The Governor stops cities from growing because the 'halt growth' setting actually stops growth rather than Just changing tile allocation.
 
The latest build helps mid/late game AI productivity a bit. Perhaps it was stuck with tons of unhappiness from not having Emancipation?

Beyond that, I think the newest Governors are using marginal tiles too frequently rather than assigning specialists. I see a lot of 2f/1com Sea tiles being used. Is that really beneficial for larger cities? Wouldnt the city be far better served with a specialist than a single extra commerce? I can see when the city is small and wants to keep growing that it might be worthwhile, but for a size 12+ city, its growing slowly enough that I cant see the payoff compared to a specialist. Am I just missing the boat on how effective these otherwise 'marginal' tiles are?

The AI also seems quite farm happy. The 2f/1h farms are everywhere and being worked quite a bit. Again, I would ask if these tiles really pay off being worked? To me, higher pop = more expense, harder to keep happy and healthy, and not generating GPPs. How is that 1 hammer worth that (even late in the game its only 2 hammers with all the trimmings).

All that said, I am getting my head handed to me in my latest game (same difficulty as before where I was mopping up, but with the latest build). I do attribute a lot of that to my poor start (stuck on a penninsula with more limited expansion area). But even with this 'poor' start, I'm still leading in production (barely) in the late Middle Ages (I'm at 85, next AI is 80). I am a fair number of techs behind though.

Perhaps the AI is just putting more emphasis on teching rather than production now?

Thanks!
 
The latest build helps mid/late game AI productivity a bit. Perhaps it was stuck with tons of unhappiness from not having Emancipation?

Beyond that, I think the newest Governors are using marginal tiles too frequently rather than assigning specialists. I see a lot of 2f/1com Sea tiles being used. Is that really beneficial for larger cities? Wouldnt the city be far better served with a specialist than a single extra commerce? I can see when the city is small and wants to keep growing that it might be worthwhile, but for a size 12+ city, its growing slowly enough that I cant see the payoff compared to a specialist. Am I just missing the boat on how effective these otherwise 'marginal' tiles are?

The AI also seems quite farm happy. The 2f/1h farms are everywhere and being worked quite a bit. Again, I would ask if these tiles really pay off being worked? To me, higher pop = more expense, harder to keep happy and healthy, and not generating GPPs. How is that 1 hammer worth that (even late in the game its only 2 hammers with all the trimmings).

All that said, I am getting my head handed to me in my latest game (same difficulty as before where I was mopping up, but with the latest build). I do attribute a lot of that to my poor start (stuck on a penninsula with more limited expansion area). But even with this 'poor' start, I'm still leading in production (barely) in the late Middle Ages (I'm at 85, next AI is 80). I am a fair number of techs behind though.

Perhaps the AI is just putting more emphasis on teching rather than production now?

Thanks!



Additional population doesn't only increase the costs. Yes 1 additional population will increase civic costs by something like 0.4 gold (heavily dependent on which civics are being used) and it will increase city upkeep costs with a few tenths of a gold (heavily dependent on whether a courthouse is present and how far away the city is located). But it will also increase the free unit upkeep (it pays for 1/4 th of a unit and even more if you're using vassalage and pacificm) and it increases the output of each of the trade routes with something like a (few) tenth(s) of a commerce point (heavily dependent on the presence of foreign trade routes and harbor). The end result will usually be that an additional point of population costs a little, but not always. Sometimes it brings in money.

All of these changes are of course independent of what you do with the population point (whether you use it to enable a specialist or to use a tile).

If you choose to use a marginal tile, then the city could grow on to use even more marginal tiles. A lot of marginal tiles together are not so marginal anymore. For me the choice depends on whether I want some production/commerce now or if I want to have a more productive city later. The production/commerce now could also be used to help your empire grow further so it's not an easy decision.
 
As for building Nothing, just delete the units.
.

No, go for "research", "culture" options etc. Better than totally wasting your hammers. Build missionaries if you have multiple religons is another idea.
 
No, go for "research", "culture" options etc. Better than totally wasting your hammers. Build missionaries if you have multiple religons is another idea.
Early in the game you sometimes have none of those options. It's just an annoyance when you don't want more units (because they'll waste money) to have to build them and delete them. It happens when a city has grown to its happy cap, has built all its buildings, no units are needed, there's nothing around to improve, et cetera.
 
Early in the game you sometimes have none of those options. It's just an annoyance when you don't want more units (because they'll waste money) to have to build them and delete them. It happens when a city has grown to its happy cap, has built all its buildings, no units are needed, there's nothing around to improve, et cetera.

There are still tons of options:

- build Great wonders: If you are close to finish and you need it, finish it; if you don't need it, take it out of your build queue (you may even build the same wonder in another city once you remove it from your build order). You'll get the cash when an AI eventually finishes it.

- build the most costly unit, then before you finish, put it lower on your build order and build something else. You don't need to pay their maintainence till you finish the units. It may require a bit micromanaging, but it comes handy when you need a rush.
 
It happens when a city has grown to its happy cap, has built all its buildings, there's nothing around to improve, et cetera.
Very very rare occurrence. If, however, you find it happening to you a lot, why not prioritize techs that allow additional infrastructure buildings?

Alternately, aren't missionaries pretty much always a good option?

no units are needed...
I don't know that's ever happened to me. Since AI relations are based off the power graph, it's pretty much always a good option to build more units. If they start to cost upkeep, why not gift them to an AI that you suspect will be swallowed up by one of your rivals? i.e., the enemy of your enemy is your friend?

Wodan
 
There are still tons of options:

Also, one option is really to put city to build nothing. I'm not sure if this is what game is supposed to do put if you have supress popups option turned on its easy just to ignore city that has completed its most recent build and leave it to build nothing. Just do not open the popup asking what to build next and press end turn. I'm not sure if the hammers that build nothing will queu up so this might be an exploit.
 
A quick question - I typically use Smartmap for my games. I'm trying to figure out if your tweaks to starting points will be in effect or not when I'm using that custom map script.

I know Smartmap.py overrides assignStartingPlots() and as far as I can tell never overrides or calls normalizeRemoveBadTerrain().

It's my understanding that the tweaks to starting points are basically all to normallizeRemoveBadTerrain(), right? It's also my understanding (from the CvMapScriptInterface entrypoints file) that that function should still be called even without the smartmap script doing so explicitly. If both of those are true, then your changes should still be in effect even when I'm using the smartmap script to do map generation, right?
 
Settler blue circle question here. Using 06-12-09 build. Small map, normal speed, custom game always peace, no barbarians, no tech trading, settler difficulty.



I don't really know why it's suggesting one plot north of the current position. I would have thought it obvious to settle where the settler is, to get at the gem. The only thing I'm suspecting would be distance maintenance. The city radius seen at the top of the image belongs to the capital. I don't know how maintenance is calculated, though. :<

Maybe it's the jungles giving a penalty? Perhaps the jungle penalty should be set even lower, or perhaps completely scrapped once Iron Working is known?

Also, the AIs aren't aware of the implications of Always Peace, I think. They're still going for archery quite early and having multiple defenders in their cities. Hm...but Always Peace is too much of a special case, I wonder if anyone actually plays this?
 
A quick question - I typically use Smartmap for my games. I'm trying to figure out if your tweaks to starting points will be in effect or not when I'm using that custom map script.

I know Smartmap.py overrides assignStartingPlots() and as far as I can tell never overrides or calls normalizeRemoveBadTerrain().

It's my understanding that the tweaks to starting points are basically all to normallizeRemoveBadTerrain(), right? It's also my understanding (from the CvMapScriptInterface entrypoints file) that that function should still be called even without the smartmap script doing so explicitly. If both of those are true, then your changes should still be in effect even when I'm using the smartmap script to do map generation, right?

I'm not intimately familiar with Smartmap, but generally, if a map script does perform normalization, it uses the SDK normalizing functions, although there's nothing stopping a particular map script from using custom normalizers or disabling particular normalizers.

Settler blue circle question here. Using 06-12-09 build. Small map, normal speed, custom game always peace, no barbarians, no tech trading, settler difficulty.



I don't really know why it's suggesting one plot north of the current position. I would have thought it obvious to settle where the settler is, to get at the gem. The only thing I'm suspecting would be distance maintenance. The city radius seen at the top of the image belongs to the capital. I don't know how maintenance is calculated, though. :<
Take blue circles with a grain of salt and remember they aren't drawn in fog. I think in this case the Ivory/Banana/2xSugar site is scoring very highly - that's probably the site it wants you to found at.

Maybe it's the jungles giving a penalty? Perhaps the jungle penalty should be set even lower, or perhaps completely scrapped once Iron Working is known?
As of this latest version it is giving a penalty - for having 4 different unique bonuses within the radius. That's considered "Too Greedy", it's not a big penalty but it's there to generally discourage founding a city smack bang in the middle of 4 or 5 resources, often with none being immediately workable. The advice it's giving is certainly not bad and would fit a dotmap just fine.

Also, the AIs aren't aware of the implications of Always Peace, I think. They're still going for archery quite early and having multiple defenders in their cities. Hm...but Always Peace is too much of a special case, I wonder if anyone actually plays this?
They are barely aware of it... it's very low priority to improve.

No.. I'm using custom continents, 3 continents, sea level low.
Also.. did you.. by any chance.. touch the tribal villages spawn frequency? I'm finding many more than what I used to do.
No change to the tribal villages.


But it never happened to me to see a food resource under a settler in the starting point... nor in 2.08 or in old build of the mod.

There was a set of circumstances which made it possible for two starts to be very close together... I've fixed this from two different approaches - directly fixing the issue that led it towards close starts and adding an additional factor to drive apart close starts. I can't outright forbid close starts, but they should be much less likely...
 
Back
Top Bottom