A Better AI.

[sveint]

Hmm, can you give some more detail? I do not see artists assigned much under normal circumstances. Were there open spots for engineers, scientists, merchants, or priests still? I assume you were at or above your happy cap? Did you have emphasize commerce clicked? (Emphasize all 3 counts here). What was your global culture slider at?

I was playing Mansu Musa, fractal world and ended up on my own little island. There was no foreign culture anywhere nearby. I ran Caste System coupled with Pacifism. In just about every large city I would get a few artists when running the governor on default (no emphasis of any kind - I do this to play faster/test). At no time did I exceed happy cap.

My point is that artist great persons are much less useful than the others. Personally I think a smart AI should always use any other kind of specialist and ONLY use artists when the fat cross culture is threathened (I don't even agree on using them to 'fix' excess population beyond happy cap).

Not only will this help the AI but it will take away some micormanagement for fast play if we can trust the governor not to assign artists.

I think I still have the savegame, so I can post a few screenshots later today.

Final note: emphasising either food or production would remove all the artists.

 

[Jaybe]

Regarding ships I think a reason the AI leaves so many combat ships in port is because they're effectively acting as escorts for the troop carriers. If the troop carriers are doing nothing, i.e. not transporting troops, they tend to remain docked in cities in an idle state. I remember from the days of CivIII when frigates and destroyers etc were created by the AI they were either labelled as attack ship or escort ship. If the latter then that's all they ever did and their role didn't change. Maybe there's something similar going on in CivIV? Regardless, it's daft when my ships are off pillaging the AI's sea resources when they have a multitude of ships doing diddly squat in their ports.

It may be daft in the game, but history is FULL of examples of daftness! There should always be at least a small probability of daftness by the AI.

Remember, unless we are playing an MP game, we often don't catch the occasions where we do daft actions ourselves!

But yes, we should not be able to DEPEND on AI daftness.

 

[Iustus]

I was playing Mansu Musa, fractal world and ended up on my own little island. There was no foreign culture anywhere nearby. I ran Caste System coupled with Pacifism. In just about every large city I would get a few artists when running the governor on default (no emphasis of any kind - I do this to play faster/test). At no time did I exceed happy cap.

My point is that artist great persons are much less useful than the others. Personally I think a smart AI should always use any other kind of specialist and ONLY use artists when the fat cross culture is threathened (I don't even agree on using them to 'fix' excess population beyond happy cap).

Not only will this help the AI but it will take away some micormanagement for fast play if we can trust the governor not to assign artists.

I think I still have the savegame, so I can post a few screenshots later today.

Final note: emphasising either food or production would remove all the artists.

I would love to see a save game. Please post it if you still have it.

I have not really tested with caste system, I almost never use it myself, but I will do so. I still do not see why it would choose artists over scientists or merchants though. Something strange must be going on.

-Iustus

 

[Iustus]

New build 12/02 now available for download.

release notes:

- financial wars! warlike leaders will now consider starting wars to solve their financial problems. When a warlike leader is in war, if his finances are bad or he has chosen dagger strat, he will be more reluctant to accept peace.
- major bug that caused city attack stacks to not merge, which prevented war declarations prior to seige weapons
- if there is danger, wonder builds will be canceled when queue is made dirty
- in the high production cities, leaderhead wonder preferences for wonders is doubled, this should increase wonder builds overall
- the city governor will now grow cities faster all the way to the target level, and to 85% full in that level, before slowing down
- the city governor emphasize food case emphasizes food even more
- the city governor will always let a city go one over the happy cap when in slavery
- once it reaches its target level and 85% full, the city governor may work lower food tiles in order to stop growth
- AI cities will only let them starve to 80% rather than 50%
- found value used to found cities updated: bonus resources in the outer ring will be correctly valued now, blocked bonuses will now be counted against city found locations, high opponent culture in the founding site will all be taken into account now
- granaries get even higher building priority
- in high water maps, dagger strat will build ships
- "plot danger" changes, when workers decide to stop working changed
- for cultural victory, AIs will not settle artists in cities that have not popped borders at least once
- for cultural victory, AIs will try even harder to found a religion if they have zero holy cities
- units that are given the mission of defending cities will not get city raider promotions
- a lot more info under chipotle cheat code, particularly war info

edit: bug fixed, make sure you download the 12/02 build, not the 12/01 build.

enjoy,

-Iustus

 

[kettyo]

There was a bug recently that caused AIs to not attack early, prior to getting siege weapons. That will be fixed in the forthcoming patch. In addition warlike leaders may choose war to solve their financial troubles. If anything, war may become too frequent and we will have to turn it down a bit

Are you sure it's a good idea to go conquest in financial trouble?
A civ is in financial trouble probably because it has a lot of underdeveloped, underimproved cities.
Freshly captured cities are of that same kind.
So i think it will pull the civ in even deeper financial troubles.
The only way war could help with gold problems is when you attack a city, pillage it then let the enemy retake it, then pillage again etc. but i think it's too complicated to program 'a war for booty'.
In general i think civs in treasury problems should develop cities and cottages/towns.
I'd stop expanding if tax rate must be above 50% to reach gold balance and continue when the financial trouble is outdeveloped.

EDIT: Now i see there is the new version out
We shall test it anyway

 

[Iustus]

If you want to do detailed testing of the new war code, I would encourage you to turn on chipotle mode so you can better understand what is happening.

Here are some screenshots of the extra information (alt key down):

This first one shows the power of every civ in the game. The second number is the summed power of a civ, all its vassals and all those with which it has defensive pacts or permenant alliances. It also shows those civs which are in financial trouble, with the ($$$). Do not pay much attention to the human player having this symbol, it does not mean anything in this case (nor does it really mean anything in the barbarian case).

When mousing over a civ in the scoreboard, it shows the value a team places in going to war with another civ. Only those civs with whom the leader will entertain war are listed. Here, Montezuma will consider war with Alexander (barely, 70), but not with Washinton or Iustus (because their power is too high currently). It it will say 'near war' or 'distant war' depending on the situation, the power ratios are different for each.

Ok. Now here we see the case where a civ has decided to go to war, but has not actually declared war yet. The war declaration will happen when a city attack stack reaches the border. The possible types of war are: preparing total war, total war, preparing limited war, limited war, dogpile war, (there are a few others, but they should never occur here). The number shown is the value the civ placed on starting this war. Note, the value is much higher here (800).

Once war has been officially declared, the values for declaring peace are shown. Higher numbers mean a higher value is placed on peace. Here we can see that Alexander values peace much higher than Washington does:

Unfortunately, since Washinton is a vassal, he cannot negotiate peace himself. The peace values between Iustus and Alexander are much closer, so the price for peace will be about even if negotiated at the current time.

-Iustus

 

[Pellanor]

One quick request:
Include the version number when you hold Alt while mousing over your score. This should be fairly simple to add

So far I haven't had the chance to sit down and do a good trial run of the better AI (Cursed FFXII), but I'm looking forward to it.

 

[sveint]

Nevermind this, sadly I was not testing the latest build.

 

[Iustus]

Ok, thanks to my screenshots, which seemed a bit wrong to me, I went back and found 2 bugs in the end war value code!

A new build, 12/02 is posted which should correctly value ending wars.

-Iustus

 

[Iustus]

Nevermind this, sadly I was not testing the latest build.

Ok, glad to hear that the latest build fixes the problem then. That is what you mean?

-Iustus

 

[Pfeffersack]

in high water maps, dagger strat will build ships

Does this mean that the AI is now also able and willing to load their frigthening stacks of land units on armadas of ships? The poor execution of naval invasions annoyed me since SMAC and CivIII...would be a dream if this mod solves this AI flaw as well.

Simply great work...can't wait to play with the new build

 

[V. Soma]

I tried "chipotle" but I see no info when mousing over civ names with ALT pushed down

BTW: I see no "===better AI===" either...

 

[Bursk]

I just checked back here to report on my game using the 1st Dec build, and I find that fresh 2nd Dec build!

I was going to report that there's been some fighting in my game - continents, Prince, large map, 9 civs - but none of it has taken place on my continent, I think due to us all sharing the same religion and being of a similar power.

I have seen two, maybe three Great Generals spawned throughout the game, and I just got a message at 1220AD that the Malinese empire is no more! Perhaps the bugfix came too late for poor Mansa Musa. Heh.

Anyway, I'll now switch over to using the 2nd Dec build. It's good to see that there's some AI war taking place, that's for sure.

 

[Iustus]

I tried "chipotle" but I see no info when mousing over civ names with ALT pushed down

BTW: I see no "===better AI===" either...

It sounds like you do not have it installed correctly, look at my post here in this thread for more info on installation.

If you are not even getting better AI when alt-mousing over your name in the score list, then you do not have BetterAI installed properly.

When I say the score list, or scoreboard, I mean this thing:

To add the chipotle thing, you need to find your CivilizationIV.ini file (Windows lists it as "Configuration Settings"), inside ~MyDocuments/MyGames/Warlords/, open it with notepad (double click), and change the cheat code:

I hope this helps,

-Iustus

 

[Iustus]

Does this mean that the AI is now also able and willing to load their frigthening stacks of land units on armadas of ships? The poor execution of naval invasions annoyed me since SMAC and CivIII...would be a dream if this mod solves this AI flaw as well.

Simply great work...can't wait to play with the new build

Only slightly. Naval warfare still needs some work to be up to par.

-Iustus

 

[HardCoder]

I think we're both right: due to Blake further enhancing the AI's tech lust, I believe the tech tree needs a little work with the possibility of increasing the research requirements for the latter half techs.

Something about being able to bypass entire phases of military units doesn't seem quite right to me. I'd prefer it, I think, if for example:

1) You have to acquire Guilds before you can research Military Tradition, and you have to acquire Rifling before you can research Assembly Line, and so forth (these are the most obvious gaps to me)

2) You have to have built a Knight x number of turns (where x is ... 10 or 20?) before you can build a Cavalry (or must have had Military Tradition for that number of turns in), and you have to have built a Rifleman x number of turns before you can build an Infantry

In other words, in general you have to make a perfunctory stop at each successive evolution in military units. I'm not saying you need to build an axe, a spear, and a sword to build a mace, but perhaps before you can build a unit you need to build one of the units that can upgrade to it ... or wait a number of turns after acquiring the technology. So you would have to build a spear before a pike, or wait a while if you didn't.

This wouldn't inconvenience the human player as much, I think, as the AI, and it has an aura of fairness to it because the AI gets such huge discounts on its upgrades. Requiring a little more thorough path through the tech tree, for military units at least, would put a stop to finding Napoleon with tanks 10 turns after he didn't have cavalry.

 

[Meatbuster]

Something about being able to bypass entire phases of military units doesn't seem quite right to me. I'd prefer it, I think, if for example: (snip)

I don't think modding the tech tree is part of this project. This only involves AI. You can always edit the tech tree yourself.

 

[HardCoder]

The governor in the new release (Dec 02) is choosing tiles that build a settler more slowly for me, apparently because of commerce.

I'd like to post a screenshot but that would be difficult because I am playing on a different machine (that's connected to the LCD TV). Anyway, here is the setup:

I'm playing Prince as Elizabeth. I have a city on the coast with 3 coast tiles, three forested grassland, one mined gold desert hills, and one oasis in addition to my city. The fastest build is city + oasis + forest (30 turns at epic), but it wants to use the gold (38 turns). The difference in commerce is huge - 5 vs 13, which at 985BC is a big deal. But I'm wondering why it doesn't always try to build settlers and workers as fast as possible.

 

[phungus420]

Great, will get to testing.

Was the 50 turn end war clause cancelled in this build? I think that clause alone makes AI domination impossible.

 

[kniteowl]

Hey Iustus, I Loaded BetterAI Using "Load a mod" but the ===Better AI=== Doesn't show up on the score list on the bottom right of the screen when I hold down ALT, does it mean the Better AI DLL file isn't installed in the game?

Do I have to use your method of installation to properly Use Better AI mod?