[MOD] More Naval AI

Hey guys, just downloaded the mod and took it out for a test run, and hit two issues

1) If I try and play Somnium I get a repeatable crash to desktop as soon as I end turn
2) Does the 'playing to win' in the mod allow friendly civs to declare war on you? I just got a rather unexpected declaration from someone I'm at +16 relations with at a time when we are both already at war with someone else.
 
If it helps I just tested and the somnium thing only occurs after I play for diplo modifier (playing a friendly game works fine) so I guess something calculating AI attitudes goes wonky
 
2) Does the 'playing to win' in the mod allow friendly civs to declare war on you? I just got a rather unexpected declaration from someone I'm at +16 relations with at a time when we are both already at war with someone else.

aaand while I remember, this war dec doesn't have the normal "now you shall pay for your many insults" -> So be it screen - it seems almost like one of the forced war decs that comes from e.g. starting to build the tower of mastery, but I can't think of what it could be from. I'm nowhere near the tower and Basium is built and dead (and we are both AV) so I don't think it's any order vs AV interaction. Is there anything else that forces a war dec?

edit: here's the save in case anyone wants a play; Tasunke declares when I end turn
 

Attachments

What was the Armageddon Count at?

AC is at 40 (It's been there for a while, Stephanos is already dead).

Something is definitely forcing the war declaration. I begged one gold to get the 10 turn unbreakable peace (It's still unbreakable for the one that gives in to the demand right?) and get the cannot declare war status but he still declares at end of turn.
 
That's it then. In Customfunctions.py there is a function called warScript. At AC40, any evil civ will try and find a weaker civ to declare war on if they aren't already at war. And yes, it's hardcoded to simply go to war. This is part of base FFH.

Huh. Never ran into that before. He is currently at war but I guess this is the root of the issue anyway :p Thanks for the help
 
He is currently at war

Interesting. You are correct. He is already at war with the elves. I don't see anything wrong with the code, so I'm not exactly sure what's going on. Interestingly enough, Tasunke is still 'Pleased' with you after declaring war!

EDIT: Think I figured it out. Looks like a Hippus Mage Scorched your Silk just southeast of Clar Marrachir, resulting in a war declaration. Looks like I need to add an extra python check to make sure they don't do that to their friends. Thanks again for the report!
 
One thing that's been jarring in my last couple games: the AI severely undervalues Iron as a resource. In three different games, I was able to trade a single luxury resource for one of the spare Iron resources from the AI who had built the Mines of Gal-Dur. I would have gladly paid everything I had for iron, since I was in the middle of a war in two of those games, and the instant +1 strength really turned things around.

In game #2, Varn actually dialed me up to offer Iron for Dye; in game #3, I asked Falamar what he'd give for Dye, and his answer was Iron + Pigs!

Dye is a resource that certainly seems to have a good value.

Anyway, not only Iron should have more value but the techs that work the metals should have it as well.

The AI takes too long to research Iron working when it could.

Btw, horses and axes should have priority over hunters for most civilizations, unless there is no metals or horses.

from the AI who had built the Mines of Gal-Dur

This reminds me that the AI will sometimes rush Arete without education or code of laws, meaning that it takes too long to boost commerce (and without bronze working or archery).

Those commerce techs should have more priority for everybody before further pursuits in the tech tree, I think.

An early religion will many times do more harm than good to the AI.

The AI should be better in bulbing religions to pursue the tech tree more effectively in the early game.


Btw, Tholal, is there anything at all that is possible to take of my game diaries regarding the AI? I'm sure that you have your own test games, but is there something specific that you would like to see reported?

Do continue your great work. We all appreciate the effort.
 
An early religion will many times do more harm than good to the AI.

Interesting. Is it because they spend too much time researching it? How early is too early for a religion?


The AI should be better in bulbing religions to pursue the tech tree more effectively in the early game.

Also interesting. I almost never use GP for researching techs (but then again, I play on Monarch level myself). When and why do you use this option rather than settling them or getting a golden age or building?


Btw, Tholal, is there anything at all that is possible to take of my game diaries regarding the AI? I'm sure that you have your own test games, but is there something specific that you would like to see reported?

Reading your diaries is useful. Though it's a little easier when I get specific comments such as the ones quoted above, or 'the AI doesn't value iron enough', or "more Pyre Zombies", etc. Then I have a specific spot in the code that I can go to and start working from.

But any and all feedback is great, whether it be comments, bug reports, game reports or even haikus!

Thanks to everyone for the continuing feedback. Hoping to release version 1.8 later this month. Here's a sneak preview of a few of the changes.

  • Major overhaul of the AI teching routines. The AI now takes a lot more factors into account and should be better at reacting and adapting to its current situation. This will be an ongoing area of work.
  • AI will actively seek out and explore Lairs and Dungeons (land and sea)
  • AI will be better about picking up its own equipment
  • Cast and resist messages will display the spell name and the name of the casting unit (also fixed a bug that was causing extra resist messages to appear)
 
Interesting. Is it because they spend too much time researching it? How early is too early for a religion?

It's because they many times ignore some of the basic technologies and take too long to boost commerce and get a proper defense. Many times they will even go straight towards the technology that is unlocked by the religion ignoring everything else.

I guess that this is also why on low difficult levels you sometimes encounter an AI spending too much time with just warriors.

On my Clan of Embers game diary, Decius researched Runes even without Calendar (and I don't remember him having any wine or gold so that crafting and mining was worth it in terms of commerce boost).
Without the Diety bonuses, what would he do to grow, increase his happy cap and get some basic commerce from resources early on? Even on diety he was not able to get away with that bad decision-making.

The three early religions must be bulbed always. There isn't a reason to not bulb them. That's how you can get them fast without falling behind in commerce and defense.

One always need mysticism to get a religion, so research that, switch to God King and pacifism (with God King you can build warriors fast enough without nationhood), start building a temple and afterwards run a priest. Meanwhile, you research with your land commerce the other necessary technologies to be able to get the religion.
If you want Runes you need mining, so you go towards that in the normal way and the priest will do his thing - you are basically doing two researches at the same time.
Depending of the traits of your leader and your land, you may be able to get runes at more or less the same time that you finish researching mining.

This is much faster, way more effective and also safer.
If the AI becomes able to do this (no idea how) it will be much harder up to rush it and it will tech more effectively early on.

Also interesting. I almost never use GP for researching techs (but then again, I play on Monarch level myself). When and why do you use this option rather than settling them or getting a golden age or building?

Bulbing strategies are a little bit more complicated than just simply farming everything and running aristocracy but they are fine with a philosophical leader.
You have to plan ahead and know the bulbing order.

A philosophical leader using guilds with big cities will be able to research multiple paths at the same time using bulbs. Since you're just basically using mostly fat farms and specialists, golden ages will not boost your commerce (but they do give a boost to GP growth).
A financial leader will just continue to use land commerce and Golden ages to boost that.

I can't give you a specific answer because it depends on your goal. Just plan ahead so that your research progression is faster with the help of bulbs.

On my White hand game, even though auric is not philosophical, I'll demonstrate a little bit of bulbing for late game technologies.

Reading your diaries is useful. Though it's a little easier when I get specific comments such as the ones quoted above, or 'the AI doesn't value iron enough', or "more Pyre Zombies", etc. Then I have a specific spot in the code that I can go to and start working from.

I am wondering, though, if you don't have too many written notes and things to do already.
Sometimes it's complicated to be specific since the AI screws up not because of just one mistake but because of a serie of small things.

You know that they don't tech effectively, you know that they don't handle their troops well enough still, and you know that they don't evaluate power properly (they're way better than vanilla, of course:)).

The AI is also incapable of "foreseeing" summons which makes it incapable of reacting properly to them.

Iron working should definitely have more value to the AI.

I also once commented that the current emphasis on drill promotions is probably not good. Warriors with just drill promotions are just asking to be rushed. Warriors are too weak and need combat promotions. Drill promotions are mostly for units that are already strong enough.

I also wish that the AI would consider more stuff when settling (like culture pressure, the overall quality of the tiles (they will settle on the ice if there is just one resource there)) instead of mainly the resources.

Make Auric AI stop spamming javelin throwers early on To attack cities and research Iron working and I'm sure that he will be one of the strongest AIs in the game.

Well, I better stop now. I am just repeating some of the things already said. There is no point in adding more stuff if there is still a lot of planned stuff to do.

  • Major overhaul of the AI teching routines. The AI now takes a lot more factors into account and should be better at reacting and adapting to its current situation. This will be an ongoing area of work.
  • AI will actively seek out and explore Lairs and Dungeons (land and sea)
  • AI will be better about picking up its own equipment
  • Cast and resist messages will display the spell name and the name of the casting unit (also fixed a bug that was causing extra resist messages to appear)

Awesome:)
But hopefully the AI will explore lairs with strong enough units;).
 
Another crash report at end of turn.

Hey Doug. This is the same crash bug as your previous report. An AI unit is trying to cast an area effect spell near the map edge. This will be fixed in the next version (hopefully coming soon), but until then, all I can recommend is that you play maps without a hard edge (or that dont have any landmasses near the map edge).

Apologies for the difficulties!
 
1) If I try and play Somnium I get a repeatable crash to desktop as soon as I end turn

If it helps I just tested and the somnium thing only occurs after I play for diplo modifier (playing a friendly game works fine) so I guess something calculating AI attitudes goes wonky

I'm unable to replicate this issue. Do you have a saved game with the repeatable crash that I can look at?
 
After spending about fifteen minutes regenerating maps and watching the AI city position circles, I have several remarks:

1) Farmed food specials are *much* better than pastured/plantationed ones. The tech comes earlier, the yield is usually higher, and they benefit from Aristograrianism. Seafood resources, which the map generator seems to prefer, are pretty poor. Floodplains are as good as rice or wheat, and frequently come in bunches. Settling near a swath of floodplains can really make the early game take off.

2) Commerce specials are better balanced than food. Calendar usually comes earlier, but the yields are lower. This gets distorted, however, by the map generator clearing off the jungled resources (dyes, sugar, gems), which are normally not accessible until bronze working. Starting by several floodplains and several de-jungled dyes is the jackpot in the start position lottery.

3) Copper is a huge deal in the early game. Warriors with bronze weapons = axes without.
 
Hey Tholal, can you make the AIs not use the conquest civic while running slavery? That'll probably be for the best, I believe.
 
Back
Top Bottom