Improving AI Germany vs AI USSR in 108

[unmerged(86927)]

It appears that the handicap of the Soviet AI versus Germany AI is not enough to simulate the historic initial gains of Germany, as in my test games the front reaches a stalemate soon after the German attack somewhere along the line west of Riga - west of Kiev - Odessa.
So if you have any suggestions of improving this, please contribute.

 

[Tegetthoff]

Some experiences from our recent work in CORE:

A big factor seems to be the scorched earth setting. The Soviets scorch as they retreat and this stops the Germans fairly quickly. Lowering the scorch setting has helped us a lot.

T.

 

[Pang Bingxun]

A higher priority for 1941 Inf might help to increase the firepower and the speed and might therefore reduce the duration the soviets can use scorched earth. So the tech 1040 needs to be researched earlier, the priortity of building (light) armour might be reduced in favour of more infantry and the IC allocated for upgrading needs to be increased a lot. Still this is likely to only increase the fighting capabilties. If the ESE i very low this will not help a lot. The IC allocated for repairing infra might be increased. Maybe the repairspeed itself might be increased. Currently in misc.txt we have the values 0.25 and 0.033. I suggest to change that by a factor of 4:

# Cost to repair Province buildings
0.25 -> 1.00
# Time to repair Province buildings
0.33 -> 0.0825

 

[Denniss]

That's too much of a change, too many IC would then be bound for repairs and those repairs would be lightning-fast. I'd rather use .33 for cost and .25 for time

 

[Fu.Th.Sy.]

How about increasing garrison priorities for Moscow, Stalingrad and Leningrad in SOV ai file, so that the main army joins the fight later on when Germany has already broken through thin Soviet defence lines. I think I saw this in WIF mod, not sure.

 

[Pang Bingxun]

Romania and bulgaria start with cavalry but are set to not research cav techs. This needs to be changed.


Edit: How about simply reducing the strenght of the soviet army by reducing its numbers? A higher priority to IC and Infra in the early years should do the trick. If the soviets have 300 divisions in late 1940 advances against them will be costly ones.

 

[Titan79]

Another suggestion would be - if possible - to reduce the Soviets' Ground Defence Efficiency from, say, mid-1941 until November/December 1941.

 

[froglegs]

I use some custom events for AI GER/AI SOV games playing as USA:

##########################
# Poorer ORG IC for SOV
##########################

event = {
id = 666040
random = no
country = SOV
trigger = {
exists = SOV

}

name = "AI_EVENT"
style = 0


date = { day = 3 month = january year = 1936 }
offset = 1
deathdate = { day = 29 month = december year = 1944 }

action_a = {
name = "AI_Event"
command = { type = max_organization which = land value = -10 }
command = { type = industrial_modifier which = total value = -20 }
}
}
##########################
# Poorer ORG IC for SOV wears off
##########################

event = {
id = 666041
random = no
country = SOV
trigger = {
exists = SOV

}

name = "AI_EVENT"
style = 0


date = { day = 2 month = january year = 1942 }
offset = 1
deathdate = { day = 29 month = december year = 1944 }

action_a = {
name = "AI_Event"
command = { type = max_organization which = land value = 10 }
command = { type = industrial_modifier which = total value = 20 }
}
}

I think this helps a lot with balance.

 

[Cakes]

Would havet thought the only way to do it is to cheat..
How about limiting initial Russian front line infantry and reducing the Russian ORG until December 1941.
Then maybe increasing the penalties for cold weather that Germany gets (and make cold weather equipment harder to research - or unavailable until 1942) whilst and also increasing the bonus Russia gets for their cold weather gear - from November until, say, the following spring 1942.

Hindsight.. A wonderful thing for a 'historical' military game. ;-)

 

[unmerged(86927)]

How about simply reducing the strenght of the soviet army by reducing its numbers? A higher priority to IC and Infra in the early years should do the trick. If the soviets have 300 divisions in late 1940 advances against them will be costly ones.

Then maybe increasing the penalties for cold weather that Germany gets (and make cold weather equipment harder to research - or unavailable until 1942) whilst and also increasing the bonus Russia gets for their cold weather gear - from November until, say, the following spring 1942.

I'd like to see these two
Anybody would like to script those things?

 

[Pang Bingxun]

I am a bit skeptical about those additional cold weather modifiers. How should that be achieved? Reduce the german morale in cold weather? Increase soviet org and morale during winter in order to make them more aggressive during winter? Is there the danger to "overskript" the hole thing? That can reduce the motivation of the human player. He wants to be the one playing the the game and not the one to be played.
Another option is to delay the USSR Great Patriotic War events till november. Right at the moment those events fire so early that germany gets an unnecessary advantage in the initial combats but as soon as it advances 1-3 provinces deep into soviet territory the soviet fight without any disadvantage left.

How many division did the soviets have at the beginning of the historic war? How many airplanes? How many did they lose in the early war? Balancing the IC invested into army or aircrafts and the amount of men and planes lost in the early war might be of help. How about giving germany a surprise attack bonus as japan gets it?

 

[unmerged(86927)]

I am a bit skeptical about those additional cold weather modifiers. How should that be achieved? Reduce the german morale in cold weather? Increase soviet org and morale during winter in order to make them more aggressive during winter? Is there the danger to "overskript" the hole thing? That can reduce the motivation of the human player. He wants to be the one playing the the game and not the one to be played.

I was thinking to apply any winter penalty/bonus only in AI versus AI scenario. None should occur if Germany or URSS is player.

 

[Cakes]

Are our expectations unrealistic - given that the Soviet AI is always 'expecting' Germany to attack and preparing as such.
It's not very historical, so why should we expect it to play out like it did historically?

 

[froglegs]

Are our expectations unrealistic - given that the Soviet AI is always 'expecting' Germany to attack and preparing as such.
It's not very historical, so why should we expect it to play out like it did historically?

Right on! This is why something like reducing Soviet production and reducing combat ability for the first 6 months of the war is one of the better ways to game this.

It sort of reminds me of my old board war game days playing a 1914 scenario where France had to be forced by the rules of the game to make a bunch of stupid attacks and lose a lot of men in order to give Germany half a chance of getting through Belgium and close to Paris.

 

[Cakes]

Only way you're going to get anything close to historical is by starting in June 1941! ;-)
(Something I would love to see in CORE)

 

[Mr_B0narpte]

How many division did the soviets have at the beginning of the historic war?

It had a total of 202 divisions on or near the Soviet-German front (and 303 divisions in total).

How many airplanes?

20,000 aircraft, with at least 7,133 of them operational in or nearly in range with the German border.

How many did they lose in the early war?

The Luftwaffe (claimed) to have destroyed 1,800 aircraft on the first day of the invasion alone. But this is hardly going to be replicated due to the combat system of the game.
In regards to the Red Army, on its 'western front' (i.e. against Army Group Centre alone) by 9th July 1941, it had lost (either destroyed or left behind in retreat) 4,799 tanks, 9,427 guns and mortars and 1,777 combat aircraft. Also it had lost (including all KIA, MIA, WIA, PoW and illness) 341,073 of its original 627,300 strength.
This is also inherently difficult to simulate, partly because of the way AoD models military divisions in the first place.

How about giving germany a surprise attack bonus as japan gets it?

That's easily doable, I created such an event a while ago but have since lost it. However it would take less then 5 minutes to create.

 

[kane279]

well how about re-setting the "peace modifier" for IC after the winter war? that would hamper the preparation of the USSR quite a bit.

 

[bestmajor]

as for the GPW idea from Pang, what about rasing the amount of lost territory?

#########################################
# USSR Great Patriotic War events by Lothos
#########################################

event = {
id = 2651
random = no
country = SOV


trigger = {
NOT = { event = 2657 }
OR = {
year = 1940
atwar = GER
lost_national = { country = SOV value = 2 }

eg from 2 to 5?

 

[Pang Bingxun]

Germany is likely to match the 8 percent for the last stage now already. Delaying them till the first winter at war gives the soviets some time seems more helpful to me.

 

[unmerged(86927)]

Tweaking Great Patriotic War events to allow USSR a slower recovery also seems a good idea, unless it makes human Ger vs AI USSR too easy. Consider including checks for Germany AI. Can someone provide a full set of events for evaluation? (a modified USSR.txt from the last Beta) I hope a new Beta version would be up soon and I'd like to include some changes that makes Ger AI advance a little further into USSR before stalling.