tweaking 1.07 - modding ai files/switches

[Weserübung]

since I am too impatient to wait for 1.08 I went ahead and took the liberty to change the ai files from all major powers as well as Can and Ast.

After some testing, it goes quite well. The ai now builds a more resonable force with different brigades and actually researches artillery techs. No more 50 Arm for Ger or no Inf for Jap and Sov prior to the war. also the omnipresent eng brigade are the past now. war has a more realistic outcome too.

still some help needed.

I notice that Eng simply adds to their existing queue of ships and keeps adding even though it hasnt enough IC to build them all. This leads to CV III still in queue in 44. Also Jap runs out of manpower by 39 and the US by 44 even though their force isnt that great? If I include light arm into the construction coding the ai still builds them even thoug it has already arm V. ai generally spends too little IC on modernization.

Id appreciate advise how to further tweak this great game

 

[Fu.Th.Sy.]

Could it be that ''max_batch_peace'' and ''max_batch_war'' are set to low for UK. Other nations have a longer serial production.
Only guessing :?

 

[Denniss]

Some countries also build ships by event, either from their ai event file or from reactive ai event file. If countries do not reinforce or upgrade enough you have to change the factors in the main ai file and check the switch files for those settings. The eng 1936 ai file has 0.15 as factor for upgrades = 15% of IC max and 0.2 = 20% of IC max. ENG is also set to build a lot of ships in this file, several switch file may change the build priorities more towards land unit builds.

 

[unmerged(151142)]

Yes, the batch sizes are a bit tricky since the AI cannot upgrade existing production lines, like a human player would. If you increase their number the Ai will gain serial boni but will have tons of outdated stuff, very bad with ships of course, I guess that's why UK has lower batch sizes?

Whooo, I never saw Japan run out of MP actually, how many divisions does he have at that point though? I tried a balance where main focus remains on naval and air forces, so Japan will take the Chinese coast but not all China until it declares war on US. (24% Inf in the build scheme, and 735 MP in the scenario file to begin with)

 

[Kirkegaard]

The AI upgrades a serial line, every time a unit is produced if it can.

 

[unmerged(151142)]

The AI upgrades a serial line, every time a unit is produced if it can.

Are you sure? Every time I look into the AI's production queue I see him producing outdated units that he could take one step ahead. Should this be the case that he just recently researched the necessary technology for this upgrade so no unit of that line had been finished meanwhile? Must have happened pretty often then. If so, one should really increase batch sizes a bit, no?

 

[Kirkegaard]

Yes, I reported it as a bug myself but "Leonartos" said it was working as I descript above.

 

[RogueLeprechaun]

since I am too impatient to wait for 1.08 I went ahead and took the liberty to change the ai files from all major powers as well as Can and Ast.

After some testing, it goes quite well. The ai now builds a more resonable force with different brigades and actually researches artillery techs. No more 50 Arm for Ger or no Inf for Jap and Sov prior to the war. also the omnipresent eng brigade are the past now. war has a more realistic outcome too.

still some help needed.

I notice that Eng simply adds to their existing queue of ships and keeps adding even though it hasnt enough IC to build them all. This leads to CV III still in queue in 44. Also Jap runs out of manpower by 39 and the US by 44 even though their force isnt that great? If I include light arm into the construction coding the ai still builds them even thoug it has already arm V. ai generally spends too little IC on modernization.

Id appreciate advise how to further tweak this great game

Sounds like you have spent a fair bit of time considering AI files.

Obviously the Major powers are the most important area for modification to make sure they have good AI files. However, I have found quite a few minors have some strange build schemes. Manchuria for example, producing mainly/only? Garrison without the tech required to strategically redeploy them. Resulting in a massive pile of Garrisons unable to move in one or two provinces.

I am also in the process of rejigging build schemes, etc to create more plausible OOBs for different powers. Are you able to post some of your edited AI files for us to have a look at?

 

[unmerged(236565)]

Indeed the sheer number of garrison units in 1.07 is amazing. Manchuria is only one example. Others are Persia, Mejing, Saudi-Arabia and Irak. I have a feeling this is to make them stronger since GAR is better in defence then MIL. Without rear area supply dump however they are useless. And some nations should not build GAR at all like USA but their AI builds them too often IMHO. They even attach MP brigade to it. That is stupid because it takes 2 MP and USA never needs that since they have far lower partisan levels due to their slider setting.
I sometimes think the AI builds only by unit stats. So they go ahead and build GAR (which is quite good if you compare stats and cost) and heavy subs (yeah they build them instead of type IV sub because they are better but more easily found so in fact worse) The waters are filled with heavy subs which is quite ahistorical. There should be a ratio of 1 heavy to every 5-6 normal subs at most.

 

[Mr_B0narpte]

Indeed the sheer number of garrison units in 1.07 is amazing. Manchuria is only one example. Others are Persia, Mejing, Saudi-Arabia and Irak. I have a feeling this is to make them stronger since GAR is better in defence then MIL. Without rear area supply dump however they are useless. And some nations should not build GAR at all like USA but their AI builds them too often IMHO. They even attach MP brigade to it. That is stupid because it takes 2 MP and USA never needs that since they have far lower partisan levels due to their slider setting.
I sometimes think the AI builds only by unit stats. So they go ahead and build GAR (which is quite good if you compare stats and cost) and heavy subs (yeah they build them instead of type IV sub because they are better but more easily found so in fact worse) The waters are filled with heavy subs which is quite ahistorical. There should be a ratio of 1 heavy to every 5-6 normal subs at most.

I completely agree.

 

[Weserübung]

as example here is the build sheme for eng_36

#### Divisions etc...
infantry = 12
cavalry = 0
motorized = 8
mechanized = 5
light_armor = 1
armor = 6
paratrooper = 1
marine = 1
bergsjaeger = 0
garrison = 0
hq = 1
militia = 0
# 35 %
interceptor = 5
multi_role = 9
cas = 0
strategic_bomber = 9
tactical_bomber = 5
naval_bomber = 1
transport_plane = 1
flying_bomb = 0
flying_rocket = 0
# 30 %
battleship = 0
carrier = 10
escort_carrier = 5
destroyer = 10
light_cruiser = 5
heavy_cruiser = 0
battlecruiser = 0
submarine = 0
nuclear_submarine = 0
transports = 15
# 35 %

#### Brigades
artillery = 45
sp_artillery = 20
rocket_artillery = 5
sp_rct_artillery = 5
anti_tank = 10
tank_destroyer = 5
light_armor_brigade = 1
heavy_armor = 1
super_heavy_armor = 0
armored_car = 2
anti_air = 1
police = 0
engineer = 5
# 100%

I also added tech endgoal to make them research arty brigades

# Artillery

11110 11020 11040 11060 11070 11090 2360

also disabled ignore techs

 

[unmerged(236565)]

Good stuff!
But I'd take off 1 point each of carrier and escort carrier and give 2 points to battleships. After all they build 4 King George V. So like that they might build 1. Great move with the tech goal but maybe it would be a good idea to make all nations (except GER and USA) ignore RAIL brigade. Everybody loves to research that - it takes really long time and is not useful at all. As Germany I got blueprints for that from Italy in 1940. From Italy !!!!! Damn this must have taken them an awful long time.
Btw. do you have the the values set to ratio or to absolute numbers? I have heard that absolute is better because that way the actual builds stay closer to your intentions.

 

[Weserübung]

I think it is absolute numbers. Eng builds 1 BB IV anyway and they have enough BBs too. not need to waste ai resouces that way. I am not too concerned about the railbrigade reasearch makes enough progress as it is but ai doesnt upgrade and produces right..

to illustrate what I mean

 

[Weserübung]

this stacking of queue lets them lose manpower fast as it is deducted when production is assigned

results in the longer run

Jap made superb progress in CHI and at one point controlled the entire coast but got pushed back due to lack of manpower

even worse with the allies

USA and ENG have NO manpower without building a big army

nevertheless the results are still better than without the tweaks, but germany and italy actually gaining ground again against the SOV

 

[unmerged(236565)]

Yeah UK still has to short serial builds it seems. Now I finally understand where all the manpower goes I'll send you my event chain fixes per mail now. Without them you can try as you might JAP will never win against über CHI. Same goes for winter war. SMEP now uses the historical date to end it. That is fine but SOV usually wins before that. That means SWE always joins the axis and scandinavia becomes a major battlefield. I dont like it because it takes away troops from the other fronts.
Any way to make UK AI set the research slider to 75% and not 100% ??? Because that is what a human player does to free IC from the massive consumer goods production and build units instead. 75% still gives you results in a reasonable time and later US can help with blueprints.

dude the red flood is coming and the allies cant make a working landing - Europe will be soviet for sure
and what the heck goes on in Spain? and why is Gibralta not taken when all else around is axis? I hate it. I have seen it in my german game as well. Bitter peace - NAT SPAIN joins axis and does not take Gibralta even when having a massive superiority of troops there with mountaineers too. So I took military control just so that they get their butts moving there before UK could send 100 divisions as usual.

 

[Julle64]

this stacking of queue lets them lose manpower fast as it is deducted when production is assigned

results in the longer run

Jap made superb progress in CHI and at one point controlled the entire coast but got pushed back due to lack of manpower

I also noted this in my modified test games. I added 800 MP for Japan (AI) and it worked quite well. Maybe the MP modification should be done via difficulty.csv so that all AI controlled nation have more manpower?

 

[Julle64]

... Same goes for winter war. SMEP now uses the historical date to end it. That is fine but SOV usually wins before that. That means SWE always joins the axis and scandinavia becomes a major battlefield. I dont like it because it takes away troops from the other fronts. ...

I also modified some files due to this and the Soviets haven't won the Winter War since. I gave all the Finnish INF division an ENG brigade and gave Finland a GDE of 0.90 (IIRC I copied the GDE values from Mjarr's Mini Mod).

EDIT: And another reason might be that I have also modified the Soviet AI files so that they don't build so many tanks and motorized anymore. So no more Winter War where the Soviets are attacking with 20+ ARM divisions

 

[Weserübung]

dude the red flood is coming and the allies cant make a working landing - Europe will be soviet for sure
and what the heck goes on in Spain? and why is Gibralta not taken when all else around is axis? I hate it. I have seen it in my german game as well. Bitter peace - NAT SPAIN joins axis and does not take Gibralta even when having a massive superiority of troops there with mountaineers too. So I took military control just so that they get their butts moving there before UK could send 100 divisions as usual.

the soviets are actually pushed back right now. they already had Greece, Milan, Danzig and Breslau but axis recovered by moving almost all they have to the east. It wont seave them but it allowed allies to land. there are almost no units in france so uk will spread out soon. but US wont be helping as they have only five transports left. so the landing will fail to liberate europe before it is red.

spain was still in civil war when rep spain joined comintern in 45 so nat spain joined axis. then italy landed in valencia. Italy makes lots of landings, see the reconquering of greece. Uk actually took all that is german now in spain. italy briefly took gibralter but a german landing in seville ended UKs progress, now they have some troops there and bomb axis too. so stalemate for now.

 

[Weserübung]

anyway what do I have to do to make everybody build longer serials and assign more IC to modernize?

 

[unmerged(236565)]

Not only modernize but also to repair their provinces. As Germany I can strat bomb UK back to the stone age with TAC (12) (with some STRAT docrines researched that is) Their flak is useless and their fighters are no match (too few) once INFRA and IC is down it takes ages to recover. They even lose a tech team because of that. So I already play with the house rule no strat or infra bombing - only IC bombing. With an intact infrastructure the rebuild is at least a bit faster.