Why does the Russian stay at home?

[OneArmedMexican]

It might be coincidence, but in my games the AI shows a strange pattern while using the Russian troops. While playing the 56-63 campaign, the Russian attacks Königsberg in 57 with all his might (beautifully executed and virtually irresistible). After conquering the city, however, the Russian retreats back to the east; never to be seen again. Huge amounts of troups are just sitting in the rodina. Each summer I expect a great offensive on Kolberg, but nothing ever happens. In my last game, I reacted to that pattern and left nothing but a small token force in Kolberg. Still, the Russian couldn't be tempted.

Is there something wrong with the AI or is it just a coincidence?

[Hohenlohe]

It might be coincidence, but in my games the AI shows a strange pattern while using the Russian troops. While playing the 56-63 campaign, the Russian attacks Königsberg in 57 with all his might (beautifully executed and virtually irresistible). After conquering the city, however, the Russian retreats back to the east; never to be seen again. Huge amounts of troups are just sitting in the rodina. Each summer I expect a great offensive on Kolberg, but nothing ever happens. In my last game, I reacted to that pattern and left nothing but a small token force in Kolberg. Still, the Russian couldn't be tempted.

Is there something wrong with the AI or is it just a coincidence?

Which version are used by you??

greetings

Hohenlohe

[Clovis]

It might be coincidence, but in my games the AI shows a strange pattern while using the Russian troops. While playing the 56-63 campaign, the Russian attacks Königsberg in 57 with all his might (beautifully executed and virtually irresistible). After conquering the city, however, the Russian retreats back to the east; never to be seen again. Huge amounts of troups are just sitting in the rodina. Each summer I expect a great offensive on Kolberg, but nothing ever happens. In my last game, I reacted to that pattern and left nothing but a small token force in Kolberg. Still, the Russian couldn't be tempted.

Is there something wrong with the AI or is it just a coincidence?

There was a problem with my AI files for 58 and after. I've fixed this. I don't know if Russian AI will be better but at last there's one trouble gone

Files are located here:

http://cid-f2e0d7dbcedffac0.office.live.com/self.aspx/.Public/ROPAI.zip

They must be used with the last Open beta patch 1.03b...

Unzip in the ROP /events directory. It would be better to start a new game. Or wait a full year before assessing if Russians are better.

[OneArmedMexican]

Which version are used by you??

greetings

Hohenlohe

1.02, I haven't tried the open beta for 1.03 yet.

When I finished my last campaign as Prussian, I took a peek. It was the end of 59: ~100.000 men were sitting around in Russia. Surprisingly I found another 85.000 doing the same in France and finally 120.000 French and HRE troops having a tea-party in Straßbourg. After taking a bloody beating in Westfalia in 58, the French hadn't shown his face all through 59. Guess the tea party took a lot of preparations.
It was as if the AI was only playing with its Austrian and Swedish troops.

[Clovis]

1.02, I haven't tried the open beta for 1.03 yet.

When I finished my last campaign as Prussian, I took a peek. It was the end of 59: ~100.000 men were sitting around in Russia. Surprisingly I found another 85.000 doing the same in France and finally 120.000 French and HRE troops having a tea-party in Straßbourg. After taking a bloody beating in Westfalia in 58, the French hadn't shown his face all through 59. Guess the tea party took a lot of preparations.
It was as if the AI was only playing with its Austrian and Swedish troops.

Russianas have a big supply problem in the 1.02

[OneArmedMexican]

There was a problem with my AI files for 58 and after. I've fixed this. I don't know if Russian AI will be better but at last there's one trouble gone

In 58, the AI seemed fine to me. It used the French and Austrian troops coordinated and rather agressive.

I guess I will have to try the beta patch after all.

Thank you very much.

[OneArmedMexican]

Russianas have a big supply problem in the 1.02

I noticed that when playing the other side. Supply problem is an understatement, I would call that famine!

[Baris]

There is also famine for ai(Aus) for attacking Dresden. It suffers from to much attrition and lack of supply. In 3 turns there was total of 200 or more notifications for ai.
Stettin was captured by swedish troops under French General. French are doing well.
Russians captured Königsberg in 1757 in september if I remember correct.. But they waited for 1758 for marching to Kolberg. They are on the way. But Im afraid they will perish from attrition and lack of supply. Which I think is the most important problem for ai.

not related but ai likes to move fort batteries like normal field artilerry. They move individually.

ıt is after Hands off game as prussians. As supply depots not working.

[Clovis]

There is also famine for ai(Aus) for attacking Dresden. It suffers from to much attrition and lack of supply. In 3 turns there was total of 200 or more notifications for ai.
Stettin was captured by swedish troops under French General. French are doing well.
Russians captured Königsberg in 1757 in september if I remember correct.. But they waited for 1758 for marching to Kolberg. They are on the way. But Im afraid they will perish from attrition and lack of supply. Which I think is the most important problem for ai.

not related but ai likes to move fort batteries like normal field artilerry. They move individually.

ıt is after Hands off game as prussians. As supply depots not working.

With my latest files?


Edit: Austrian are peculiarly exposed to supply problems during winter as LOS are blocked in mountains. But I've just devised something to help AI to avoid such a miserable fate (maybe! not tested, nor written )

[Baris]

Yes,
They are near to Kolsberg but supply wagons depleted. Another Stack is going to join the march from far east. But generally the logic of ai is good. But marches from far can be devastating. Im like watching a movie with hands off game if they can not conquer in first attempt they are counted as Martyr


I wonder what they going to do after(İf) they capture Kolsberg.
I move just some small troops to kolsberg to see if Russians will be out of suppy while sieging.

Edit: My mistake Stettin wasnt captured. Swedish troops captures wittenberge and they are on the way to stettin.

Russian cossacks captured Berlin, I missed that. Siege by russians in kolsberg continues. Russians lost half of the troops during 1 or 2 turn siege.

Troops captured Berlin(russian) are sieging Frankfurt, it is all well.
1 breach achieved in kolsberg but russian main army is now 20000 troops.
They are all red.

Edit 2: stettin under siege
Im defeated in Kolsberg but russian main army retreated 2 regions far(if I saw right). They are positioned and supplied from Gdansk and coming to attack again to kolberg.

Russian army looks active but supply problems I think is the main problem. . Not necessary for Russian ai, Aus Ai can also lose many forces when it is winter. But this AI looks very promising, if only supply problems can be solved in some way. As in the picture below Hadik's army transformed into supply wagon

Austrian and Russians had the most stacks concentrated and united.
Russian main army lost many time sieging kolberg. When there is failure in siege or winter approaches "Armiya Poltschi return to Gdansk. If he can forget it and march to inner Germany, he would have created more havoc against Prussians.

Stettin wasnt captured even there was only 1000 men and some artilery in it. And I think AI does not get support from ships when sieging near river forts.


[ATTACH]12863[/ATTACH]

[ATTACH]12864[/ATTACH]

There is also CP penalty in biggest forces. There are 10 formed armies by 3 star generals. Most the armies are have either CP penalty or they have very few men in some cases. Artilery and supply wagons and fort batteries mostly travel alone in the map. They can be easy targets.
But AI is promising with the moves. If Russians can have more efficient army corps in some cases and dont have some obsessed cities as target selection, it will be more challenging for the player. But it can change game to game.

[Clovis]

Yes,
They are near to Kolsberg but supply wagons depleted. Another Stack is going to join the march from far east. But generally the logic of ai is good. But marches from far can be devastating. Im like watching a movie with hands off game if they can not conquer in first attempt they are counted as Martyr


I wonder what they going to do after(İf) they capture Kolsberg.
I move just some small troops to kolsberg to see if Russians will be out of suppy while sieging.

Edit: My mistake Stettin wasnt captured. Swedish troops captures wittenberge and they are on the way to stettin.

Russian cossacks captured Berlin, I missed that. Siege by russians in kolsberg continues. Russians lost half of the troops during 1 or 2 turn siege.

Troops captured Berlin(russian) are sieging Frankfurt, it is all well.
1 breach achieved in kolsberg but russian main army is now 20000 troops.
They are all red.

Edit 2: stettin under siege
Im defeated in Kolsberg but russian main army retreated 2 regions far(if I saw right). They are positioned and supplied from Gdansk and coming to attack again to kolberg.

Russian army looks active but supply problems I think is the main problem. . Not necessary for Russian ai, Aus Ai can also lose many forces when it is winter. But this AI looks very promising, if only supply problems can be solved in some way. As in the picture below Hadik's army transformed into supply wagon

Austrian and Russians had the most stacks concentrated and united.
Russian main army lost many time sieging kolberg. When there is failure in siege or winter approaches "Armiya Poltschi return to Gdansk. If he can forget it and march to inner Germany, he would have created more havoc against Prussians.

Stettin wasnt captured even there was only 1000 men and some artilery in it. And I think AI does not get support from ships when sieging near river forts.


[ATTACH]12863[/ATTACH]

[ATTACH]12864[/ATTACH]

There is also CP penalty in biggest forces. There are 10 formed armies by 3 star generals. Most the armies are have either CP penalty or they have very few men in some cases. Artilery and supply wagons and fort batteries mostly travel alone in the map. They can be easy targets.
But AI is promising with the moves. If Russians can have more efficient army corps in some cases and dont have some obsessed cities as target selection, it will be more challenging for the player. But it can change game to game.

I can't do anything forCP penalty as there is no scripting command for this; Ageod is welle aware of the point, and is working on.

I belive the current ROP exe has yet the suppply bug starving Russians. In any case, I will look at winter AI mechanisms, because there's obviously something to be done...

It seems French are preforming well too.

Thanks for the report, because it's good news for ROP...and for RUS as the AI is based on the same logic.

[Baris]

By the year 1759 february or march, 38 units(and additional forces)of Russian army(40000-50000 troops) destroyed on the way to Gdansk. There is no russian threat now. But supply is also difficult for the player as playing russians. As AI not building depots and the distance is to far to handle it is inevitable. There is still some winter movement of troops. That is undermining the smart moves of AI.

3 star Russian Generals was commanding supply wagons in central Germany while they starve. They need to form corps so they wont have stack penalty and move penalty.

Swedish ai is mixing forces with french, They can be more interested in Stettin with her own forces as It was not defended by me. If swedish can maintain more concentrated forces and Russians able to handle supply and more reasonable priority of cities when attacking it will be good challange for the player. IMHO concentration parameter of some forces can be tweaked further for the AI to have much greater army size. The CP tweak is also needed. AI tends to form army from all 3-star generals, resulting some small armies. 3 star generals can asist CP penalty stacks by forming corps more.
But AI is very promising and much better generally.

[Clovis]

By the year 1759 february or march, 38 units(and additional forces)of Russian army(40000-50000 troops) destroyed on the way to Gdansk. There is no russian threat now. But supply is also difficult for the player as playing russians. As AI not building depots and the distance is to far to handle it is inevitable. There is still some winter movement of troops. That is undermining the smart moves of AI.

3 star Russian Generals was commanding supply wagons in central Germany while they starve. They need to form corps so they wont have stack penalty and move penalty.

Swedish ai is mixing forces with french, They can be more interested in Stettin with her own forces as It was not defended by me. If swedish can maintain more concentrated forces and Russians able to handle supply and more reasonable priority of cities when attacking it will be good challange for the player. IMHO concentration parameter of some forces can be tweaked further for the AI to have much greater army size. The CP tweak is also needed. AI tends to form army from all 3-star generals, resulting some small armies. 3 star generals can asist CP penalty stacks by forming corps more.
But AI is very promising and much better generally.

I will produce shortly something for winter. But AI will keep a possibility, albeit less often, to be active during winter. Total passivity would be predictable to human player, and predictability is killing AI.

Force mixing: I can't do nothing for that, because of a limitation of the AGE engine. Indeed, all the coalition forces are united under the same AI: there is no possibility to assign different objectives let's say for Swedish and French armies.

Supply: After the next patch, without supply bug, maybe AGEOD wil have to consider adding some AI helper events for depot building.I've not introduced that waiting for considering its necessity after next patch.

[Baris]

Yes If there is harsh rules, winter moves can be very predictable by the player.
I observed it happens when sometimes Austrians or French march from HRE area to Dresden or other near forts. When they come to siege they have already lost most of their troops if winter started. Russians on the other hand are totaly or partially eliminated if they can not conquer kolberg in the new files. If they are eliminated they are out of the war in early 1759.

Russians invasion about memmel or könisberg is good. They wait 1758 summer to start marching to Kolberg. In kolberg the theater getting bigger and they have to make better priority about the attack or maintaining supply. Or they can have have different priority about region.

not necessary AI to each nations but I think there is some kind of Region AI or priority of regions or theaters.(educated guess or just guess )Region or map priority balance for AI is the most difficult thing to balance I guess. But when the theater gets bigger it can not evaluate the situation or the time takes to finish some marches or calculate her supply.
I have played with high LOS bonus and I agree AI is smart. I see AI very promising thats why I feel it has pottential to be better. But supply is the main problem seems, especially for Russian part.

[Clovis]

Yes If there is harsh rules, winter moves can be very predictable by the player.
I observed it happens when sometimes Austrians or French march from HRE area to Dresden or other near forts. When they come to siege they have already lost most of their troops if winter started. Russians on the other hand are totaly or partially eliminated if they can not conquer kolberg in the new files. If they are eliminated they are out of the war in early 1759.

Russians invasion about memmel or könisberg is good. They wait 1758 summer to start marching to Kolberg. In kolberg the theater getting bigger and they have to make better priority about the attack or maintaining supply. Or they can have have different priority about region.

not necessary AI to each nations but I think there is some kind of Region AI or priority of regions or theaters.(educated guess or just guess )Region or map priority balance for AI is the most difficult thing to balance I guess. But when the theater gets bigger it can not evaluate the situation or the time takes to finish some marches or calculate her supply.
I have played with high LOS bonus and I agree AI is smart. I see AI very promising thats why I feel it has pottential to be better. But supply is the main problem seems, especially for Russian part.

I've just uploaded new AI files to cure the Colberg case, hopefully.

About theaters and regions: you're right, but when upping AI interest for let's say Stettin, you will create attractions to Stettin for Russian, Swedes and even Austrian or French...Fortunatly, Austrian and French aren't moving very often to Stettin, but it's how it works in ROP ( not in RUS).

Supply bug solved, I'm pretty confident Russians will perform much better. After all, Cossacks have tajken Berlin, don't they?

[Baris]

I think you hit the jackpot this time
(High LOS bonus,+1 activation bonus)
it is september 1758 and very few AI troops lose men from winter or movement. Russian couldn't capture Könisberg in first attempt but they withdrew and get supply carefully from the mother Russia.

But from my half hands- off game(just did small tweaks to my troops locations) AI tends to lose men sieging forts near the river. When sieging Stettin France lose half of the troops.After some time Austrian force came and starved for lack of supply while sieging.
When Austrian troops siege Glogau fort(level 2) they were also destroyed from lack of supply.(Both happened in summer) . Glogau fort chosen by AI because it had very weak defensive forces. But it is a Fort after all.
Russia is now captured in Könisberg with huge army. What she will do will be interesting to see. But depot problem is undermining more accurate play tests.

Edit: (1759 february) 3 star general Aus commander and 3 corps meet and concentrate forces in Poznan(poland level 1 city)) and they continue their way north to Torun, they lost 1/4 forces on the way. Torun is a level 1 Fort only.They will lose more. 20000 troops are red about supply.

France and Russians part is ok(russians didnt attack kolberg yet but at least moving to border depot and sometimes to memel in winter from Könisberg) but Austrian forces near Prag really struggle to have concentrated and supplied troops.

IMHO there should be some ai helper about building depot maybe half the price of what player pays. Also supply production from the depots can be 2 times more efficient then players. If in case of threat own depots can be destroyed by AI. Otherwise it is too much for ai to handle supply. It can change game to game but Austrians struggled to have supplied and orginized troops. French and Russian organization and supply is good.
Thanks for your work, despite the negatives it is improved from yesterdays hands off game.

[Clovis]

Thanks for you: such a post saves me several hours of test. I'm short on time, any way to get more time for scripting is precious.

Now for supply, I will wait more feedback. Creating such events isn't difficult, but before doing something that could be viewed as AI cheating I will need both proof of its absolute necessity and AGEOD agreement.

[Baris]

Movements in winter is ok,
Actually the problem I have observed(with 2 latest AI files ) are siege procedure of the AI. From my experience 70 % of the out of supply stacks are the result of long sieges especially near river forts. AI estimates that she can win but when sieges get longer there must be some procedure of retreating not only in winter. French and Russians(Russia still könisberg) didnt disseapered out of supply because they didnt have long sieges.Once French siege Stettin then they lost men for out of supply.
On the other hand Austrian forces in the main German land lost many while sieging. I will test what will Russia do. it is 1759 april.

Daun(best commander of coalition)Kollowrat are out of supply while sieging.
Kheul army(30000 men) returning from Torun to Poznan partially out of supply But even the desicion is wrong to go there he didnt lost many many troops like the 2 other. Edit: He reurned to Torun and 5000 troops left.


[ATTACH]12898[/ATTACH]

[Bertram]

Pocus confirmed that player (and AI) build depots dont receive and send supply. This confirmed what Squarian and me saw in our pbem campaign. Both the Russian depots as well as depot north and south of Kassel and a rebuild depot at Pirna stayed empty.
It looks like the depots you build in an already existing structure (like a town or fortress that didn't have a depot, or whose depot is destroyed) DOES function correct. I am not sure about that - I have only one example (a rebuild depot at Kosel).

Untill a patch is released to correct this, it is pointless to have the AI "learn" how to build depots, or tweak the AI to make the Russians more effective - they are hampered more by the depot issue then by bad routines at the moment. Infact any routine that gives an optimal result under the current conditions is probabely less than optimal once the supply issue is fixed.

[Baris]

Yes Bertram, Thats why it is mostly hands-off game. I m just observing the movements. AI does not build depot at all so it should not effect much. When Russian captures Könisberg I think they get double amount of supply.

Russian was obsessed with Kolberg back and forth attacking. Another point is long sieges, it can create lack of supply for AI. Maybe there is no need for AI to build depots, just sieges can be problem.

Depot built in current structures are also problem I observed. Im waiting eagarly for the next patch, But I just played to see how efficient AI can move troops or maintain supply.

[Clovis]

Yes Bertram, Thats why it is mostly hands-off game. I m just observing the movements. AI does not build depot at all so it should not effect much. When Russian captures Könisberg I think they get double amount of supply.

Russian was obsessed with Kolberg back and forth attacking. Another point is long sieges, it can create lack of supply for AI. Maybe there is no need for AI to build depots, just sieges can be problem.

Depot built in current structures are also problem I observed. Im waiting eagarly for the next patch, But I just played to see how efficient AI can move troops or maintain supply.

We'll see then. I've posted a new version in the beta thread lowering some values to attenuate indeed the siege obseession mentioned , and noticed by others. But my first aim was to force Russians to move. Now, let's see if they move with lower values, and if thse new values helps AI to choose resupply rather than pursuing sieges.

I've maybe found a solution to AI organization problem. MAYBE. If you want to give a try just place this file in the ROP/Settings directory

[Baris]

We'll see then. I've posted a new version lowering some values to attenuate indeed the siege obseession mentioned , and noticed by others. But my first aim was to force Russians to move. Now, let's see if they move with lower values, and if thse new values helps AI to choose resupply rather than pursuing sieges.

I've maybe found a solution to AI organization problem. MAYBE. If you want to give a try just place this file in the ROP/Settings directory

I did all the procedure again(put again the new files and lodilefty's hotfix). Started the game in 1757 april. I got 2 warnings in script report in first turn. If you can check it, I can continue to test the Russian AI,supply and AI organization files.

[Clovis]

I did all the procedure again(put again the new files and lodilefty's hotfix). Started the game in 1757 april. I got 2 warnings in script report in first turn. If you can check it, I can continue to test the Russian AI,supply and AI organization files.

these 2 warnings aren't related to my events , but for Daun's replacement. In any case, they just hinder these events to apply properly, but as they are here since some months, you may play without problem .

BTW, you should post this to AGEOD as bug report.

I'm very interested to see how AI organization will evolve with these settings. I don't belive you will not see some CP problems, but they could be less numerous.

[Baris]

1 warning this time and,

does "Too many occurences for event evt_nam_CMN_BattlePlassey1757 further processing aborted." means event fired twice. is about the hotfix? I can just pass those but Im asking if it effects gameplay? if no critical for ai I can continue.

[Clovis]

1 warning this time and,

does "Too many occurences for event evt_nam_CMN_BattlePlassey1757 further processing aborted." means event fired twice. is about the hotfix? I can just pass those but Im asking if it effects gameplay? if no critical for ai I can continue.

Once again, it's not an event created by me. No relation to AI, and here since many months. Once again you shoud report it to AGEOD, and continue testing AI

[Baris]

Once again, it's not an event created by me. No relation to AI, and here since many months. Once again you shoud report it to AGEOD, and continue testing AI

Then I must say Orleans(2 star general French) lost all of his troops for out of supply while sieging Hannover in summer. Area was crowded or outnumbered by French troops but he decided to siege alone with command penalty. I think that if it ever happens ,it will be excellent for AI to have troops orginized first and then attack without command penalty when there is not urgent situation. As French has many numerical advantages in the west they can attack more patiently. The most bad thing is AI losing troops while sieging. On the other hand Russians stormed Könisberg , I hope their faith will be well

[Clovis]

Then I must say Orleans(2 star general French) lost all of his troops for out of supply while sieging Hannover in summer. Area was crowded or outnumbered by French troops but he decided to siege alone with command penalty. I think that if it ever happens ,it will be excellent for AI to have troops orginized first and then attack without command penalty when there is not urgent situation. As French has many numerical advantages in the west they can attack more patiently. The most bad thing is AI losing troops while sieging. On the other hand Russians stormed Könisberg , I hope their faith will be well

For better work, I need 2 files when a problem arises:

- the Scriptreport in the Scripts folder
-the AI log file in the ROP/logs directory.

Thanks again. I fear AI will always lose some troops as you mentioned, I just hope it will be les often.

[Baris]

Thanks it is Looking good! winter sieges were stoped at some point.But I have to look at the summer sieges.
Only Russian army(biggest stack) and one big stack in Bavaria(Nuremberg) area have Command penalty.
As depots not giving enough supply Daun's big army needs to move around to find supply. Rus are going to Kurland as depot supply pruduction or distribition problems.

[Clovis]

Thanks it is Looking good! winter sieges were stoped at some point.But I have to look at the summer sieges.
Only Russian army(biggest stack) and one big stack in Bavaria(Nuremberg) area have Command penalty.
As depots not giving enough supply Daun's big army needs to move around to find supply. Rus are going to Kurland as depot supply pruduction or distribition problems.

I've no tools to force AI to stop a siege in summer, or I would just force AI to full passivity . I can just hope better supply, lesser attrition and lowering of the AIinterestvalues for cities will help to limit the occurence of such decimations...

For CP, I've seen less CP challenged stacks. But that's until now only a feeling.

Russians needs to capture the 2 objectives. Sometimes they fail, but in many games, they usually achieve this first step by autumn 1757.Considering how terrible is the land and precarious their supply lines from the start, it's just a sort of miracle

[Baris]

I've no tools to force AI to stop a siege in summer, or I would just force AI to full passivity . I can just hope better supply, lesser attrition and lowering of the AIinterestvalues for cities will help to limit the occurence of such decimations...

Then we are in agreement . But I have some mix feelings about the ai desperate continuing of sieging. some good are some bad.If it benefits other games it will be excellent.

yes, lowering interest is the word. Sometimes cohesion can be problem. I play with high bonus LOS settings for ai. Even with high bonus LOS Russian cossacks marched to prag then decided to support siege in Stettin without resting.
I try to send AI logs in some situations, thanks