Information about developments at snowballing problem

Currently viewing this thread:

Dabos37

Knight at Arms
Best answers
0
Great analysis as always!

Aside from the great points you have mentioned, the thing is that I find Sturgia mostly week due to some other factions overperforming (Battania and Khuzait). As I have mentioned before, I am disabling the cavalry bonus in simulated battles in all my new campaigns and Sturgia is doing ok in the first 5-10 years (not sure what happen after that because I just start a new campaign).

In all my new campaigns, Sturgia is able to defend from Vlandia without much problems, and they just start getting wrecked if facing Battania and/or Khuzait (like any other faction).

The fact of Battania being able to wreck Vlandia in some campaigns is something which has surprised me a lot if we think about Vlandia amount of fiefs and clans.

Talking about the map redesign or adding some few villages for Sturgia, I am not sure if this is something which could be do easily and it probably requires an insane amount of work (maybe I am wrong here), and not sure if it is something which devs could see as an option, but this would probably help Sturgia a lot.
 

five bucks

Regular
Best answers
0
@mexxico I've been thinking about why Sturgia is one of the normal losers and something that stood out to me was that most of the sturgian towns only have 2 villages. This puts them at a massive disadvantage for several reasons especially considering their heavy geographic handicap. Compared to kingdoms like Battania (one of two most common winners)
Great writeup! My two cents: Considering that in the setting, Battania is meant to be an underdog like Sturgia is now, perhaps Battania could lose some villages while Sturgia gains some villages? In other words, nerf Battania at the same time as buffing Sturgia?
 

Terco_Viejo

Spanish Gifquisition
Master Knight
Best answers
0
[...]

Talking about the map redesign or adding some few villages for Sturgia, I am not sure if this is something which could be do easily and it probably requires an insane amount of work (maybe I am wrong here), and not sure if it is something which devs could see as an option, but this would probably help Sturgia a lot.
I believe that one of the problems with Sturgia is that the expansion/invasion points are too conditioned by agressive bottleneck narrowing passages. In this thread, I proposed a series of modifications to make the map a little closer to the warband heritage.

Original NativeModified
 

Itisausername

Sergeant Knight at Arms
Best answers
0
So since main PC broke where I had all the progress on my mod I really should have used a USB, I did tests through the versions to see what happened wince if you remember the first patches it was always an empire faction that would snowball I had to run all these tests on my ****ty laptop btw and the crashes I faced fast forwarding were enough to give me cancer anyway lets get to the real deal:

1.0: This was when the first of the snowballing began this time by the empire or battania never Khuzait actually so I wanted to see what caused the problem so after the fast forward N.E had one town left to take so I checked through everything and I find that every defeated kingdom has 1 clan left being the ruler one not to surprsing aye.

Test 2 1.0: This was a 5 year test with no fast forward and the main thing I noticed is that people started to defect as soon as the first town was lost and always directly to the winning faction instead of spreading out to multiple kingdoms which left factions with around 3 to 4 clans while still controlling 5 castles and 3 towns and this tame it was S.E who sonowballed

Now this is where I started making my hypothesis on the older versions it seened that clans would defect to one kingdom immediately unlike now where they can keep 3-4 clans with one city so I started thinking now that the Khuzait have time to not defect since it wasn't just the losing faction it was also neutral factions but as soon as the faction who got empowered declared war they're entire country defected so now that they don't defect as much they keep they're original clans with a bunch of cav who had the bonus.

I skipped all patches leading up 1.2.0 since nothing crazy changed

1.2: This is when I first began to notice the Kuzait snowball and they rolled through everybody except battania and vlandia now my original thought was of they were too far away for 20 years but vlandia also has a lot of cav and looking through battanian parties the average was 30 cav hmmm

1.3 not much changed from snowballing in older versions.

1.4: This is where Vlandia started snowballing it seems as on my 3 tests on this version they never once lost they would be stuck in an endless war against the Khuzait and after fast forwarding 20 years more they were still in eternal war.

1.5 not much change

1.5.1: Big patch for snowballing it made it slower but it was not made close to obselete

1.5.2 there was not much difference execpt nicer looking sheep

1.5.3: This is wherenI noticed a change in the tests Battania was not just holding its own it was kicking ass and this is where vlandian snowballing also stopped.

1.5.4: Much of the same
1.5.5: Here we are today with it being somewhat improved but again not close to obselete

This is basically just a bunch of useless data so let me explain what I saw as the main difference so big one between the first few like 1.0 to 1.2 was cland defectionthrough 1.3 to 1.5 it was mostly noticed in how armies and individual lords behaved and 1.5.1 to 1.5.5 main difference was that snowballing would happen a bit slower.

Main problem Khuzait culture bonus

I did a few other tests in versions being much shorter but no fast forward if I saw anything interesting.

Again as has been mentioned by @Blood Gryphon before we see in your tests that battania makes a comeback but its lost no clans and from what I saw from these tests

Losing clans= You die
 
Last edited:

Worlok

Sergeant at Arms
Best answers
0
Here are those settings to run your own test mexxico mentioned
I am trying to run this test on my new campaign 1.5.5... Battania and Sturgia inevitably eats up Vlandia and Khuzait snowballing (I know it's still expected in 1.5.5).
However there is a thing I miss a lot in the game. Heroic AI vs AI battles. All these battles are simply mathematical calculation. Bigger number wins all the time.
Maybe TW could introduce some hero spawns with high tactical skills. This could lead to several heroic battles where the weaker faction randomly defeats the strong which could trigger comebacks.
 

Apocal

Sergeant Knight at Arms
Best answers
1
1.2: This is when I first began to notice the Kuzait snowball and they rolled through everybody except battania and vlandia now my original thought was of they were too far away for 20 years but vlandia also has a lot of cav and looking through battanian parties the average was 30 cav hmmm
...
Main problem Khuzait culture bonus
The Khuzait bonus wasn't working at all until 1.3 or thereabouts, then got dialed back around 1.4.3. I'm not sure on which specific patches because they were both unannounced.
 

Itisausername

Sergeant Knight at Arms
Best answers
0
The Khuzait bonus wasn't working at all until 1.3 or thereabouts, then got dialed back around 1.4.3. I'm not sure on which specific patches because they were both unannounced.
Cool cool gonna read all the patch notes now could be different these are just my tests thanks for the info
 

sniparsexe

Squire
Best answers
0
@mexxico Here's a 1.5.4 20 year test I ran last week.
FactionsFiefs
Aserai11
Vlandia48
Sturgia1
Battania0
Khuzait39
N.E0
S.E1
W.E20


1.5.5 20 year tests I ran today

Test 1)
FactionsFiefs
Aserai14
Vlandia8
Sturgia3
Battania23
Khuzait31
N.E1
S.E2
W.E37


Test 2)
FactionsFiefs
Aserai12
Vlandia18
Sturgia7
Battania29
Khuzait37
N.E0
S.E14
W.E3
 

sniparsexe

Squire
Best answers
0
@Blood Gryphon you told me to use campaign."multiply_campaign_speed 10" but it was really slow (Idk if it's a bug or not but in 2 hours only 2 years have passed) so I used "multiply_campaign_speed 50" I hope it doesn't effect anything lol

(I can't edit my previous message to add this if any mod can merge these 2 comments please merge them)
 

mexxico

Developer
Best answers
0
@mexxico Here's a 1.5.4 20 year test I ran last week.
FactionsFiefs
Aserai11
Vlandia48
Sturgia1
Battania0
Khuzait39
N.E0
S.E1
W.E20


1.5.5 20 year tests I ran today

Test 1)
FactionsFiefs
Aserai14
Vlandia8
Sturgia3
Battania23
Khuzait31
N.E1
S.E2
W.E37


Test 2)
FactionsFiefs
Aserai12
Vlandia18
Sturgia7
Battania29
Khuzait37
N.E0
S.E14
W.E3
Thank you. According to your tests snowballing score is :

9+13+19+20+4+20+19=104 for 1.5.4
6+12+17+19+18+2=74 for 1.5.5 first test
8+2+13+2+20+6+17=68 for 1.5.5 second test

Also we have two tests @Blood Gryphon did for 1.5.5 and he found snowball score 135 and 140.

At initial post I gave results of 2 tests from @Blood Gryphon and @Apocal their results were 116 and 129 for 1.5.4 20 year tests.

So it seems average snowball score average is 104 for 1.5.5 (4 samples) and 116 for 1.5.4 (3 samples) We will see what this value dropped to at 1.5.6

For ones who do not know how to calculate snowball score is :
we run game for 20 years without player interaction. Then for each faction if fief count (towns are counted 2 and castles are counted 1) is out of 20-35 values we add distance to 20-35.

Lower scores are better.
 
Last edited:

Apocal

Sergeant Knight at Arms
Best answers
1
@Blood Gryphon you told me to use campaign."multiply_campaign_speed 10" but it was really slow (Idk if it's a bug or not but in 2 hours only 2 years have passed)
If you're playing on a slow laptop years can take that long, yes.

so I used "multiply_campaign_speed 50" I hope it doesn't effect anything lol
It does effect some things but that doesn't mean your results are invalid.
 

sniparsexe

Squire
Best answers
0
If you're playing on a slow laptop years can take that long, yes.



It does effect some things but that doesn't mean your results are invalid.
I actually have a decent computer and it wasn't lagging, I honestly don't know why it was so slow maybe a bug or I wrote 1 instead of 10 don't know

Thank you. According to your tests snowballing score is :

9+13+19+20+4+20+19=104 for 1.5.4
6+12+17+19+18+2=74 for 1.5.5 first test
8+2+13+2+20+6+17=68 for 1.5.5 second test

Also we have two tests @Blood Gryphon did for 1.5.5 and he found snowball score 135 and 140.

At initial post I gave results of 2 tests from @Blood Gryphon and @Apocal their results were 116 and 129 for 1.5.4 20 year tests.

So it seems average snowball score average is 104 for 1.5.5 (4 samples) and 116 for 1.5.4 (3 samples) We will see what this value dropped to at 1.5.6

For ones who do not know how to calculate snowball score is :
we run game for 20 years without player interaction. Then for each faction if fief count (towns are counted 2 and castles are counted 1) is out of 20-35 values we add distance to 20-35.

Lower scores are better.
Okay I only wrote total fiefs for these tests, from now on I'll calculate the snowball score first and then let you know.
 

Nalgasucia507

Sergeant
Best answers
0
I believe that one of the problems with Sturgia is that the expansion/invasion points are too conditioned by agressive bottleneck narrowing passages. In this thread, I proposed a series of modifications to make the map a little closer to the warband heritage.

Original NativeModified
+1. TW, please consider going over the map before EA finishes.
 

Blood Gryphon

Sergeant Knight
WBVC
Best answers
1
Great analysis as always!

Aside from the great points you have mentioned, the thing is that I find Sturgia mostly week due to some other factions overperforming (Battania and Khuzait). As I have mentioned before, I am disabling the cavalry bonus in simulated battles in all my new campaigns and Sturgia is doing ok in the first 5-10 years (not sure what happen after that because I just start a new campaign).

In all my new campaigns, Sturgia is able to defend from Vlandia without much problems, and they just start getting wrecked if facing Battania and/or Khuzait (like any other faction).

The fact of Battania being able to wreck Vlandia in some campaigns is something which has surprised me a lot if we think about Vlandia amount of fiefs and clans.

Talking about the map redesign or adding some few villages for Sturgia, I am not sure if this is something which could be do easily and it probably requires an insane amount of work (maybe I am wrong here), and not sure if it is something which devs could see as an option, but this would probably help Sturgia a lot.
You bring up some great points. Cav bonus gives too much power to cav dominate kingdoms.

I think Battania is an outlier to most of the general reasons Kingdoms are strong because they are so condensed that their defensive response time is fast and their ability to recover from loss is fast (its not that they don't lose fights). This is why I think Sturgia is at such a disadvantage, all of their settlements are so disperse that they take forever to respond to defensive needs and can't build up forces fast enough after a defeat, which is often due to lack of inf advantage in autocalc (having some of the best inf in battle doesn't mean anything in autocalc).

Vlandia suffers from not having a giant lake and having the Battanians as neighbors, although are better off than Sturgia due to their high town/village numbers. Aserai also shows us a long kingdom that does fine because they have the highest amount of towns and villages and also have strategic cutoff points at the edge of their kingdom instead of the middle of it.

I truly hope they considering adding new villages/castles as I don't think mexxico can do anything with the map, it is not his decision (just like cav bonus and Khuzait faction traits).

Great writeup! My two cents: Considering that in the setting, Battania is meant to be an underdog like Sturgia is now, perhaps Battania could lose some villages while Sturgia gains some villages? In other words, nerf Battania at the same time as buffing Sturgia?
You would think that right? Battania is supposedly one of the weaker factions that dont survive to Warband. Khuzaits strength seems acceptable in 1.5.6, I mean if we go by Warband lore they defeat the empire in a couple hundred years (not 20), but not the Sturgians (who split into nords/Vaegir).

I believe that one of the problems with Sturgia is that the expansion/invasion points are too conditioned by agressive bottleneck narrowing passages. In this thread, I proposed a series of modifications to make the map a little closer to the warband heritage.

Original NativeModified
Terco will you promise to make a mod with all your suggestions if they aren't taken in by the end of EA? So many of your suggestions are spot on. Sturgia looks a lot more open like Aserai instead of bottlenecked in the middle of their own territory.

I am trying to run this test on my new campaign 1.5.5... Battania and Sturgia inevitably eats up Vlandia and Khuzait snowballing (I know it's still expected in 1.5.5).
However there is a thing I miss a lot in the game. Heroic AI vs AI battles. All these battles are simply mathematical calculation. Bigger number wins all the time.
Maybe TW could introduce some hero spawns with high tactical skills. This could lead to several heroic battles where the weaker faction randomly defeats the strong which could trigger comebacks.
This is a great point and something I've suggested before is letting weaker kingdom AI focus on training up their troops by going into manhunter/patrol/recruit mode until their strength level is decent. Something that is missing from Warband is a kingdom focus on gathering their strength again, instead in 1.5.5 they form armies once they get enough influence and other lords have 40% of their party with just recruits. But this is one of the fixes mexxico has made in 1.5.6 with ensuring parties actually fill up their parties.

Something that @Apocal has suggested which is think is great is making lords have a higher focus on retreating once a portion of their troops are dead, I think that would only be possible with a change to armor/health/dmg resistance, like @Terco_Viejo has tested, to let the battles last long enough for the AI to retreat (right now battles are over so fast their is no chance to strategically retreat for AI). I think in autocalc the AI do a decent job of retreating when severally outnumbered, but it wont happen when the player is involved.

snip- hitting word count limit

Again as has been mentioned by @Blood Gryphon before we see in your tests that battania makes a comeback but its lost no clans and from what I saw from these tests

Losing clans= Uou die
Good notes. Sorry about your pc that sucks.

I really do think the clans deserting thing is the major issue in kingdoms dying. I think when they lose a clan they need to focus on getting them or another clan back at a high priority, like increase the kingdom wallet tax rate to gather funds for Clan recruitment. We can even see in mexxicos 1.5.6 test at year 30 the sturgians start to die off once the Kuloving clan desert over to the Western empire handing over Varcheg (or being giving Varcheg after losing it).

I also think maybe they should try out a version where deserting because you are surrounded is turned off, that seems to be a big factor for Sturgians deserting once they lose one or two of their settlements (once again due to how dispersed all of their settlements are). Its honestly almost like a penalty to specifically screw over the Sturgians because of their geography.


I actually have a decent computer and it wasn't lagging, I honestly don't know why it was so slow maybe a bug or I wrote 1 instead of 10 don't know
I have a feeling it was this, the typical fast forward button speed is 3 so you likely went below that. mexxico does his test at 10 speed, apparently above 20 has shown some issues in his tests.
 
Last edited:

Flesson19

Sergeant
Best answers
0
Snowball has no relation with open-close at same moment voting bug, this voting bug only effects player other voting bug (if player is in kingdom no new peace is happening) is fixed. And I am not dealing with that open-close at same moment voting bug this is another dev’s area. I sent your posts link to him.

Snowball fixes will be at 1.5.6 for comparision its good different players make their own 20 year test at 1.5.5 and post results here.
ok but the lords running out of money is happening with no players in their kingdom, did you address that as well?
when I joined S Empire they had 7 lords in Vostrum sitting in the keep broke and it seems most games that is what happens that the S empire seems to only have Vostrum left so I am assuming most games that is happening.
 
Last edited:

Terco_Viejo

Spanish Gifquisition
Master Knight
Best answers
0
@mexxico One question i would like to ask you...Would it be possible to bring back Zendar as an independent (non-conquerable) settlement and have it develop commercial activities? Could this be applicable to the simulation of the current economic system?
 

Blood Gryphon

Sergeant Knight
WBVC
Best answers
1
Created a kingdom budget and any clan having 100K+ money start to give 0.5% of their (money - 100K) to kingdom wallet daily. For example if a clan leader has 200K he paid 500 denars daily to kingdom budget (200K - 100K) x 0.5 = 500. This collected money is used to help poor clans in same kingdom if any exists. Each day 500 gold money is given to any clan having less than 30K money, 1000 gold is given to any clan having less than 15K gold.
Few questions on the kingdom budget.
  • Are the 500/1000 donations to poor clans hard capped or do they scale with the total amount of the kingdom budget? (ex. if there is a clan with 1 million they would pay 4.5k to kingdom budget, but if there is only 1 poor clan do they only get 1000 or all 4.5k from the millionaire clan?
  • If they only get 1000, does the kingdom start saving for times when maybe they are all poor?
  • Maybe a kingdom budget savings could be used for other features once you are king? maybe like recruiting other clans?
  • What happens when all clans are poor?
  • You should encourage people who make kingdom policies to have a policy that increases the rate (but never gets rid of this base 0.5%), kingdoms could activate this policy when they have a majority of poor clans (once you/someone implements dynamic kingdom policies).
 

hieronymos2

Recruit
Best answers
0
Couldn't snowballing just be seriously slowed or even stopped by enacting a serious autocalc siege attack casualty debuff? With also changing AI to postpone actual assault until loss ratio is more favorable?

I mean isn't stopping loss/gain of real estate the main driver of snowballing?

Regarding defections, isn't there a cultural bias that influences clans defecting to other non-culture nations? Couldn't that be increased to lower defection % ?
 
Last edited:

Apocal

Sergeant Knight at Arms
Best answers
1
Couldn't snowballing just be seriously slowed or even stopped by enacting a serious autocalc siege attack casualty debuff? With also changing AI to postpone actual assault until loss ratio is more favorable?

I mean isn't stopping loss/gain of real estate the main driver of snowballing?
This came up really early in EA:
Making stable world is not hard. We just close defections, decrease sieges, make garrisons huge and give ai extra money to feed his garrison then its easy to make game stable. However it is not entertaining and it is not player-like ai playing. In every game you should see map is dynamic and take you in different direction. Of course one faction controlling all map is not something we also want to see. However we do not want to see stable world also.