Information about developments at snowballing problem

[Blood Gryphon]

This is something I have been always curious about... I usually use x20 time speed multiply but it looks like not much changed in your test.

Same, but yeah it doesn't seem to have impacted overall results considering i landed right in the middle of the general snowball scores (thankfully, not sure my computer was too happy about running bannerlord for 10 hours).

It is weird because there are some great campaigns where snowballing is really low, while other campaigns where snowballing is much worse than ever, strange...

I know that I sound like a broken record but I really do think it has to do with clan defection. Once it starts the snowballing cant be stopped, the factions who recruit the most clans are always the snowballed factions (obviously lol).

Spoiler: Test 1 with clans

1.57​	Test 1 10 years	Test 1 20 years	20 yr Strength	Original Clans	Recruited Clans	Rebel Clans	Mercenaries	Total Noble Clans	Total Clans w/ MERCs
Aserai	20​	23​	7171​	9​	2​	0​	0​	11​	11​
Battania	33​	48​	8891​	7​	2​	0​	0​	9​	9​
Khuzait	34​	38​	4993​	9​	0​	1​	0​	10​	10​
Northern	26​	13​	5011​	9​	0​	1​	1​	10​	11​
Southern	15​	13​	4029​	9​	0​	0​	1​	9​	10​
Western	10​	7​	2849​	7​	0​	1​	2​	8​	10​
Sturgia	21​	20​	4894​	9​	0​	0​	0​	9​	9​
Vlandia	14​	11​	4116​	9​	0​	0​	0​	9​	9​
Snowball score	21​	47​

Spoiler: Test 2 with clans

1.57​	Test 2 20 yrs	20 yr Strength	Original Clans	Recruited Clans	Rebel Clans	Mercenaries	Total Noble Clans	Total Clans w/ MERCs
Aserai	23​	7044​	9​	3​	0​	1​	12​	13​
Battania	46​	10012​	7​	4​	0​	0​	11​	11​
Khuzait	44​	10046​	9​	5​	0​	0​	14​	14​
Northern	7​	3887​	6​	1​	0​	0​	7​	7​
Southern	12​	3406​	7​	1​	4​	3​	12​	15​
Western	23​	6077​	9​	1​	0​	1​	10​	11​
Sturgia	19​	7046​	9​	0​	0​	0​	9​	9​
Vlandia	0​	604​	1​	0​	0​	5​	1​	6​
Snowball score	62​

Look at the clan counts in these two tests its pretty clear to me that when clans are balanced snowballing is in an acceptable range (test 1), but once defections start to take place the snowballing begins (test 2). Would be great if others could start counting clans.

@mexxico it does seem your changes to make rebel clans not join the strongest worked well (we see no rebel clans in the snowballed factions). Of course, we see that recruited noble clans is the opposite and they join the snowballed faction. Also seems like the mercenary debt issue is what truly killed Vlandia in Test 2 (I know 1.5.8 will fix).

 

[Apocal]

@Apocal 's tests have some pretty bad snowball scores; 89, 116, and 67. Can you upload the what the maps looked like?

Sure thing.

Spoiler: Test 1

Spoiler: Test 2

Spoiler: Test 3

 

[mexxico]

@mexxico it does seem your changes to make rebel clans not join the strongest worked well (we see no rebel clans in the snowballed factions). Of course, we see that recruited noble clans is the opposite and they join the snowballed faction. Also seems like the mercenary debt issue is what truly killed Vlandia in Test 2 (I know 1.5.8 will fix).

Yeah that mercenary debt issue will be fixed with 1.5.8 and it is something making weakened factions poor also then clans with no money leave their kingdom. This can be related. I was thinking this too.

More samples are always welcomed by the way. Thanks for all test runs.

 

[Blood Gryphon]

Yeah that mercenary debt issue will be fixed with 1.5.8 and it is something making weakened factions poor also then clans with no money leave their kingdom. This can be related. I was thinking this too.

More samples are always welcomed by the way. Thanks for all test runs.

No way to push that change into a hotfix right? (I ask too much I know )

I want to see more data on clans but I already feel like that is probably the straw that breaks Humphreys back and activates snowballing. You made it obvious once you reminded me that poor clans defect easily.

With that fix, I think we are almost there

 

[Apocal]

More samples are always welcomed by the way. Thanks for all test runs.

I'll run some more tests tomorrow. It is possible that we are just unlucky.

I think we are almost there

Well, right now, a lot of the tools we used to curb snowballing are actually "nerf Khuzaits, specifically" instead of robust balancing systems. And even now what we have is fairly precarious. That means any mods that tweak numbers, add new features or make major changes are likely to have their own issues with snowballing.

 

[Blood Gryphon]

I'll run some more tests tomorrow. It is possible that we are just unlucky.



Well, right now, a lot of the tools we used to curb snowballing are actually "nerf Khuzaits, specifically" instead of robust balancing systems. That means any mods that adjust that make major changes are likely to have their own issues with snowballing.

Honestly, I think if we can get clans to rarely defect which seems to be the case until a clan goes into permanent mercenary debt, snowballing is not a big factor. The luck part right now comes with if a weak kingdom will let itself go into a irrecoverable debt, which is a bug that mex has already found and fixed for the next version. If they don't go into debt they keep their clans and we see minor comebacks.

You are right though, the next step for a robust balancing system is real diplomacy. We don't see powers shift at all, no major comebacks or falls from the top. Only way to get that is through player intervention right now.

In truth have we really nerfed specifically Khuzait yet? There is no change to their faction bonus (yet) and the cav speed/party changes impacted everyone (although clearly hitting khuzait the most, but we know that was broken). Also with mods hasn't that always been the case?(I know bs cop out answer )

 

[Apocal]

In truth have we really nerfed specifically Khuzait yet? There is no change to their faction bonus (yet) and the cav speed/party changes impacted everyone (although clearly hitting khuzait the most, but we know that was broken).

The culture bonus was effectively nerfed around 1.5/refactor; it was originally implemented (not intended, I'm sure) as a flat 10% party speed bonus, all conditions. The Khuzaits also received at least two rounds of economy nerfs and their starting war opponents were given 'free' boosts to garrisons and fortifications.

Also with mods hasn't that always been the case?(I know bs cop out answer )

Yeah, some Warband mods had broken balance. That's why I think there should be a fairly large toolkit for addressing it.

 

[lunelune2413]

Sure thing.

Spoiler: Test 1

Spoiler: Test 2

Spoiler: Test 3

That Batannia. Damn...
I thought that longer wars was a sacrifice to make and all the other improvements would help. But it would appear that those longer wars just allow the AI to expand much more and thus expand way faster.

The first 10 years everything seems fair and balanced. Then 10 years later theres 2-3 "dead" factions. Even more frequent that what 1.5.6 seems to have.
Mexxico, we wish you good luck in figuring out what's directly causing all these snowballs to form.

 

[Blood Gryphon]

The culture bonus was effectively nerfed around 1.5/refactor; it was originally implemented (not intended, I'm sure) as a flat 10% party speed bonus, all conditions. The Khuzaits also received at least two rounds of economy nerfs and their starting war opponents were given 'free' boosts to garrisons and fortifications.

Ah that's right, thanks for reminding me (hard to keep all the changes straight).

Yeah, some Warband mods had broken balance. That's why I think there should be a fairly large toolkit for addressing it.

I bet mex would love that toolkit too

Also I'm sure this isn't the case but I gotta ask considering you had the most drastic results by far. You for sure you hadn't let any of your modifications in? I know you were trying out some different percentages for certain aspects in the previous patch (I might be getting you mixed up with someone else tho, sorry).

The first 10 years everything seems fair and balanced. Then 10 years later theres 2-3 "dead" factions. Even more frequent that what 1.5.6 seems to have.
Mexxico, we wish you good luck in figuring out what's directly causing all these snowballs to form.

I felt the same about my first test, first 10 years was pretty even but then by the next 10 the snowballing had started. This is actually a massive improvement compared to previous versions where the snowballing would happen in the first 10 years.

 

[Bummerlord]

Here is my 20-year test. Not sure how to calculate the snowball score. Hit me up if more info is needed.

Factions	Cities	Castles
Aserai	8	11
Battania	3	0
Khuzait	15	22
NE	1	2
SE	3	4
Sturgia	9	11
Vlandia	12	16
WE	2	1

First time I saw Sturgia expand. They even managed to take back Tyal after a brief occupation by the Khuzait. Both them and Vlandia were giving Battania a very hard time. Apart from that, it was a slow but inexorable Khuzait expansion.

 

[Sarissofoi]

I noticed that in Vlandia vs Battania the one that get ganged ion 1v2 or 1v3 war get shafted and the other win hard.

 

[Apocal]

Also I'm sure this isn't the case but I gotta ask considering you had the most drastic results by far. You for sure you hadn't let any of your modifications in?

I always do a fresh install before I test, to prevent exactly that.

 

[Ferisko]

@Blood Gryphon Today I made a test that I'll update later (didn't have time) and I had similar results that @Apocal

 

[Blood Gryphon]

I always do a fresh install before I test, to prevent exactly that.

I figured, but that 116 score had me like

Also thanks for the maps!

@Blood Gryphon Today I made a test that I'll update later (didn't have time) and I had similar results that @Apocal

Good to know its not an outlier

 

[Pejot]

Bigger cav ratio plus no need for horses for ai cheat does the thing. Winning side is getting a hell lot of a cav. Battanian are wrecking Vlandians (Battanian cav is for some reason stronger than Vlandian). Since they only need money to upgrade winners get better troops outing looses in a loop of loses.

 

[Dabos37]

Yes, judging by some people's tests, it looks like snowballing got actually worse, and I feel somehow gilty because maybe some of my suggestions have made it worse :'-(.

On the other hand, 10-20 days war lenght average is too short and it is good that it has got increased.

 

[Blood Gryphon]

Yes, judging by some people's tests, it looks like snowballing got actually worse, and I feel somehow gilty because maybe some of my suggestions have made it worse :'-(.

On the other hand, 10-20 days war lenght average is too short and it is good that it has got increased.

If anything 1.5.7 seems to be the most variable, we have also seen the lowest scores. Its a good indication that we are almost there, im not too sure the long wars is such an issue. We'll know for sure once we get the merc debt issue fixed (my bet is this).

 

[Sarissofoi]

If anyone interested I could test some games with Party templates changed(reduced in size).
Would this be helpful?

 

[Dabos37]

If anyone interested I could test some games with Party templates changed(reduced in size).
Would this be helpful?

Not sure, maybe could slow it down a bit but who knows... I am really interested to test the result of defeated armies trying to fleeing while getting disbanded automatically after trying to flee. This would drastically increase the parties chance to escape. If someone knows how to mod this, please let me know it.

 

[Sarissofoi]

Getting extra speed buff on defeat would be nice.
But consider that it may actually weaken defeat faction as they will not get free troops from respawn.