Information about developments at snowballing problem

Users who are viewing this thread

So to my original question:- Mechanically, how does AI factions recruiting clans work? Do they have to pay just as much as the player does? Does their king/leader have to meet them themselves just like the player does? or is it a bit "cheaty" for the AI?
AFAIK, they do not need to converse to recruit; take this with a grain of salt as my knowledge may be outdated
From my limited understanding the AI uses a lot of cheats. Just from reading around on the forums I see stories of AI lords being released from captivity, walking into the nearest village and recruiting 80+ troops. Just yesterday I watched a guy on Twitch besiege a Town and an enemy lord had just got there before he did. The AI lord recruited over 40 troops while the siege camp was being constructed. So "a bit cheaty" might be an understatement lol
I doubt a lord recruited 80+ troops from a village. Do you have a quote on this (I am not doubting you, I am doubting your source)? AI lords "cheat" by having an extra recruitment slot but they are still bound by relations with the notables.
In the second case, when the AI lord recruited the troops did the garrison numbers go down? The AI could very well recruit from the notables and then drain the garrison. The latter is usually why you have constant armies attacking you, and lords that you captured->released can come back with big numbers.
 
I do think lords take from the garrison so after they are released they can get troops fast but you have that option also, except TW forces us to try and get out of enemy territory to get back to our fiefs after being released. But as for the siege I am sure while the siege is being setup they are also recruiting just being in town like you can or parties in your army, just go to town with your army and wait you will see them auto recruit
 
From my limited understanding the AI uses a lot of cheats. Just from reading around on the forums I see stories of AI lords being released from captivity, walking into the nearest village and recruiting 80+ troops. Just yesterday I watched a guy on Twitch besiege a Town and an enemy lord had just got there before he did. The AI lord recruited over 40 troops while the siege camp was being constructed. So "a bit cheaty" might be an understatement lol
Let's avoid spreading false information and base our judgement on reliable information.

[EDIT] broken link fixed
 
Last edited:
I doubt a lord recruited 80+ troops from a village. Do you have a quote on this (I am not doubting you, I am doubting your source)? AI lords "cheat" by having an extra recruitment slot but they are still bound by relations with the notables.

Turns out it was a Town, my mistake. But during the siege I didn't notice any change to the garrison numbers, just the Lords troop numbers increasing
 
Sounds like they fixed the main issue of AI suicidal armies. I think the suicidal nature of offensive armies had a big part to play in stopping snowballing as its been very hard to gain ground for factions. Still gotta test it but if true this is a major step for the campaign. My theory before any testing is that these changes will likely increase the snowball scores (hopefully not to the levels of it being an issue again) from the very low and stagnant scores we've been seeing. I'm gonna run some snowball tests to see if these changes have any impact on the world state.

You'll all see updates later today in the typical snowball thread. I encourage others to run some of their own.
Well so far my theory is not playing out. The first 20 year test had very little to no town conversions. Sieges did not seem to last long but im not sure if the siegers themselves left or if they got chased away from a defending army.

Here are the results of my first test of 1.7.2. With a snowball score of 6 the map is essentially stagnant and looks the same as it did at the start of the game. Most territory taken is border castles that just flip back and forth.

pnhBJ.jpg

Test 1 1.7.2Year 20Strength
Aserai
25​
10453​
Battania
14​
3535​
Khuzait
20​
6973​
Northern
22​
9287​
Southern
20​
8827​
Western
24​
7927​
Sturgia
24​
9217​
Vlandia
24​
9325​
Snowball score
6​


Definitely need a few more tests before coming to any conclusions as this test could just be an outlier, but before i do another im going to take a closer look at AI behavior. I would suggest others start to watch AI armies and see if the recent changes improved things so much that the AI are too good at defending or if the AI is so much worse that they cant successfully prepare for a siege.
 
Ok so here are the results for two 20 year snowball tests for patch 1.8.0. This patch has definitely shook things up and the map is much more spicy now.

Here are the results of Test 1.

5xpUM.jpg
RXM4k.jpg
tXAE0.jpg
Z9l3g.jpg
LSDPS.png
_aJAT.png
HlT8v.png

Here are the results of Test 2.

-kcEj.jpg
2lsQM.jpg
ek4BG.jpg
6oi4u.png
TflGK.png
IUmfO.png

Here are observations I've made while throwing all the data together for both tests @SadShogun .
  • 20 year snowball scores have jumped up to 80 and 71, numbers we haven't seen since 1.5.6 (generally have averaged about 30-40 the past several patches). The map is definitely not stable anymore after 20 years, but we do see a relatively calm early game with 5 year snowball scores only being 13 and 12. We saw different mid games with test one having a snowball score of 41 after 10 years and test two having a score of 19. So even though test two hadn't escalated as fast as test one in 10 years they both end up relatively high by 20 years.
  • My guess on why snowballing is much higher now is something to do with defection. If you look at the recruited clans column in the clan status table you'll see we have 19 and 22 clans that have flip kingdoms, in previous patches defections would only happen if the kingdom was losing territory and going broke, but ive noticed even the strong kingdoms are having clans defect. If you compare starting clans with original clans(@20 years) you'll notice every kingdom has had at least 1 clan defect.
-RwsZ.jpg
EkLzN.jpg
  • Wealth is still a mirror of the state of the world, with more clear winners there are a lot of rich clans, but with more losers there are also lots of poor clans. Good to see there are barely any very poor clans as that's normally an indication that clans are too poor to build full armies.
  • I noticed there seems to be about a 50% chance that a rebel group that becomes an official clan will get recruited by a kingdom. Most of them join the kingdom that is the same as their culture which is a good sign. The ones who dont get recruited seem to get themselves into wars with everyone in the world leaving them with no one to join.
RlwbY.jpg
DfVQJ.jpg
SD1cA.jpg
V5IGe.jpg
D3Mvr.jpg
  • There seems to be an issue with marriage, i wasnt seeing many if any at all and there didnt seem to be any young children left by 20 years.
KU731.jpg
Htltv.jpg
KZ5iN.jpg
QMhOg.jpg
9Td6m.jpg
2AjVr.jpg
cTUfv.jpg
bpsTh.jpg
66Th5.jpg
2v6tp.jpg

Only two tests so take everything with a grain of salt.
 
Last edited:
Ok so here are the results for two 20 year snowball tests for patch 1.8.0. This patch has definitely shook things up and the map is much more spicy now.

Here are the results of Test 1.

5xpUM.jpg
RXM4k.jpg
tXAE0.jpg
Z9l3g.jpg
LSDPS.png
_aJAT.png
HlT8v.png

Here are the results of Test 2.

-kcEj.jpg
2lsQM.jpg
ek4BG.jpg
6oi4u.png
TflGK.png
IUmfO.png

Here are observations I've made while throwing all the data together for both tests @SadShogun .
  • 20 year snowball scores have jumped up to 80 and 71, numbers we haven't seen since 1.5.6 (generally have averaged about 30-40 the past several patches). The map is definitely not stable anymore after 20 years, but we do see a relatively calm early game with 5 year snowball scores only being 13 and 12. We saw different mid games with test one having a snowball score of 41 after 10 years and test two having a score of 19. So even though test two hadn't escalated as fast as test one in 10 years they both end up relatively high by 20 years.
  • My guess on why snowballing is much higher now is something to do with defection. If you look at the recruited clans column in the clan status table you'll see we have 19 and 22 clans that have flip kingdoms, in previous patches defections would only happen if the kingdom was losing territory and going broke, but ive noticed even the strong kingdoms are having clans defect. If you compare starting clans with original clans(@20 years) you'll notice every kingdom has had at least 1 clan defect.
-RwsZ.jpg
EkLzN.jpg
  • Wealth is still a mirror of the state of the world, with more clear winners there are a lot of rich clans, but with more losers there are also lots of poor clans. Good to see there are barely any very poor clans as that's normally an indication that clans are too poor to build full armies.
  • I noticed there seems to be about a 50% chance that a rebel group that becomes an official clan will get recruited by a kingdom. Most of them join the kingdom that is the same as their culture which is a good sign. The ones who dont get recruited seem to get themselves into wars with everyone in the world leaving them with no one to join.
RlwbY.jpg
DfVQJ.jpg
SD1cA.jpg
V5IGe.jpg
D3Mvr.jpg

Only two tests so take everything with a grain of salt.
Thanks for the testing; can I ask, do you consider 1.8 good from the perspective of gameplay pacing for the player (ample time to interact with or join kingdoms before they are destroyed, but a decent amount of movement on the world map)?

The way you're talking, it sounds like the early game is where it should be, the mid game might be a bit too much and the late game is good?
 
@Blood Gryphon : i can only surmise that you're serving out a moderate to lengthy prison term somewhere -and the guards have left you unattended in the computer lab.
giphy.gif

im just a data scientist who likes simulation, I also like having longer playthroughs so snowballing is an important issue for me. also might be running these 2 hour tests while i work :lol:
Thanks for the testing; can I ask, do you consider 1.8 good from the perspective of gameplay pacing for the player (ample time to interact with or join kingdoms before they are destroyed, but a decent amount of movement on the world map)?

The way you're talking, it sounds like the early game is where it should be, the mid game might be a bit too much and the late game is good?
Yeah early game is where it should be but its unlikely to ever change as its mostly held up by the fact all clans start out with average wealth, the world only starts to shift once clans start to become poor. Test 1 mid game happened to fast for sure, i would expect that score around 20 years, but test 2's mid game was pretty good, so cant definitely say either way for mid game. I think late game is probably happening because of a bug with defection, personally I think it might be a bit much for 3-4 kingdoms to be dead by 20 years, id like to see it be more like 2 dead kingdoms which is more around the 40-50 snowball range. There seems to be a bias against empire and sturgia dying right now, Battania seems to have gotten out of its funk (was dying all the time in 1.7.2) but its only 2 tests so not enough to say that with confidence.
 
Last edited:
Only two tests so take everything with a grain of salt.
This is good, right?

The earlier snowballing almost always happened by 1 faction. Now it seems that there's not so much snowballing but rather a dominant faction in the game. And not always the same faction. So each playthrough reacts and responds differently. Which is what you'd want in a new playthrough.
 
This is good, right?

The earlier snowballing almost always happened by 1 faction. Now it seems that there's not so much snowballing but rather a dominant faction in the game. And not always the same faction. So each playthrough reacts and responds differently. Which is what you'd want in a new playthrough.
I don’t think we’ve hit that point where each playthrough can end up vastly different yet. The common theme has been for a long time that either khuzaits, aserai, or vlandia are the dominant factions.

I’ve never seen the Aserai as a losing kingdom and I think that’s a problem.

This is all assumption based on two tests tho so still need to see more.

Also RIP pinned status :dead:
 
Ok so here are the results for two 20 year snowball tests for patch 1.8.0. This patch has definitely shook things up and the map is much more spicy now.

Here are the results of Test 1.

5xpUM.jpg
RXM4k.jpg
tXAE0.jpg
Z9l3g.jpg
LSDPS.png
_aJAT.png
HlT8v.png

Here are the results of Test 2.

-kcEj.jpg
2lsQM.jpg
ek4BG.jpg
6oi4u.png
TflGK.png
IUmfO.png

Here are observations I've made while throwing all the data together for both tests @SadShogun .
  • 20 year snowball scores have jumped up to 80 and 71, numbers we haven't seen since 1.5.6 (generally have averaged about 30-40 the past several patches). The map is definitely not stable anymore after 20 years, but we do see a relatively calm early game with 5 year snowball scores only being 13 and 12. We saw different mid games with test one having a snowball score of 41 after 10 years and test two having a score of 19. So even though test two hadn't escalated as fast as test one in 10 years they both end up relatively high by 20 years.
  • My guess on why snowballing is much higher now is something to do with defection. If you look at the recruited clans column in the clan status table you'll see we have 19 and 22 clans that have flip kingdoms, in previous patches defections would only happen if the kingdom was losing territory and going broke, but ive noticed even the strong kingdoms are having clans defect. If you compare starting clans with original clans(@20 years) you'll notice every kingdom has had at least 1 clan defect.
-RwsZ.jpg
EkLzN.jpg
  • Wealth is still a mirror of the state of the world, with more clear winners there are a lot of rich clans, but with more losers there are also lots of poor clans. Good to see there are barely any very poor clans as that's normally an indication that clans are too poor to build full armies.
  • I noticed there seems to be about a 50% chance that a rebel group that becomes an official clan will get recruited by a kingdom. Most of them join the kingdom that is the same as their culture which is a good sign. The ones who dont get recruited seem to get themselves into wars with everyone in the world leaving them with no one to join.
RlwbY.jpg
DfVQJ.jpg
SD1cA.jpg
V5IGe.jpg
D3Mvr.jpg
  • There seems to be an issue with marriage, i wasnt seeing many if any at all and there didnt seem to be any young children left by 20 years.
KU731.jpg
Htltv.jpg
KZ5iN.jpg
QMhOg.jpg
9Td6m.jpg
2AjVr.jpg
cTUfv.jpg
bpsTh.jpg
66Th5.jpg
2v6tp.jpg

Only two tests so take everything with a grain of salt.
I imagine the higher snowballing score is due to armies cooperating in sieges. In 1.7.2, besiegers don't have enough to take towns most of the time, and will get torn apart by reinforcing armies at a certain threshold.
 
I imagine the higher snowballing score is due to armies cooperating in sieges. In 1.7.2, besiegers don't have enough to take towns most of the time, and will get torn apart by reinforcing armies at a certain threshold.
I haven't made this assumption yet because armies also cooperate in defending sieges as well now. This is why you don't see much happen in the first 5 years.

The biggest anomaly so far in 1.8.0 compared to other versions is the high defection rate. The weirdest part is clans in kingdoms that are winning are defecting as well, which makes me think there is a bug with it. Also I can tell you that defections are the fastest way to snowballing, one of the big reasons snowballing came down in the past is that defections essentially stopped.
 
I haven't made this assumption yet because armies also cooperate in defending sieges as well now. This is why you don't see much happen in the first 5 years.

The biggest anomaly so far in 1.8.0 compared to other versions is the high defection rate. The weirdest part is clans in kingdoms that are winning are defecting as well, which makes me think there is a bug with it. Also I can tell you that defections are the fastest way to snowballing, one of the big reasons snowballing came down in the past is that defections essentially stopped.
Could very well be. The only time I ever witnessed a defection in 500 hours of playing was when the game first released, and everyone, including ruler clans, would join other kingdoms once it was looking bad. And that playthrough was only a few years long.
 
I haven't made this assumption yet because armies also cooperate in defending sieges as well now. This is why you don't see much happen in the first 5 years.

The biggest anomaly so far in 1.8.0 compared to other versions is the high defection rate. The weirdest part is clans in kingdoms that are winning are defecting as well, which makes me think there is a bug with it. Also I can tell you that defections are the fastest way to snowballing, one of the big reasons snowballing came down in the past is that defections essentially stopped.
Hey @Blood Gryphon, can you send a "natural" save file before mass defections happen? You may submit it using our ticket system. Thank you!
 
Hey @Blood Gryphon, can you send a "natural" save file before mass defections happen? You may submit it using our ticket system. Thank you!
I only have year 4, 9, 14, and 20 for test 2, the defections seem to start to happen between years 4 and 9 but the majority of them seem to happen after year 9. Going to drop yall 4, 9 and 14. Will get you some more when i run another test later today.

Ticket submitted.
 
Last edited:
Thanks a lot @Blood Gryphon we will analyze what's causing the problem.
Going to submit another ticket with my most recent test. Northern empire some how loses 3 starting clans after year 10 even though by year 10 they own almost all the empire. Going to send year 10, 15, and 20 so you can see the progression.

Will post snowball stats later today for this and another test running right now.

Ticket submitted under "Snowball Test 3 defection saves".
 
Back
Top Bottom