Information about developments at snowballing problem

Users who are viewing this thread

five bucks

Squire
Hello, the lancers just need to be a real anti-cavalry unit. No more cavalry, but just a real utility for the lancer and pikeman. the Pike Wall to break charges, like the Shield Wall to break archers. This is enough to reduce the impact of cavalry in the game.
All battles the AI fights against each other are auto-calculated, the way troops act in a real battle doesn't influence it.
 

Dabos37

Sergeant Knight at Arms
I have not had much time to play the game until this week, and I am really glad to see how the snowballing is completely fixed as other people have mentioned, and pretty happy about how enjoyable my new campaign is for this reason.

@mexxico it is certainly a shame that you are going to leave the company soon, and it is really sad for us that TW is nor doing anything to try to keep you in the team. I am really concerned about snowballing coming back in the future due to some changes and additions, who would assist us with this in the same way you did it. Tournaments AI has been downgraded since 1.0.9-1.0.10 patch or so and nobody in TW cares about fixing it, it would be really sad if same would happen with campaign AI in the future :’-( .
 

mexxico

Sergeant Knight
I have not had much time to play the game until this week, and I am really glad to see how the snowballing is completely fixed as other people have mentioned, and pretty happy about how enjoyable my new campaign is for this reason.

@mexxico it is certainly a shame that you are going to leave the company soon, and it is really sad for us that TW is nor doing anything to try to keep you in the team. I am really concerned about snowballing coming back in the future due to some changes and additions, who would assist us with this in the same way you did it. Tournaments AI has been downgraded since 1.0.9-1.0.10 patch or so and nobody in TW cares about fixing it, it would be really sad if same would happen with campaign AI in the future :’-( .

We are trying to add different test mechanics to keep game world keep running properly. So if any problem happens these auto test systems will give alerts. So company can find which addition broke game. For example we are counting how many towns / castles change hand in first 20 years and we find average town / castle hand change per year, currently this average should be between 3-7.5 for both castles and towns. We are counting how many raids happening each year and find an average, it should be between 30-70, success rate of raids should be 40-60% (so if any lord started raid it should completed without problem with 40-60%). We also count gold amount in towns and poor clan leader's ratio (with money < 30K) and tier1-2 troop distribution (should be between 40-55%). And so many things like these...

If in future something is broken and this town / castle change hand per year average go above 7 or go down 3 or raid counts go out of limits or raid success rate go out of its limits these test systems will alert. So even I am not here they can find out which addition created problem and broke game world balance. Also these test systems calculate snowballing scores and it should be between 10-50. If score go out of this range test again give alert. We count militia count averages, garrison averages, starving towns, rebellions per year. For example last tests show rebellion per year average was below limits then we lowered decay effect of loyality to increase rebellions (it was (x - 55) * 0.0125 now it is (x - 50) * 0.01), so if loyality go down decay effect will give less positive loyality bonus. After this change rebellions again be 1.5 per year so it is a number between expected limits (limits for rebellion average per year is 1 - 1.75). We also count successfull rebellion ratio and it should be between 15-45%, this ratio was nearly 1-2% in 1.5.9 and we give some bonuses to rebel clans to increase their survival chance after they rebel (+1 garrison per day to town rebelled, more initial money, 100 initial militias to town rebelled, food stocks of rebelled town is now filled).

So before I leave (August 1) we are trying to completely finish this system so this system will try to keep game world running properly without need to anyone. If any change damage anything in sandbox world these tests will give alerts. I do not think but if still there happen a big problem which could not be solved for a long time I could help team but I do not think it can happen.
 
Last edited:

Dabos37

Sergeant Knight at Arms
We are trying to add different test mechanics to keep game world keep running properly. So if any problem happens these auto test systems will give alerts. So company can find which addition broke game. For example we are counting how many towns / castles change hand in first 20 years and we find average town / castle hand change per year. Currently this average should be between 3-7.5 for both castles and towns. We are counting how many raids happening each year and find an average. If in future something is broken and this town / castle change hand per year average go above 7 or go down 3 these test systems will alert or there happen less raids than 5 per year or more raids than 20 per year system again give alert. So even I am not here they can find out which addition created problem and broke game world balance. Also these test systems calculate snowballing scores and it should be between 10-50. If score go out of this range test again give alert. We count militia count averages, garrison averages, starving towns, rebellions per year. For example last tests show rebellion per year average was below limits then we lowered decay effect of loyality to increase rebellions (it was (x - 55) * 0.0125 now it is (x - 50) * 0.01), we also count successfull rebellion ratio and it should be between 15-45%, this ratio was nearly 1-2% in 1.5.9 and we give some bonuses to rebel clans to increase their survival chance after they rebel (+1 garrison per day, more initial money, 100 initial militias to town rebelled). So before I leave (August 1) we are trying to completely finish this system so this system will try to keep game world running properly without need to anyone. I do not think but if still there happen a big problem which could not be solved for a long time I could help team but I do not think it can happen.

Ok, pretty good news to me, I have been really concerned about some new changes breaking the snowballing balancing in the future, after the huge effort you have invested on this issue to fix it. Especially thinking that aside from some exceptions, it is really hard to get other devs' attention when reporting broken things. It is still pretty sad for me you leave the company because most of my favorite changes and ideas have come from you, but at least it is good to know that there will be a system to avoid step-backs on this matter. Thank you very much and I hope that a miracle happens and you finally decide to stay :smile: .
 
We are trying to add different test mechanics to keep game world keep running properly. So if any problem happens these auto test systems will give alerts. So company can find which addition broke game. For example we are counting how many towns / castles change hand in first 20 years and we find average town / castle hand change per year, currently this average should be between 3-7.5 for both castles and towns. We are counting how many raids happening each year and find an average, it should be between 30-70, success rate of raids should be 40-60% (so if any lord started raid it should completed without problem with 40-60%). We also count gold amount in towns and poor clan leader's ratio (with money < 30K) and tier1-2 troop distribution (should be between 40-55%). And so many things like these...

If in future something is broken and this town / castle change hand per year average go above 7 or go down 3 or raid counts go out of limits or raid success rate go out of its limits these test systems will alert. So even I am not here they can find out which addition created problem and broke game world balance. Also these test systems calculate snowballing scores and it should be between 10-50. If score go out of this range test again give alert. We count militia count averages, garrison averages, starving towns, rebellions per year. For example last tests show rebellion per year average was below limits then we lowered decay effect of loyality to increase rebellions (it was (x - 55) * 0.0125 now it is (x - 50) * 0.01), so if loyality go down decay effect will give less positive loyality bonus. After this change rebellions again be 1.5 per year so it is a number between expected limits (limits for rebellion average per year is 1 - 1.75). We also count successfull rebellion ratio and it should be between 15-45%, this ratio was nearly 1-2% in 1.5.9 and we give some bonuses to rebel clans to increase their survival chance after they rebel (+1 garrison per day to town rebelled, more initial money, 100 initial militias to town rebelled, food stocks of rebelled town is now filled).

So before I leave (August 1) we are trying to completely finish this system so this system will try to keep game world running properly without need to anyone. If any change damage anything in sandbox world these tests will give alerts. I do not think but if still there happen a big problem which could not be solved for a long time I could help team but I do not think it can happen.
this seems like an excellent system to have :smile: (I actually thought that besides the steamroll testing, a system like this was working in the background)
and as others have said, thank you for taking forum members suggestions seriously and your continuous in-depth communication on the game systems, very interesting
 
Last edited:
We are trying to add different test mechanics to keep game world keep running properly. So if any problem happens these auto test systems will give alerts. So company can find which addition broke game. For example we are counting how many towns / castles change hand in first 20 years and we find average town / castle hand change per year, currently this average should be between 3-7.5 for both castles and towns. We are counting how many raids happening each year and find an average, it should be between 30-70, success rate of raids should be 40-60% (so if any lord started raid it should completed without problem with 40-60%). We also count gold amount in towns and poor clan leader's ratio (with money < 30K) and tier1-2 troop distribution (should be between 40-55%). And so many things like these...

If in future something is broken and this town / castle change hand per year average go above 7 or go down 3 or raid counts go out of limits or raid success rate go out of its limits these test systems will alert. So even I am not here they can find out which addition created problem and broke game world balance. Also these test systems calculate snowballing scores and it should be between 10-50. If score go out of this range test again give alert. We count militia count averages, garrison averages, starving towns, rebellions per year. For example last tests show rebellion per year average was below limits then we lowered decay effect of loyality to increase rebellions (it was (x - 55) * 0.0125 now it is (x - 50) * 0.01), so if loyality go down decay effect will give less positive loyality bonus. After this change rebellions again be 1.5 per year so it is a number between expected limits (limits for rebellion average per year is 1 - 1.75). We also count successfull rebellion ratio and it should be between 15-45%, this ratio was nearly 1-2% in 1.5.9 and we give some bonuses to rebel clans to increase their survival chance after they rebel (+1 garrison per day to town rebelled, more initial money, 100 initial militias to town rebelled, food stocks of rebelled town is now filled).

So before I leave (August 1) we are trying to completely finish this system so this system will try to keep game world running properly without need to anyone. If any change damage anything in sandbox world these tests will give alerts. I do not think but if still there happen a big problem which could not be solved for a long time I could help team but I do not think it can happen.
Great job for creating such monitoring system.
Respect!
 

Dabos37

Sergeant Knight at Arms
@mexxico While snowballing is now fixed, I am noticing too short wars again. Only 10-15 days average as much most of the time and the message log is showing continuously how the kingdoms go to war and peace every few days. Not sure if it has started after the last hotfix or before, but currently the wars are too short and there is no time to do much.
 

mexxico

Sergeant Knight
@mexxico While snowballing is now fixed, I am noticing too short wars again. Only 10-15 days average as much most of the time and the message log is showing continuously how the kingdoms go to war and peace every few days. Not sure if it has started after the last hotfix or before, but currently the wars are too short and there is no time to do much.
Thanks for report. I will check it at monday how 1.5.10 stats are. We should add average war duration to monitoring system also.
 

Dabos37

Sergeant Knight at Arms
Thanks for report. I will check it at monday how 1.5.10 stats are. We should add average war duration to monitoring system also.

Yes please, I think that sacrificing some snowballing score in order to have some long wars would improve the gameplay and make it feels less random. Currently, kingdoms are declaring random wars and making peace really fast, which also makes buying workshops extremely risky if you are playing as mercenary for a kingdom.

It is ok to see some 1-10 days wars if a kingdom is fighting against multiple enemies, but what I am seeing a lot, is two kingdoms fighting a 1vs1 war and still making peace really fast. Plus reducing a bit the new war declarations chance would be also welcome. Thank you!!
 

mexxico

Sergeant Knight
Sir, any update?
Currently average war duration at 1.5.10 (which will be beta in 1 week hopefully) is about 30 days and 1-2-3 day wars has a ratio of 10-15%. We added a new test for monitoring this. If this value goes below 30 days it will give alert. I will try to increase this duration. I wanted a wars schedule from team (this schedule shows wars in a big table). When I receive it I will post but not yet received.
 

Irfan Ross

Regular
Currently average war duration at 1.5.10 (which will be beta in 1 week hopefully) is about 30 days and 1-2-3 day wars has a ratio of 10-15%. We added a new test for monitoring this. If this value goes below 30 days it will give alert. I will try to increase this duration. I wanted a wars schedule from team (this schedule shows wars in a big table). When I receive it I will post but not yet received.
Thank you.
 

Blood Gryphon

Master Knight
WBVC
Just some historical data for everyone to consider when thinking about average days for wars. (data provided by Mexxico)

n3fbs.png


@mexxico using average (mean) for the check will be helpful, but using the median might be more telling. Looking at 1.5.1 upcoming hotfix stats, it would have passed the average >30 check even with 53% of wars being less than a season, if you used median it would be flagged.

Some notes to help read the image:
Negotiation = 1 day or less
Month = 7 days
Season = 21 days
Year = 84 days
 
Last edited:
Just some historical data for everyone to consider when thinking about average days for wars. (data provided by Mexxico)

n3fbs.png


@mexxico using average (mean) for the check will be helpful, but using the median might be more telling. Looking at 1.5.1 upcoming hotfix stats, it would have passed the average >30 check even with 53% of wars being less than a season, if you used median it would be flagged.

Some notes to help read the image:
Negotiation = 1 day or less
Month = 7 days
Season = 21 days
Year = 84 days

It always makes me happy when folks suggest using the median over the mean. It certainly helps protect the analysis from being skewed.
 

Dabos37

Sergeant Knight at Arms
Currently average war duration at 1.5.10 (which will be beta in 1 week hopefully) is about 30 days and 1-2-3 day wars has a ratio of 10-15%. We added a new test for monitoring this. If this value goes below 30 days it will give alert. I will try to increase this duration. I wanted a wars schedule from team (this schedule shows wars in a big table). When I receive it I will post but not yet received.

Thank you very much Mexxico. Would be great if you could get something like 50 days average war duration or at least 40. We are currently having great snowballing scores, so it won’t probably affect much if you could manage to increase the war duration average, which would improve gameplay for sure. Especially if you could manage to increase the amount of wars greater than a season. Thanks!

@Blood Gryphon interesting point. Agree with that.
 
Last edited:

tlowey

Sergeant at Arms
I have a question about how many kingdoms you should be fighting at once. My game started in 1.5.7 through t 1.5.9 so not sure if that has a bearing. I am about 37 years into the game and am fighting 5 kingdoms at once.My power level is probably equivalent to the 5 kingdoms I am fighting.
Is this normal as it is a bit of a pain and have to rely on the vassals AI?
 

Blood Gryphon

Master Knight
WBVC
I have a question about how many kingdoms you should be fighting at once. My game started in 1.5.7 through t 1.5.9 so not sure if that has a bearing. I am about 37 years into the game and am fighting 5 kingdoms at once.My power level is probably equivalent to the 5 kingdoms I am fighting.
Is this normal as it is a bit of a pain and have to rely on the vassals AI?
It is normal, right now your kingdom will be willing to fight an equal amount of enemies as its current strength. If you take some loses and your strength drops below the 5 kingdoms total strength your lords will start to ask for peace. Also if you are king even if they are ready for peace they may wait a bit on voting on it as they have a reduced chance to start votes, so the player has more control over their kingdom.
 

tlowey

Sergeant at Arms
It is normal, right now your kingdom will be willing to fight an equal amount of enemies as its current strength. If you take some loses and your strength drops below the 5 kingdoms total strength your lords will start to ask for peace. Also if you are king even if they are ready for peace they may wait a bit on voting on it as they have a reduced chance to start votes, so the player has more control over their kingdom.
Thanks for the quick reply I am the King so looks like I will have to rely on my vassals to protect my towns, while I am in the deep south fighting the Aserai.
 
Top Bottom