Diplomacy Developments

[Bannerman Man]

Right but what i meant by that question is - is that 20%an absolute or is it regulated by other meaningful factors like suffering penalties for Cav in the mountains or cav in Snow or enemy pikeman etc..

No, it's just pure upside, other than the fact that they cost more to recruit and, in the case of the player, to upgrade.

Maybe the solution could be to decrease the strenght of the Triabal Warriors and make the higher tier cavs harder to get.

Troop stats don't factor into auto calc at all currently, so this would not affect anyone but the player and any AI the player battles in real time. Unless you mean replace some of the cavalry options from their troop tree.

Maybe one final test, if possible. Give Khuzait the speed bonus but still don´t give any faction the +cav bonus for autocalc.

I've actually tested this already in both 1.5.0 and 1.5.1. Here are the results:

1.5.0

Spoiler: 1.5.0 No Auto Calc Cav Bonus

1.5.1

Spoiler: 1.5.1 No Auto Calc Cav Bonus

This is only one test a piece, but I don't think the cavalry bonus is as strong as people make it out to be. I think it's the sum total of their advantages, but the extra speed is probably the biggest factor.

**By the way, a confounding factor in these tests is the "King Bonus" for influence applying to everyone, which would make armies much more common (I can't remember if that was an issue in 1.5.0, though).**

 

[froggyluv]

No, it's just pure upside, other than the fact that they cost more to recruit, and in the case of the player, to upgrade.

I just dont get this. How hard would it be to apply a few factors that would mitigate some of their over-poweredness? Seems the most reasonable rather than a binary on/off 20% auto calc bump. Thats the kinda features even the most rudimentary strategy games possess

 

[StewVader]

Mexxico, correct me if I'm wrong, but doesn't tax corruption happen already based on security and loyalty? I'm talking about the CalculateSecurityGoldCutDueToCorruption method that is called by CalculateDailyTaxInternal. Security needs to be below 75 for that to happen, and at the moment it's very easy to keep security high.


That's not a bad idea. Though for tools specifically, towns already get a small (mostly negligible, needs a buff) boost to their construction stats for each tool they consume per day. You'd want to make sure that towns and their bound villages are not competing for certain resources if they produce an effect. You would also need to work out a formula for how many resources villagers would buy and consume each trip, and for that you'd need to give every applicable resource some demand values so they are not consumed to quickly or slowly. There is already one such list for towns, and if you add another one for villages it becomes harder to tune.

Couldn't you just put an artisan in each village? (And then apply some buffs depending on the consumption thresholds).

I'm thinking tools for construction speed
food variety for militia morale/strength
luxury items could improve quality of troops

I don't know. Just seems kinda bad that villages don't buy anything from the towns. Would probably make workshops better too.

 

[Blood Gryphon]

Spoiler: Mexxico's Auto calc and khuzait bonus removed test

No, it's just pure upside, other than the fact that they cost more to recruit and, in the case of the player, to upgrade.


Troop stats don't factor into auto calc at all currently, so this would not affect anyone but the player and any AI the player battles in real time. Unless you mean replace some of the cavalry options from their troop tree.


I've actually tested this already in both 1.5.0 and 1.5.1. Here are the results:

1.5.0

Spoiler: 1.5.0 No Auto Calc Cav Bonus

1.5.1

Spoiler: 1.5.1 No Auto Calc Cav Bonus

This is only one test a piece, but I don't think the cavalry bonus is as strong as people make it out to be. I think it's the sum total of their advantages, but the extra speed is probably the biggest factor.

**By the way, a confounding factor in these tests is the "King Bonus" for influence applying to everyone, which would make armies much more common (I can't remember if that was an issue in 1.5.0, though).**

You are right, especially when comparing those to Mexxico's world state makes it pretty clear the extra speed is a big factor and not to be ignored. Although looking at Mexxicos again it seems to be untainted by the king bonus bug so thats probably playing a heavy influence.

 

[froggyluv]

Like every Cavalry unit has an Infantry counter-part, and whenever the horse die, they become infantry in the next battle unless you re-upgrade them with a horse?

Yes exactly. Now add to that certain regions having a booming Horse markets while others drying up - Raid the Stud Farms!!

 

[Stromming]

Yes exactly. Now add to that certain regions having a booming Horse markets while others drying up - Raid the Stud Farms!!

Your idea combined with T2 khuzait troops having 50/50 (or whatever percentage you want) chance of getting a horse (just like how khergits t2 work in warband) could be what the game needs, after applying mexxico's fixes as well as getting rid of the cav speed and calc bonus.

 

[Worlok]

Troop stats don't factor into auto calc at all currently, so this would not affect anyone but the player and any AI the player battles in real time. Unless you mean replace some of the cavalry options from their troop tree.

I thougt the logic of the autocalc takes into consideration the composition of the party/army (cav, inf, arch, T2, T3, T4 etc...). If the system of autocalc does not differentiate Triabal warriors, raiders, horse archers in the calculation, well that's too bad.

 

[MostBlunted]

I've actually tested this already in both 1.5.0 and 1.5.1. Here are the results:

Thanks.

But you still don´t have the same options as Mexxico has and his internal builds I guess?

But of course the sample size is too small, but when it comes down to AI vs AI I guess +20% is more important than the movement bonus, in my opinion.

Also one screenshot is from 1090 and one from 1092. You can´t compare those in my opinion. The screenshot from Mexxico is from 1093.

I still believe the cav autocalc bonus is an issue. Doesn´t make a lot of sense to compare different game versions at all in my opinion.

No offense.

 

[Stromming]

I thougt the logic of the autocalc takes into consideration the composition of the party/army (cav, inf, arch, T2, T3, T4 etc...). If the system of autocalc does not differentiate Triabal warriors, raiders, horse archers in the calculation, well that's too bad.

Surely it must?!

 

[Old-Bull]

I enjoy the fact that cavalry dominant armies can expand quicker and take territory faster. That's accurate to history. That's what happens when an empire has a dominant cavalry advantage. They move faster. They fight faster. They expand faster. That's how it works.

I think the solution is to have realistic dimplomatic solutions to deal with it. If two countries are losing territory rapidly to the same enemy they should be a hell of a lot more likely to ally against that enemy. Khuzait is currently taking territory rapidly from multiple enemies at the same time. They expand into both Sturgia and Empire on every play through. Why don't Sturgia and Empire organize, ally with each other, and drive Khuzait back?

Economically, Khuzait needs some changes. The Khuzait come from pretty barren high plains. Their home territory should be pretty poor.

I hope Khuzait combat performance doesn't get nerfed because their behavior is accurate to history. I just hope someone can think of an economic nerf to reign them in.

 

[MostBlunted]

Why don't Sturgia and Empire organize, ally with each other, and drive Khuzait back?

Because this option isn´t in the game right now. If alliance become real, we can talk about it.

Nobody knows if this option will ever be in vanilla.

"We" can balance the game on future dreams and hopes or balance it as it is right now.

Right now, only one faction steam rolls every time.

 

[froggyluv]

I thougt the logic of the autocalc takes into consideration the composition of the party/army (cav, inf, arch, T2, T3, T4 etc...). If the system of autocalc does not differentiate Triabal warriors, raiders, horse archers in the calculation, well that's too bad.

So this is what i found as an explanation of how how Vanilla Auto Calc works at a mod site called Better Auto Calc:

What does the game do?

The way auto resolve works is (basically) like this:
- get a random troop from a party.
- get the power level of this troop, and apply it as damage to another defending troop.

The problem with native, is that the power levels of all troops are actually quite close to each other, as far as the game is concerned, there's not THAT much difference between a tier 1 and a tier 5. There is a difference, but it isn't a lot, which is why you often see high tiers be killed by low tier trash.

So its makes a very slight distinction it seems. Man this has bugged me since Mount and Blade with so many mods doing it better -i thought for sure they would have upped their game on this aspect. Not asking for Panzer General level mechanics but surely they can do better than using a basic 5 point tier system.

Honestly im kinda shocked this isnt a premiere issue with all the complaints of Snowballing. Its Auto-Calc that IS THE DEFINING MECHANIC for AI battle outcomes -how can it be this simplified and people not expect simple results

 

[froggyluv]

Just to add -using terrain effects on battle auto-calc would open up so much -such as ever notice how parties just chase each other randomly around the map with no thought of where they are going or why -only care is that they get away. Now imagine if the terrain played into their travel plans calculation. Say a party of Forest Bandits would do their best to get back to the Woods -giving them a battle % bump. Sturgians would try and get back to snow or mountains tc.. this would give semblance and reason to the world map rather than watching a the GoldFish Bowl world of big fish eat little fish we have now.

 

[Revenant342]

Anyone have the actual power level numbers different troops types get assigned? I'm curious about this. Balancing it properly seems like a real double-edged sword though.

On the one hand, it definitely does feel like higher-tier troops need to be more effective in autoresolve. If a Legionary is only slightly better than a peasant and costs 10x the upkeep, that's a problem.

On the other hand, making higher-tier troops more effective in autoresolve without any other changes would make strong factions even stronger and weak factions running around with peasant armies even worse off, so it would be even harder for factions that take significant early losses to compete and recover.

 

[Bannerman Man]

I just dont get this. How hard would it be to apply a few factors that would mitigate some of their over-poweredness? Seems the most reasonable rather than a binary on/off 20% auto calc bump. Thats the kinda features even the most rudimentary strategy games possess

Don't know ?‍. If we really sat down and thought about it I'm sure we could probably come up with all sorts of reasons.

So, the GetPower method is what calculates the strength of an individual unit in auto calc. This code is also used to compare two parties/armies to see if an AI should attack or flee, or where to besiege, etc. It executes a lot. The more variables you add to it, the more complex the rest of the AI must become to account for it. If they're situational variables, then it's even more complex.

Say you give cavalry a 0.8 penalty in auto calc for fighting in the sand or snow only. When two hostile parties apporach each other at the border of two terrain types, one might be on snow and the other on grass. How should they apply the power formula in that case? Should they consider what terrain they are currently on, or where they're likely to make contact, or where the other party is? Once one of them crosses the boundary to the other terrain, will they now sense they are weaker and turn and flee? Will it become a cat and mouse game as they go back and forth over terrain types?

If you give certain troop types a bonus or penalty against another type (e.g. cav vs. pikes), you now need to compare all of the unit types in each party in some way to see who is superior. The number of calculations on a piece of code that is supposed to execute frequently increases significantly, which might cause performance problems.

As a human it's easy to develop an intuition for these things as a shortcut, but it's really hard to program a computer to have that same kind of intuition, so instead you need to do tons of calculations to compare every aspect.

Those are just some things that may or may not have been brought up at the design table. Something might seem like a no-brainer to us, but I just find it hard to believe that a group of people could sit around and think about these things for 8 years and not have good reasons for doing the things they did. The auto resolve power formula that tournaments use is quite a bit more complex than the general use formula, so it's not like they aren't capable of designing it.

Sorry if that all sounds condescending; I don't mean it to be.

 

[Bannerman Man]

Also one screenshot is from 1090 and one from 1092. You can´t compare those in my opinion. The screenshot from Mexxico is from 1093.

I know, I intentionally cut them short because Khuzait was expanding in one direction and it was clear what the outcome was going to be. The two year difference in the two tests is at least a testament to mexxico's changes to diplomacy. I also think the world state looks better in the outcome of the second test. I might run some more tests later, including tweaking the speed related aspects.

My intention was also not to compare versions, but to show that the snowballing without the cav bonus is independent of the changes made to diplomacy.

 

[froggyluv]

If you give certain troop types a bonus or penalty against another type (e.g. cav vs. pikes), you now need to compare all of the unit types in each party in some way to see who is superior. The number of calculations on a piece of code that is supposed to execute frequently increases significantly, which might cause performance problems.

Well for these type of things we look at precedent. Has it been done before and on this engine without compromising opitmization? It has. It was mod i was a beta tester for for the original M&B and I cant even remember the name right now but the author definitely devised a very complex and detailed Auto-Calc system that was far more satisfying than anything Vanilla has done by taking all troop Levels/Types/ Weapons and Armor even Terrain into a big soup of math with very detailed results coming out. The game felt like a real live strategy game. Guy even had a fully fledged Religion with all of its advantages and penalties built in. Perhaps the Sword of Damocles mod also did somethin similar if not quite as involved -the point being it can and should be done.

Say you give cavalry a 0.8 penalty in auto calc for fighting in the sand or snow only. When two hostile parties approach each other at the border of two terrain types, one might be on snow and the other on grass. How should they apply the power formula in that case? Should they consider what terrain they are currently on, or where they're likely to make contact, or where the other party is? Once one of them crosses the boundary to the other terrain, will they now sense they are weaker and turn and flee? Will it become a cat and mouse game as they go back and forth over terrain types?

Well if you had the above system in place -one party would decidely be stronger and the aggressor at first meet. So it need only be the Defender to make a choice of whether to run for it (to a perceived more advantageous terrain) or stand their ground. Of course if they were near even matched there may be a bit of back and forth type prey/predator role shifts but that sometimes happens now and only for goofy reasons -id far prefer they a least attempt some Army logic even if it turned out to be ill advised.

As a human it's easy to develop an intuition for these things as a shortcut, but it's really hard to program a computer to have that same kind of intuition, so instead you need to do tons of calculations to compare every aspect.

Of course -but thats exactly why we have computers capable of 2 Billion cycles per second.

Those are just some things that may or may not have been brought up at the design table. Something might seem like a no-brainer to us, but I just find it hard to believe that a group of people could sit around and think about these things for 8 years and not have good reasons for doing the things they did. The auto resolve power formula that tournaments use is quite a bit more complex than the general use formula, so it's not like they aren't capable of designing it.

Because it was a grueling affair -i do remember testing it over and over and this guy was a real math whiz. Dont be surprised by the Path of Least resistance should your audience afford it to you.

Sorry if that all sounds condescending; I don't mean it to be.

Not at all.

Edit: How to find that mod i was talking about its called Conquest! for M&B
Funny i found a question I asked in 2009 on his forums

froggyluv said
Sounds good. What do you mean by the auto-resolve has been fixed? Will it now include weapons and armour etc...

Yes.

You can read more about the journey here: http://forums.taleworlds.com/index.php/topic,8652.msg1575773.html#msg1575773

 

[sir Gregor Clegane]

Diplomacy worked fine in version 1.3.1. and why it was necessary to change it to not logical and stupid algorithms I do not understand. Due to diplomacy alone, I am eager to roll back the game to earlier versions, because I am very tired of all the foolishness of AI. By the way, is there a mod that returns diplomacy version 1.3.1 or which file from the old version needs to be returned? If someone tells me I will be very grateful.

 

[Bannerman Man]

Has it been done before and on this engine without compromising opitmization?

Has it on BL's engine? The only auto calc mod I know of so far for Bannerlord is this one, which just increases the power difference between troop tiers. If you say it was done in a mod for the OG M&B I'll take your word for it because I don't know how that game functioned under the hood.

Like I said, the tournament power formula is more complex and factors in weapon and armor values, horses and horse armor, weight of equipment, and skills, so maybe someone will eventually adapt it for use in the general auto calc code if it's possible.

Well if you had the above system in place -one party would decidely be stronger and the aggressor at first meet. So it need only be the Defender to make a choice of whether to run for it (to a perceived more advantageous terrain) or stand their ground. Of course if they were near even matched there may be a bit of back and forth type prey/predator role shifts but that sometimes happens now and only for goofy reasons.

Sure, in many cases it won't matter, but if you want to make these factors powerful enough to make a difference depending on the situation you'll still end up with many edge cases where the aggressor and defender flip in unexpected ways. My example was just one possibility.

-id far prefer they a least attempt some Army logic even if it turned out to be ill advised.

If you look at the tactics perk tree you'll see that they've at least considered using terrain and troop types in auto calc, but there may have been complications that caused them to scrap the idea. Or they could still be planning on implementing it at some point in the future and all of this is a moot point anyway. At the very least they've already tweaked the auto calc code, and I'm sure they'll continue to tune it if they feel they can improve it.

Of course -but thats exactly why we have computers capable of 2 Billion cycles per second.

This game is already a CPU hog though, and they're trying to develop it for a wide range of machines. There's still a limit.

I'm just trying to rationalize it from their perspective. It's okay if you don't agree with it.

 

[Bannerman Man]

Couldn't you just put an artisan in each village? (And then apply some buffs depending on the consumption thresholds).

I'm thinking tools for construction speed
food variety for militia morale/strength
luxury items could improve quality of troops

I don't know. Just seems kinda bad that villages don't buy anything from the towns. Would probably make workshops better too.

Yeah I guess you could put an artisan in a village. You'd still need to determine how much of each resource to transport back to the village, and some way to get rid of the excess so raw resources don't just stack up in a village's inventory if the rate of production doesn't quite line up.

The towns themselves are what consume the goods using a formula based on prosperity and item prices. Artisans just take raw goods and convert them into manufactured goods.