Disappointing progress.

Users who are viewing this thread

NPC99

Baron
M&BWBWF&SVC
Apparently disbanding party is now broken too. Incredible. I mean how difficult is the logic? It's really not that complex. Someone or a group of someones coded that but it's management's fault that they got the logic wrong. Right.
Spoken like someone who's never looked at the code.
 

Earth Dragon

Sergeant
Right. No one's to blame, everyone's great but something went wrong. Bad luck I suppose.

Would be nice if people actually started taking ownership of their work. Nothing wrong with admitting fault. That's actually how you start to improve but responsibility seems to be anathema to people nowadays. Everyone's a victim, no one's responsible. It was all because of XYZ.

Apparently disbanding party is now broken too. Incredible. I mean how difficult is the logic? It's really not that complex. Someone or a group of someones coded that but it's management's fault that they got the logic wrong. Right.
We’re not exactly in a merit based world society as of now, so it’s harder said then done. Even in Turkey, you’re gonna get flak for firing someone who wasn’t of a center look and lifestyle. This is the world we’ve created. This is what millennials wanted
 

Orion

Still Not Worthy
Global Moderator
M&BWBWF&SNW
Don't turn this into your political soapbox. First & last warning.
 

Apocal

Grandmaster Knight
What about terrain? Defender bonus? Troop ratio bonus/penalties? These are basic military calculations. If I have 100 soldiers vs 10 looters, I should NEVER lose a single person. Yet it's possible in autocalc for my Elite Cataphract to die.

Because the formula is so basic, it caused the Kuzait snowball. How did they fix it? Did they improve autocalc? No, they gave everyone more cavalry and endless recruits. When that didn't work, they just made the AI declare war against impossible odds.

A more elegant solution would have been to improve autocalc to factor in defender bonus and terrain bonus. Battanians fighting in forest should have the upper hand against Vladian cavalry. Sturgians fighting in snow should be able to catch up to and ambush Kuzaits.
Well, mexxico just nerfed the Cav bonuses overall and that was enough. I don't recall mexxico doing anything to wardecs to fix it. Sturgians did used to get a pretty substantial snow bonus that made it so they could catch Khuzait parties in the snow. TW nerfed it for whatever reason but it had zero effect on snowballing even wen it was gloriously high.

(Also, "basic" military calculations aren't you take 0 losses at 10:1 odds. It is you take roughly equivalent losses in an attrition exchange.)
Of course there is no specific employee to blame and I´m quite sure that the devs are not incompent or anything. But I also don´t see anyone blaming the employees.
They don't blame them by name, but posters have definitely called specific dev's design and balance changes incompetent before.
 
Spoken like someone who's never looked at the code.

2 things. Logic > code. If your logic sucks, the code will suck. If code is spaghetti, whose fault is that? Management?

Well, mexxico just nerfed the Cav bonuses overall and that was enough. I don't recall mexxico doing anything to wardecs to fix it. Sturgians did used to get a pretty substantial snow bonus that made it so they could catch Khuzait parties in the snow. TW nerfed it for whatever reason but it had zero effect on snowballing even wen it was gloriously high.

(Also, "basic" military calculations aren't you take 0 losses at 10:1 odds. It is you take roughly equivalent losses in an attrition exchange.)



Nerfing cav bonuses was certainly not enough. Sturgians had speed bonus in snow correct but they didn't have enough cav then. Battanians didn't rock around with a lot of cav either in 1.0

If memory serves, the way it's resolved is basically ATK unit 1 vs DEF unit 1. If A1 kills D1, A1 vs D2 and so on. If A1 vs D3, A1 dies, then A2 vs D3. Correct?

Which accounts for why a tier 6 cav in a 200 strong party can die in a fight against 10 looters because it was essentially 1 tier 6 cav vs 10 looters.

I don't think that's an accurate simulation.

And you're wrong. If I outnumber you 10 to 1, I would most likely not take equivalent losses. Only in fantasy can 1 man prevail against 10. I would agree that 0 losses has a none 0 probability. Like 0.01%.

They don't blame them by name, but posters have definitely called specific dev's design and balance changes incompetent before.

There's no need to name and shame. But certainly, there's nothing wrong with calling it like it is. A professional is not going to allow emotions to cloud their judgement.
 

MostBlunted

Banned
They don't blame them by name, but posters have definitely called specific dev's design and balance changes incompetent before.
But those are just a few guys I guess (MP forum?) because I never saw someone blaming specific people.

But the majority is friendly to the devs as it should be.

I personally only blame Orion for not having a good taste for dope memes.
 

NPC99

Baron
M&BWBWF&SVC
2 things. Logic > code. If your logic sucks, the code will suck. If code is spaghetti, whose fault is that? Management?
The basic logic of the game is AI independence. A sandbox with hundreds of active parties on the campaign map, interacting based on probabilities that create unique and at times unpredictable outcomes. Ditto individual agents interacting in battle scenes as opposed to basing AI control on formations as per Total War. That underlying principle is intended to make game play unpredictable to make the game replayable. However, that approach inevitably allows for unique coding conflicts and unanticipated bugs, which take time to isolate and cull as some might only appear after x hundred days etc. I'd rather have a sandbox game with its unique and challenging code problems than the hardcoded logical outcomes of a dressed up puzzle game where you move from one rigged encounter to the next with minimal player freedom. I only wish TW had ignored consoles and chosen a higher minimum pc spec to make the game less constrained, but that was probably commercially unviable.

PS while its not the underlying xml data, C# game programs or C++ engine code take a look at Bannerlord's API documentation if you want a basic glimpse of the scope of what's involved in coding the game. https://apidoc.bannerlord.com/v/1.8...r_base.html#a05ec45ba9a8707a6048fa5ba129fb438
There's no need to name and shame. But certainly, there's nothing wrong with calling it like it is. A professional is not going to allow emotions to cloud their judgement
True professionals care about their work. Any professional investing years creating a product becomes emotional engaged with it. Knee jerk criticism by players with no idea about the necessary coding/modelling compromises is bound to hurt.
 
Last edited:
The basic logic of the game is AI independence. A sandbox with hundreds of active parties on the campaign map, interacting based on probabilities that create unique and at times unpredictable outcomes. Ditto individual agents interacting in battle scenes as opposed to basing AI control on formations as per Total War. That underlying principle is intended to make game play unpredictable to make the game replayable. However, that approach inevitably allows for unique coding conflicts and unanticipated bugs, which take time to isolate and cull as some might only appear after x hundred days etc. I'd rather have a sandbox game with its unique and challenging code problems than the hardcoded logical outcomes of a dressed up puzzle game where you move from one rigged encounter to the next with minimal player freedom. I only wish TW had ignored consoles and chosen a higher minimum pc spec to make the game less constrained, but that was probably commercially unviable.

You're trying to obfuscate the argument by raising irrelevant points. Take a look at the actual bug report before commenting further.


When you've actually read it, tell me what's right or wrong with the logic. Can it be improved? If so, how. If not, why. Bonus points if you can support your statement here that the party disband logic is impacted by AI independence and/or interactions based on probabilities.


True professionals care about their work. Any professional investing years creating a product becomes emotional engaged with it. Knee jerk criticism by players with no idea about the necessary coding/modelling compromises is bound to hurt.

Not really. A professional is just someone who does something for money. Some professionals care and some don't. I don't think it's given everyone will become emotionally attached and I don't see why criticism would hurt. If it's valid, process it and if it's not ignore it.

I'm being genuine here. I cannot appreciate why anyone would get upset over words.
 

Apocal

Grandmaster Knight
If memory serves, the way it's resolved is basically ATK unit 1 vs DEF unit 1. If A1 kills D1, A1 vs D2 and so on. If A1 vs D3, A1 dies, then A2 vs D3. Correct?
Not quite. The attackers and defenders change every round, selected at random. So it would be A1 kills D1, A2 kills D2, D3 kills A3, A1 and D4 kill each other, etc. The reason high-tier units die so easy is because (or was, it has been a long time since I looked it over) is that the high-tier troops' primary autocalc advantage is in killing power, not in defending themselves; the two are separate.

Also their power is clamped so they don't have too far of an advantage over low-tier troops, like Looters.
And you're wrong. If I outnumber you 10 to 1, I would most likely not take equivalent losses. Only in fantasy can 1 man prevail against 10. I would agree that 0 losses has a none 0 probability. Like 0.01%.
I didn't say the underdog would prevail. I said the winner would take equivalent losses. In the case of 100 dudes against 10, there would be 90 victorious at the end of it. Attrition exchange: I kill you, you kill me and last man standing wins.

It would make for a worse game, because players would spend even more time recovering their numbers between battles, but that's why devs shouldn't add realistic elements (especially military ones) to everything under the sun. It wouldn't even mean the autocalc was any different in practice. Autocalc is already heavily tilted in favor of raw numbers (moreso now than on release) and it is rare for the side with less combat power to win, except when other parties jump in to save them.
 

NPC99

Baron
M&BWBWF&SVC
Not really. A professional is just someone who does something for money. Some professionals care and some don't. I don't think it's given everyone will become emotionally attached and I don't see why criticism would hurt. If it's valid, process it and if it's not ignore it.
An employee does something for money and may or may not care about it. A true professional has professional standards and is invested in his/her work.
 
An employee does something for money and may or may not care about it. A true professional has professional standards and is invested in his/her work.

Nonsense. Go check a dictionary. But I noticed you totally ignored the more pertinent question in favour of the trivial.

Not quite. The attackers and defenders change every round, selected at random. So it would be A1 kills D1, A2 kills D2, D3 kills A3, A1 and D4 kill each other, etc. The reason high-tier units die so easy is because (or was, it has been a long time since I looked it over) is that the high-tier troops' primary autocalc advantage is in killing power, not in defending themselves; the two are separate.

Also their power is clamped so they don't have too far of an advantage over low-tier troops, like Looters.

Seems to me you're supporting my argument that autocalc is too basic. Higher tier troops have better armour and higher skills which should be represented by a better ability to defend themselves. Does autocalc reference skill levels at all?

Why should their power be clamped?

I didn't say the underdog would prevail. I said the winner would take equivalent losses. In the case of 100 dudes against 10, there would be 90 victorious at the end of it. Attrition exchange: I kill you, you kill me and last man standing wins.

The attrition exchange theory is easily disproven. Battle of Cannae comes to mind. I mean, history is literally littered with battles that totally disprove it.

It would make for a worse game, because players would spend even more time recovering their numbers between battles, but that's why devs shouldn't add realistic elements (especially military ones) to everything under the sun.

Only if you base autocalc on that rubbish theory. I mean, just try it. Go to a pub and pick a fight with a group of people and see if you can take one down before the rest jump you. You might succeed if you sneak up and crack one on the skull before the rest respond so be sure to avoid doing that. You'll want lines drawn up first.

It wouldn't even mean the autocalc was any different in practice. Autocalc is already heavily tilted in favor of raw numbers (moreso now than on release)

No, I would argue it's not. A well trained force of soldiers should rarely die to a motley bunch of looters. Yes, bad luck happens so we should see it happen but not at the frequency the game is exhibiting right now. Raw numbers aren't accounted for at all. It should be a force multiplier yet it's totally disregarded.

and it is rare for the side with less combat power to win, except when other parties jump in to save them.

Again, I would argue this is a result of the shoddy autocalc. Tactics mean absolutely nothing. Or near to nothing.

May I refer you to the Battle of Cannae once more. A smaller force double enveloped a larger force and totally decimated it due to superior leadership.

Can this happen in Bannerlords? No. Why not? Autocalc is one of the most important functions of the game. I don't think they've invested sufficiently in it and it's the cause of a lot of bad decisions and bad outcomes.
 

froggyluv

Grandmaster Knight
NW
I dont understand the minimalist Auto-Calc? Why? Add a few more qualifiers like terrain, forecast, whatever...makes the game THAT much more interesting
 

Apocal

Grandmaster Knight
The attrition exchange theory is easily disproven. Battle of Cannae comes to mind. I mean, history is literally littered with battles that totally disprove it.
Sure, you can find any number of exceptional cases. Point is that military calculations don't exist to cover outliers and definitely do NOT say this:
What about terrain? Defender bonus? Troop ratio bonus/penalties? These are basic military calculations. If I have 100 soldiers vs 10 looters, I should NEVER lose a single person.
All the way from the incredibly simple (Lancester's linear law) to the extremely complex (Dupuy's quantified judgement model) say that even a grossly inferior force will occasionally inflict casualties on a superior force.

Why should their power be clamped?
So high-tier AI parties don't win everything, forever. To be clear, I don't agree with it but that's the reason the campaign devs did it.

May I refer you to the Battle of Cannae once more. A smaller force double enveloped a larger force and totally decimated it due to superior leadership.

Can this happen in Bannerlords? No. Why not? Autocalc is one of the most important functions of the game. I don't think they've invested sufficiently in it and it's the cause of a lot of bad decisions and bad outcomes.
No, it cannot happen in Bannerlord, because even outside the autocalc the AI fights like AI and not like men who are "held in a terrible balance of fear." Fully envelop a force and it will still fight nearly as effectively; all the men require is enough space to swing their weapons and they'll start exchanging off. The only way to avoid it is to use a superiority in ranged weapons and distance or (somewhat) mobility to keep your men safe.

Autocalc is, at its core, meant for the AI on AI battles. To determine them quickly and without any performance impacts on the map.
 

JunKeteer

Veteran
Autocalc should always be inferior and cause more casualties than actually playing the battle; it's a shortcut/speed run similar with Total War.
You should have better outcome playing the battle vs simulation.
The only odd thing/issue is this tendency for the high-tier troop be the singular death all the time and that's just a fault of how they did their calculations.
And I don't think they tweaked the autocalc battles accordingly with all the armor changes; as it feels like a lot more situations come up where your better bet is to autocalc for a better result which shouldn't happen; as it's more numbers dependent.
 
Sure, you can find any number of exceptional cases. Point is that military calculations don't exist to cover outliers and definitely do NOT say this:

All the way from the incredibly simple (Lancester's linear law) to the extremely complex (Dupuy's quantified judgement model) say that even a grossly inferior force will occasionally inflict casualties on a superior force.

Sure. I'm not contesting that point. Key word being occasional. As I've repeatedly said, it's occurring too frequently.


So high-tier AI parties don't win everything, forever. To be clear, I don't agree with it but that's the reason the campaign devs did it.

I'm not here to support points I disagree with. I'm not sure why you feel obliged to support something you disagree with.


No, it cannot happen in Bannerlord, because even outside the autocalc the AI fights like AI and not like men who are "held in a terrible balance of fear." Fully envelop a force and it will still fight nearly as effectively; all the men require is enough space to swing their weapons and they'll start exchanging off. The only way to avoid it is to use a superiority in ranged weapons and distance or (somewhat) mobility to keep your men safe.

I disagree. When soldiers get surrounded, they get hit multiple times for every attack they make. It's literally impossible to defend against multiple attackers.

Autocalc is, at its core, meant for the AI on AI battles. To determine them quickly and without any performance impacts on the map.

Precisely! I'm convinced you're not altogether following the premise of my argument. So let me repeat:

Because the formula is so basic, it caused the Kuzait snowball. How did they fix it? Did they improve autocalc? No, they gave everyone more cavalry and endless recruits. When that didn't work, they just made the AI declare war against impossible odds.

A more elegant solution would have been to improve autocalc to factor in defender bonus and terrain bonus. Battanians fighting in forest should have the upper hand against Vladian cavalry. Sturgians fighting in snow should be able to catch up to and ambush Kuzaits.

Giving battanians more cavalry is just cheap. So yes, perhaps management screwed up with their bad decisions on how to fix snowballing but if autocalc was better, they wouldn't have been in a position to make the bad decision.

If it's still not clear, I am saying AI vs AI battles are flawed.
 

Apocal

Grandmaster Knight
I'm not here to support points I disagree with. I'm not sure why you feel obliged to support something you disagree with.
You asked why was it clamped and I gave you the same answer that was given to me.

At any rate, it isn't something that is difficult to mod out. Any can do it for themselves in dnSpy without outright breaking anything else (that I noticed), it just makes kingdoms at peace a long time into steamrollers once they do go to war unless the player jumps in personally to stop them.
I disagree. When soldiers get surrounded, they get hit multiple times for every attack they make. It's literally impossible to defend against multiple attackers.
They sure do. But a double envelopment doesn't mean every troop is defending from multiple attackers. In Bannerlord the troops (as a whole) do attrition exchange while being outflanked just fine.

Here's a video.

Formations don't do anything and tthe morale effects are far too weak to matter. That makes it impossible to replicate Cannae, except in the sense that Cannae was an annihilation battle. But annihilation battles are standard in Bannerlord so...
Precisely! I'm convinced you're not altogether following the premise of my argument. So let me repeat:

If it's still not clear, I am saying AI vs AI battles are flawed.
That's because you keep muddling it. If your point is about AI vs. AI, then why do you give a damn what factors go into things? A player will never notice unless they go out of their way to stare at two AI parties/armies battling. The one serious negative effect (snowballing) was fixed. There's no reason to give a **** that autocalc looters occasionally kill an AI's high-tier troop, at all.

If your problem is that they kill YOUR troops, then attack the problem at its source: why the hell does anyone feel the need to engage looters when they have tier 3+ troops at all? Why does the game even allow it? Just have the looters surrender if engaged with a moderately leveled player party and gradually despawn every single last group of looters off the map once the player hits Clan Tier 3.

"But how will I level my troops???"
By time you're a vassal, there should be ways (skills, fief, money, influence, relations, companions) that you can acquire a decent starter pack of troops, already leveled up, every single time you wipe and a sustainable flow of them to replace the losses you take while campaigning. It would eliminate the literally pointless grind of having to stroll obsessively around the map, stripping every village you come across of appropriate troop types, then finangling ways of pinning large groups of looters in place in order to level. It is only grind we'd be losing: nothing interesting will ever happen, there are no decisions to be made and it takes IRL time. Just give decent troops.

Then there would be absolutely ZERO reason to even have these ****ty looters after the very start of the game and we can stop seeing complaints made by people who feel the need to obsessively grind them to level up all 400+ of their T6 Fashion Plates before they feel comfortable doing anything else.

If your issue is that autocalc doesn't produce blowouts, then yeah, right there with you. It doesn't.
 
Top Bottom