Simulation Battles are Rigged

[ἀποκάλυψις]

Simulation battles are rigged with a pseudorandom number generator and seed to produce a deterministic outcome on every game load and run.

Saving before a battle and repeatedly running a simulated battle will reproduce the same results. The save must be made in the dialog screen where the game asks to attack or send troops. If the player sends troops, the results are pre determined on each run.

Every detail is repeatedly reproduced, including death, wounded, kills, etc. for all troop units.

The only way to break the seed is to exit the game completely and then load the game pre-battle. I suspect other mechanics in the game, like conversations, may use similar seeded methods too.

If you are save scumming you may need to exit the game before re-load.

 

[durbal]

Well, yeah. So? I quite like it. It means if I'm just playing with autosave only and my game crashes or I encounter a bug I don't get different results.

Why are you savescumming autoresolve battles anyway? Doesn't that kinda defeat the purpose?

 

[Diavolo]

That sounds like a good feature imo. Auto-calc has always been suicide so it's good. It simulates that your troops have no morale as you send them to their deaths.

 

[ἀποκάλυψις]

You're making opinions based on your own likes and dislikes.

I'm simply stating a fact of how the game rules are executed. Players may assume simulated battles are not deterministic and involve some chance.

 

[AndrewBones]

This is actually good if you want to prevent save scumming as much as possible. Although I agree it should be up to players choice in the end. I think XCOM did that - upon starting the game you could choose if you want to re-roll the seed on each load.

What is worse about the autocalc imo, is the whole way it is coded right now, see:

Reddit thread

The random seed being the same is the least of it's problems right now.

 

[Woopzilla]

That reddit thread makes it sound like it's basically the same as it was in Warband then? I remember auto-calc being a thing to avoid if you didn't want Swadian Knights dying to basic bandits.

Maybe they'll update it, but we have no idea since there's no complete list of planned features or changes. We'll just have to hope that they feel the need to adjust it if enough people make noise about it. Because there are swarms of bandits in the late game that can threaten your Caravans and villagers, and I don't feel like fighting so many battles vs small stacks, but also don't want to lose my tier 5 and 6 troops.

 

[AndrewBones]

That reddit thread makes it sound like it's basically the same as it was in Warband then? I remember auto-calc being a thing to avoid if you didn't want Swadian Knights dying to basic bandits.

I never bothered to find out how the calculations were done in Warband so I can't say if it is exactly the same, but it seems like the result is the same.

The key takeaway is that it pits troops 1v1 and decides the chances to die based only on tiers and HP. The result can be that any T1 unit has a 12% chance to wound or kill a max tier 6 unit, assuming the T6 unit has 100hp. If it has more due to perks, the difference would probably not be significant as HP cap seems to be at 110hp anyway (all perks that give +hp).

This makes no sense of course as it does not consider the rest of the party composition (archers?)
But it does not even make sense for the 1v1 scenarios - if say, a T6 unit, like Vlandian Banner Knight, fights one single Looter. In a completely isolated environment, I'd bet that the Looter simply dies to one lance stab every time, while the Knight is not in danger at all. Maybe in slight discomfort after being hit by a rock into his leg or arm, but certainly not dead or incapacitated.

 

[ἀποκάλυψις]

There are two issues and both come back on the other like the ouroboros.

First, never let players make bad assumptions. Players assume games, including minigames, involve some element of chance. Most enlightened people are not fatalistic enough to believe in predetermination... even if the real universe might be laughing at them.

Second, never make players choose when there is no choice. The message on that Attack or Send Troops screen should be "you must intervene to try and save this hopeless situation" ... and also "try to escape" if there is a chance. Remember that dialog for looters to surrender? It's just bad design.

My final thought, I'm agnostic about the "fun" of saves vs not saves. What I will say, however, is I am wholly against frustrating players who paid to be entertained. A game is not a beautiful princess, it's a prostitute. It should behave like a beautiful prostitute.

 

[Rconner]

I stopped using it, I just go into battle, hit F6, and go do something else while my guys wipe everything out.

 

[Viik]

There are two issues and both come back on the other like the ouroboros.

First, never let players make bad assumptions. Players assume games, including minigames, involve some element of chance. Most enlightened people, are not fatalistic enough to believe in predetermination... even if the real universe might be laughing at them.

Second, never make players choose when there is no choice. The message on that Attack or Send Troops screen should be "you must intervene to try and save this hopeless situation" ... and also "try to escape" if there is a chance. Remember that dialog for looters to surrender? It's just bad design.

My final thought, I'm agnostic about the "fun" of saves vs not saves. What I will say, however, is I am wholly against frustrating players who paid to be entertained. A game is not a beautiful princess, it's a prostitute. It should behave like a beautiful prostitute.

The chance is still there. Basically the way it works is that you have a stream of random numbers, these numbers are generated by a specific "seed". The seed is generated at the beginning of your campaign, for example, and stored in the save, the random stream can be stored too. Every action that requires some random number takes it from the stream. This is how it's done in pretty much any professional game for a simple reason that it allows to test and fix bugs much easier. It allows to replicate some event using players save file which otherwise might be impossible to do.

 

[bigbalzzz]

to all the spaghetti heads saying this is good to prevent save scums.... its not.....
-----------------
ignore the save-load factor
this means the game decided that one looter will kill 5 tier 6 units without rng
meaning that the last auto sim update was made to punish people
meaning the tactic skill is useless
it was their hope to stop factions snowballing and safely farming looters
but this also makes mostly recruit armies stronger then ever before

and thats is the main issue from this
you can just spam recruit
before even when skipping you had to carefully build your army and farm level it for days
now any army is good as long as it bigger
as tier and type are less relevant and auto simulated battles

but i dont like aotu simulated battles?
THE AI VS AI WILL ALWAYS BE SIMULATED
and thats why this patch was the worse

the game is in early access naaaa naaa naaa

I know
and rants like this tell the devs what direction we hope that full game go to

 

[ἀποκάλυψις]

The chance is still there. Basically the way it works is that you have a stream of random numbers, these numbers are generated by a specific "seed". The seed is generated at the beginning of your campaign, for example, and stored in the save, the random stream can be stored too. Every action that requires some random number takes it from the stream. This is how it's done in pretty much any professional game for a simple reason that it allows to test and fix bugs much easier. It allows to replicate some event using players save file which otherwise might be impossible to do.

um yeah. im familiar with pseudorandom. that's what makes it deterministic.

the salient point is that players are acting to make optimal decisions. you know... "the critical path"

you dont want your strategy game to be a "maze"

ive read a few naive comments about the equality of players and npc lords too, as if such a thing could exist.

 

[Bannerman Man]

Seems to me they will expand upon the auto resolve formula in time. If you look at the tactics skill tree there are perks that give cavalry advantage against archers and spearmen advantage against cavalry. Currently the auto resolve doesn't take into account the unit type, so if they want to make these perks operational they will have to factor in those things at some point.

 

[bigbalzzz]

Seems to me they will expand upon the auto resolve formula in time. If you look at the tactics skill tree there are perks that give cavalry advantage against archers and spearmen advantage against cavalry. Currently the auto resolve doesn't take into account the unit type, so if they want to make these perks operational they will have to factor in those things at some point.

thats why i am aginst balance updates like the change to caravans and simulated battles before perks fully function
because its impossible to balance a system and test it WITHOUT ACTUALLY HAVING THE SYSTEM

 

[Pejot]

Seems to me they will expand upon the auto resolve formula in time. If you look at the tactics skill tree there are perks that give cavalry advantage against archers and spearmen advantage against cavalry. Currently the auto resolve doesn't take into account the unit type, so if they want to make these perks operational they will have to factor in those things at some point.

Probably current autoresolve is just a placeholder.
Simple calculation means better performance. Imagine that instead of this You have 500 lines of code determining every factor in battle and there are around 50-100 battles simultaneously at any given time. Your CPU melts together with motherboard.

 

[bigbalzzz]

Probably current autoresolve is just a placeholder.
Simple calculation means better performance. Imagine that instead of this You have 500 lines of code determining every factor in battle and there are around 50-100 battles simultaneously at any given time. Your CPU melts together with motherboard.

atm is an 4 line code
the change they made was to let even peasants be able to kill any unit
before this change chance of kill was your damage vs the other unit hp

 

[ἀποκάλυψις]

Seems to me they will expand upon the auto resolve formula in time. If you look at the tactics skill tree there are perks that give cavalry advantage against archers and spearmen advantage against cavalry. Currently the auto resolve doesn't take into account the unit type, so if they want to make these perks operational they will have to factor in those things at some point.

Lancaster attrition models would agree with ignoring outside factors in sufficiently sized battles... so long as you used square of relative combat power.

 

[Bannerman Man]

Probably current autoresolve is just a placeholder.
Simple calculation means better performance. Imagine that instead of this You have 500 lines of code determining every factor in battle and there are around 50-100 battles simultaneously at any given time. Your CPU melts together with motherboard.

True. In my opinion it's good enough for the time being.

 

[Pejot]

atm is an 4 line code
the change they made was to let even peasants be able to kill any unit
before this change chance of kill was your damage vs the other unit hp

I know I look at those things a lot. Previous was advantageous to NPC/player with high tier troops and was one of the factors causing snowballing cause peasant armies had no chance.

 

[bigbalzzz]

True. In my opinion it's good enough for the time being.

i can write a better code in 5 mins
tw can a code that is better place holder x100 in 10 mins
the code is super simple and bad i am not over reacting