Unit Weighted Damage Allocation v1.0

Lord Thane

UWDA is now live! If you haven't been keeping up with discussion on the topic, this update is meant to distribute damage in a more reasonable manner - so your fodder units will take the brunt of damages before your more elite units.

I've setup an initial weight value for every race based on some factors - cost to produce, maintenance, att/def, basically how valuable the unit is. Most races have a pretty standard setup of high-value units and low-value "fodder", so it's pretty easy to figure what needs a lower weight (allocate less damage), or higher weight (allocate more damage). We can tweak these values to be more or less, but for simplicity I started with a basic 1-2-3 assignment.

Some important notes:

The math doesn't care what mix of units you throw into battle, sending less fodder units won't give your elites any advantage - the incoming damage is gonna hit something.
Overkill situations where weight-3 units take more damage than they can absorb gets pass on to the next weight, and so on. We call this Trickle-Down Fuckyounomics.
Each unit type also has a “defense per unit” value. When damage is allocated to a unit type, you divide by that defense to figure out how many individuals die.
A small random factor (±5–10%) means you never get exactly the same casualty numbers every battle, keeping things feeling dynamic.

Weights I've initially assigned, and we can tweak depending on what the race experts think:
Dwarf Shieldbearer: 3
Dwarf Hammer wheilder: 2
Dwarf Axe Man: 2
Dwarf Runner: 2
Dwarf Light Crossbowman: 2
Dwarf Heavy Crossbowman: 1

Elf Mage: 1
Elf Swordman: 3
Elf Lanceman: 3
Elf Caragous: 2
Elf Archer: 3
Elf Elite Archer: 1.5

Gnome Catapult: 1
Gnome Infantry: 2
Gnome Militia: 3
Gnome Rider: 2
Gnome Rock thrower: 3
Gnome Balista: 1

Human Knight: 1
Human Infantry: 3
Human Pikeman: 3
Human Heavy Infantry: 2.5
Human Archer: 2
Human Mounted Archers: 1

Orc Shadow Warrior: 1
Orc Rusher: 3
Orc Slother: 3
Orc Wolf Master: 2
Orc Slinger: 2.5
Orc Axe Thrower: 1.5

Undead Dark Knight: 1
Undead Skeleton Warrior: 3
Undead Wraith Pikeman: 3
Undead Abomination: 2
Undead Phantom Archer: 3
Undead Wraith Rider: 1

We can tweak these values to be more or less dramatic, we can use decimals to make the differences less, we can use 1 | 1.5 | 2, as a weighting system. If there's one unit you think should get totally clapped we could give it a 4 (need community input for such drastic change). These initial starting points are just me shooting from the hip.

Real World Test #1

1:1 human v human, 80% defense via castle rule.

Before UWDA:
We managed to kill : 637 Knight 800 Infantry 800 Pikeman 695 Heavy Infantry 622 Archer 800 Mounted Archers
But we lost : 646 Knight 440 Infantry 452 Pikeman 503 Heavy Infantry 610 Archer 630 Mounted Archers
Result: Win, both sides pretty much losing around same numbers of troops 500-600 of each unit.

After UWDA:
We managed to kill : 381 Knight 800 Infantry 800 Pikeman 800 Heavy Infantry 800 Archer 700 Mounted Archers
But we lost : 212 Knight 1000 Infantry 1000 Pikeman 605 Heavy Infantry 1000 Archer 343 Mounted Archers
Result: Win, both sides getting zeroed out for the lowest weighted 'fodder' units, elite units have better survivability.

Technical analysis

Total casualties inflicted
Before: 6,635 total kills (Attacker 4,354 + Defender 2,281)
After: 8,441 total kills (Attacker 4,281 + Defender 4,160)
+1,806 more units fell once we switched to weighted damage

Balance of losses (kill differential)
Before: Defender suffered 1,073 more losses than Attacker (4,354 − 3,281)
After: Defender suffered only 121 more losses (4,281 − 4,160)
*swings 952 points back toward parity

Per‐unit shifts in total casualties
(Kills + Losses summed for both sides)
Knight: 1,283 → 593 (−54%)
Infantry: 1,240 → 1 800 (+45%)
Pikeman: 1,252 → 1 800 (+44%)
Heavy Infantry: 1,198 → 1,405 (+17%)
Archer: 1,232 → 1,800 (+46%)
Mounted Archers: 1,430 → 1,043 (−27%)

What this tells us
*Closer match. The walloping the defender took before (–1 073) is almost gone (–121), so neither side dominates purely by having weak units soak the hits.
*Elite units live longer. see far fewer total casualties, while mid and lower weighted tiers pick up the slack.
*Pulls attrition off elites, spreads it more evenly through the army, and turns a one-sided rout into a much tighter, more tactically interesting fight.
*Mid-tiers now soak up more of the damage - the weighted system shifts attrition “down the line.”
*Light skirmishers see a boost in both kills and losses, reflecting their middling weight.

MithicYort

Lord Thane Nice. Is the "test" account live so we can sim battles with the new format?

My only criticism is that defending a castle doesn't seem to offer enough of a benefit in any of these scenarios. I know its a fantasy game and all, but historically castles were incredibly hard to capture. Any chance or plan to buff castles a bit more on defense?

Leo

Lord Thane

https://github.com/cdelatorref/thardferr-issues/issues/22

In that BR; 0 Pikeman 0 Archer dead which is unexpected. Normally due to being weakest def. should die more. Strategy was set to Quick Retreat on that battle. Wiping out 800k orc units (80% of 1m i believe) except the SW with 1m human units seems excessive.

Leo

Lord Thane

Thanks for the hardwork.

Though i still think the weight (allocation percentage) to be decided according to the percentage of the unit among the army. And A pre-determined weight is more likely to bring-out complications in code and BR results. There will be many scenarios to test it out. Besides there will be dealing with community opinions about what a unit weight should be.

And about "The incoming damage is gonna hit something" -> I like the wasted damage idea more. Range units shoot and miss 50% +- 5-10% (The logic behind the range damage 50%)
In the perspective of melee units they will be assigned to some area in battle and have limited mobility to inflict all their capacity.

Weight according to percentage of the unit in army
Damage distributed according to weight. Any excess damage is wasted.

It also sounds much easier coding.
Hard part is coding the unit bonuses and strategy bonus/debuffs. Hell, that is headache even to try to figure it out.

Lord Thane

MithicYort the UWDA isn't meant to buff castle defense, it's just a way of shifting the damage in a more realistic way - in previous battles if you sent 1000 rock throwers (a guy with a rock) and 1000 catapults (a team of 5 men running a siege engine) you would end up getting something like 600 casualties to your rock throwers and 600 casualties to your catapults. I had added some entropy to make it look slightly less stupid, but it really was the same kind of result, except maybe you lose 576 rock throwers and 602 catapults lol

With UWDA, in that same kind of scenario you would now lose 1000 rock throwers and maybe 200 catapults, or whatever.

Yeah getting busted open on defense sucks, you can lose up to 80% of your army in a blink if the attacker is strong enough, which sucks. We will need some more community input on if we should make defending easier and what those changes might look like.

If you think about it, the castles are actually quite buffed as it is. If you had 10 castles and stationed 10% of your army in each, then shouldn't you only get 10% of your total force on defense? Yet we give at least 70% of your troops to defend. If we changed that to be so realistic then we'd need a counter-balance, like get rid of "protection" lol, so now it's not such a hot idea to send 100% of your garrison out to war if anyone can mosey in on your zero-defense castles... That could be a interesting and fun game mode.

MithicYort

Lord Thane Right, I understand the UWDA isn't meant to buff castles. I just have this thought that castles should offer some benefit to the defenders damage output and defensive score. Archers for example might be stationed on walls making them harder to hit, give them (at least some of them) immunity or massively reduced melee damage taken and offering them a height advantage which would increase their range, visibility, and velocity of the arrow itself.

There were often booby traps so that advancing troops would get slaughtered at certain points by raining rocks, boiling tar, ect. Even with a 20-30% reduction to force the castle ought to provide unique benefits to the defender.

I also don't think this is unique to before or after UWDA but rather a separate issue.

Not to underscore the UWDA damage weighting which I think is an awesome addition!

Xliest

Lord Thane Yeah getting busted open on defense sucks, you can lose up to 80% of your army in a blink if the attacker is strong enough, which sucks. We will need some more community input on if we should make defending easier and what those changes might look like.

In my opinion would be easier to defend/lower casualties as you get lower than the opponent no change if the attacking opponent is your size or below.

Jules

Looks good. I'll have a lot more input later after I get a good feel next round. It's a good starting point and those battles look a lot better.

trupheus

I’m really confused by shieldbearers being a 1 and curious how that will interact with the strategy

Lord Thane

MithicYort maybe we should do a new Tech that increases ranged attack for defender multiplied by castles or something - like arrowslits.

trupheus yeah that's a good catch, definitely would want to send a mass of the damage onto shieldbearers so they can do the job they're built for. Maybe something like this makes more sense:

Dwarf Shieldbearer: 3
Dwarf Hammer wheilder: 2
Dwarf Axe Man: 2
Dwarf Runner: 2
Dwarf Light Crossbowman: 2
Dwarf Heavy Crossbowman: 1

MithicYort

Lord Thane That tech sounds sweet! That actually brings me to "fortification" Being the final tech in the tree, 5% doesn't feel particularly generous to me. I have been wondering if it's something that has been tested and actually works as described and if a more generous bonus is indicated. Something that would maybe need to be voted on after solid testing? Any way, just more rambling of thoughts.

Lord Thane

MithicYort I haven't tested techs, other than reviewing the codebase to make sure it's getting run through the battle calcs. I pretty much just focused on magic, strats, and attacks (cash, grass or ass combo).

Hopefully someone else was testing those lol... Maybe if I have time in the final mini-test round next week I will.

MithicYort

Lord Thane I've got fortification but from my side I have no idea if its functioning properly. Been hit twice I think since I got it but the 5% is low enough and I don't know what was sent.

Lord Thane

Leo 5 seconds before the tick, I wonder if the attack went through as units were getting wiped from the tick or something. Gonna have to pull up the detailed logs to see if those units actually went to battle or not. Or maybe a bug with QR and UWDA 😭

Lord Thane

Leo the UWDA actually is affected by how many units of the various troops you put into the mix. The % of units X their weight. That way the exact army mix determines who soaks up how much hits - and no damage vanishes. Theres a wobble built in for variability.

Here's a simple example:

10 Infantry (weight 3, defense 10 each)
10 Archers (weight 2, defense 10 each)
5 Knights (weight 1, defense 10 each)

180 points of incoming attack damage to allocate.

Unit / Count / Weight / Total Defense /Share of Damage / Damage Taken / Units Killed
Infantry 10 / 3 / 100 / 60% / 108 / 10
Archers 10 / 2 / 100 / 30% / 54 / 5
Knights 5 / 1 / 50 / 10% / 18 / 1

Total weighted value = (10×3) + (10×2) + (5×1) = 55

Infantry’s share = (30/55) = around 55%, Archers = (20/55) = around 36%, Knights = (5/55) = around 9% (with a little fluctuation from randomness)

Damage is divided proportionally, then divided by each unit’s defense to calculate kills.

Overkill spill may kick in if any unit type dies off before using up its full share, redistributing leftover damage.

Leo

Lord Thane

Can you clarify below:

Is "10 defense each" figuratively speaking in the example?

Lets say there is 180 damage to be distributed (ruling out randomness affect to simplify)
108 damage kills 10 infantry but 8 damage is passed to next unit (lets say archer) ? Then 54+8 means 6 archers die and 2 damage passes to knight and 18+2 = 2 knights die . Is that correct?

Xliest

Let’s not throw only orc range into the fire first I suggest slothers get a 3 and rushers 3 also then maybe WM, are a 1.5 or 2 while expencive they have a mount and are expected to be front lines. There’s an argument to be made they should have 1:4 atc training too then add slingers and axthrowers as 2s

Xliest

Also elves EA needs a 1.5 and train time should come into play with this at least at the 3vs2 units.

For humans I would swap heavy’s and knights mounted units kinda more forward and they have charging as a strategy…

In general I like the idea however will this lead to battles where 0 of one unit class dies? I personally don’t like that part of it. I think there should be some amount of death especially in larger battles. And some troops should come home from the front lines as it’s morally uplifting to a kingdom… just like when front line troops come home IRL.

Lord Thane

Xliest agree with you on the elves, and partial agree on the orcs - I increased sloth/rush/wolf-m, slightly reduced slinger (although an important ranged unit they're dirt cheap and highly expendable - same defense as axe throws at much higher cost), reduced axe thrower.

Orc Shadow Warrior: 1
Orc Rusher: 3
Orc Slother: 3
Orc Wolf Master: 2
Orc Slinger: 2.5
Orc Axe Thrower: 1.5

trupheus

This one’s been rolling around in the brain.

The most simple form of this is simple distribution of damage based on army composition. Unit makes up X% of army so it takes X% of damage. That damage divided by unit defense = unit deaths. This prevents overkill. Army has to get refactored after each phase of combat.

I really like the idea of being able to assign weights to units as a player, although the coding of this would probably be a MF’er to factor it into the different generic and racial strategies. Also factoring in how those all interact with each other.

Jules

trupheus

Ya I had been mulling empowering the player with weights as well after I read your suggestion of bringing SB to 3. I like SB at 3 90% of the time, the other 10% I wish it was a 1. It really just depends what people are trying to do composition and strategy wise at a given time

Xliest

Lord Thane this would be exactly what I was thinking…

Lord Thane

Leo I guess it would have been helpful to put how much incoming damage I was using in that mock example lol. In the example I was using 180 attack damage, and the defender had 250 defense points, so the example can demonstrate how the weights and army composition effects the allocation... It’s designed to keep damage distribution fair, even if some unit types are low in count but high in weight - so everything takes a share without over-penalizing any one group.

And the example isn't pushing the overkill spillover to the next bucket, so yes in your example with the overkill added in there would be an additional casualty for the archers. Give or take with randomness wobble. And no overspill added to knights since the archers would absorb the full overspill from the infantry.