Krikkitone wrote:

I think that is fine, at least until they fix technology so that it is impossible to be more than ~2 techs ahead of anyone, you can't make that too big of an advantage.

No, it is not fine for ranged units to deal capped damage at the damage floor (which is hit at -4 relative CS, and is 75% of the damage at +0) to units of the very same era, let alone eras ahead. This renders fortifications and terrain completely redundant - what is the point of taking the high ground and manning the walls, if the enemy archers will deal the same damage to your ancient (or classical or medieval or early modern) units whether they are standing in an ideal defensive position, or down in a river with said archers firing from walls on the hills.

If you want to make it so that you can't get too big an advantage early on, then the formula should just scale more gently. But it is precisely because of the current formula that Archers beat Warriors on an open field, that they beat Chariots on an open field - both of which are units of the ancient era meant to counter them outside of high ground scenarios / siege battles.

It is because of the current formula that upgrading your units makes no sense at all, because why build your costly late classical Praetorian Guards and defend your well-placed and fortified cities if they will take the same amount of damage from each shot of those Archers as far more primitive and cheaper units would? And melee units will never be as good at dealing damage as ranged ones, because they have 1 range, are constrained by space and terrain, can be blocked by enemies and suffer retaliation damage - so what you care about with your melee units is not their damage dealt - its their damage taken. Which doesn't differ at all in most scenarios whether you recruit warriors, swordsmen, praetorian guards or halberdiers.

The formula needs granularity, it needs for each and every point of difference in Combat Strength to count. Currently this is not the case to say the least, and even within the range that does matter (a quite narrow -4 to +16), there is many dead zones, some as wide as 7 points (+8 to +15), where extra points will do exactly nothing most of the time because it will take the same amount of hits to kill enemy units. This on top of +1 CS being worth a different amount under the current formula, depending on if its from +0 to +1, or from say +15 to +16, or +16 to +17.

Currently:

Going from +0 to +1 relative CS will provide you with +12.5% damage dealt and -10% damage taken.
Going from +9 to +10 relative CS will provide you with +6.3% damage dealt and -0% (yes, no difference) damage taken.
Going from +13 to +14 relative CS will provide you with +20% damage dealt and -0% (anything above +4 does that) damage taken.

I hope you see the issue.

As far as line infantry vs peasants is concerned, having more CS is actually a bad thing (lol), because if you had less than +16 CS then the peasant would not die on attacking, and it would continue blocking space, making it so that other peasants cannot cross through its tile, making it so that the line infantry is attacked by just one unit if it holds a chokepoint, and not seven of them in one turn.

I think the main issue here is that the CS diff to damage formula is too complicated and multi-segmented so it is not scale-independent. If you plot it it 'almost' fits an exponential function but it flattens out for low (negative values). If they changed it to a simple 15*Exp(0.1 * CS Diff) - 28*Exp(0.075 * CS Diff) that would give roughly the same values for most of the range but would also work for very large/very small differences. I plot both of these on the graph below alongside the action values in game right now. With that an Archer would have a vanishing chance to damage a Tank (equation would give a range of 0.08-0.56.

PopsiclePrime wrote:

I am not sure which game you all are playing. I have 266 hours in Humankind and my Nubian Archers NEVER beat melee or chariots on an open field. EVER. Nor should they. Likewise, archers, longbowmen, crossbowmen have NEVER EVER killed a single one of my tanks. EVER. If any of you have captured video of these things happening I would love to watch it, please upload it or post a link.

Can someone please tell me what the current combat/damage formula actually is? I see a lot of talking about it, but I don't see the actual formula. I have some minor, very minor concerns about how much damage some units may or may not be doing to other units and I would like to tweak the combat strength values of some of the units myself when the mod tools come out. So I am looking for the actual formula to plug into a spreadsheet to test possible outcomes. Should lessen the number of test scenarios I will have to run.

And just FYI, there are different types of units in the game because they have different roles to play. Each one of those roles is a counter to one of the others. This is the way things are in real life as well. And the reason they are that way in real life is because that is what works. Yes the developer could give the Norse a purple and pink unicorn that has an indirect, ranged, and counter attack in every situation so that you only have to ever build one unit type/class, but that would not be fun, and it is not how the real world ever worked, how it works today, or how it will work in the future. Yes this is just a game, but it is a game based on the real world; i.e. it is not a Fantasy or SCI-FI game where you can just make new things up. Indirect fire weapons should NEVER take return fire, that is what horses and APCs are for; to run around behind the front lines and kill the archers and/or siege weapons. Yes polearm class units should do more damage to cavalry because that is what they were invented/developed to do.  Yes melee units should take damage when they attack because they are conducting close quarters hand to hand combat, you get hurt when you do that stuff. But when archers unleash a volley on advancing infantry they do NOT and should NOT take return damage, period! There is no scenario where infantry stop advancing, throw their swords, stones, or anything else at archers 100 to 200 yards away, and then continue advancing. Expecting to be able to sit idly by while siege units pound your position is unrealistic; as is wanting your defending units to magically return fire on siege weapons that are literally a hundred yards or more away. Siege units were invented to counter fortifications, archers on ramparts, burning oil falling from towers, etc. Yes the people inside the fortifications took damage, some were killed, but siege units were NOT invented to kill people. They were invented to destroy fortifications and other defenses to make breaching easier and safer for the infantry. 

While playing Humankind you should know that there are different classes of units, that each is the counter to one of the others, and you should make sure you build enough of each kind such that you have a well rounded military that can not only attack efficiently and effectively, but can defend just as well.

The current formula seems to be arbitrarily assigned damage dealt and taken values for every single combat strength difference, then an arbitrary damage floor at -4 or below (5-25), followed by lower floors at certain far larger differences (5-20 starts at a difference of -30 CS or so), and a damage cap of 100-100 at 16+ CS or above

You don't see it provided anywhere because damage does not follow an actual formula at the moment (though Arkatreides's in the post above is a close fit for much of the range), rather it is all arbitrary values that maybe used to be the output of one in the past before it got butchered it seems like. As for the actual values, see below per the wiki:


As you can see, the impact of each point of Combat Strength of difference is not even. Sometimes you can get +20% Damage Dealt per point, sometimes you can get +6%, and most of the time you get -0% damage taken because its capped at -4 already. Suffice to say, this is not an easy environment for balancing units against one another because a change aimed at balancing unit A against unit B has a completely disproportional impact on its performance against units C and D, and that's before the (flat!) bonuses to CS come into play making it all a jumbled mess.

As for research on the anachronistic formula lowering the damage bound from 5-25 to first 5-20, then 5-15, then 5-10, then 5-5, that has not been fully done, but the first of those thresholds seems to happen around a difference of -30 CS like I mentioned (hence why longbows or soviet-buffed ancient era units still work just fine versus tanks), and the last at a difference of ~60 CS, therefore it is hardly relevant for regular play.

And yes, Archers do beat an equal amount of Warriors or an equal industry cost in Chariots, on an open field, and it only gets worse from there. Let us math out both examples.

Each Chariot is 180 Industry. Each Archer is 90 industry. Therefore for every Chariot you can have 2 Archers, and population is not yet a constraint in the Ancient era.


Archers win even if the chariot has the first round. And Archers don't even have to take hills or any other defensible locations, or hide in forests on deployment so that the chariot cannot attack them first round because it can't see them.

The precise odds on an open field neutral terrain (which archers will never take) are 63.16% chance to win for archers, 25.03% chance to draw (both sides dead) and a 11.81% chance to win for chariot. This is if the chariot has the first turn and can see and hence attack the archers. If it does not, then the chance for equal industry cost of archers to win are ~99%.

Archers vs Warriors - both units are equal in industry and pop cost. A battle looks like this:

Yes, if Archers stand in place stupidly then Warriors will win. But Archers can kite Warriors by running backwards, use chokepoints to block more than 1-2 Warriors from crossing through an area, and they can focus fire on one warrior with mutliple archers at a time, whereas Warriors cannot do so due to space constraints. Archers can also take favourable locations and force the warriors to approach, and even doing just one of the above things will make them win easily.

Therefore Archers reliably win 1v1 vs Warriors regardless of who takes the turn first. And more numbers on each side favour the Archers heavily over Warriors that are constrained by range, space, terrain and line of sight, and in such a situation can be kited as well.

You do not need Nubian ones. And I do not appreciate your insults, because yes I am well aware units *should* have separate roles and counter one another. They just don't, at the moment, because of how the formula works. It is not an issue with the relative unit stats (although there are a ton of units that are just strictly better than other units - check british emblematic unit vs austro-hungarian) - It is an issue with the formula.

PopsiclePrime wrote:

What am I looking at here? Can you explain this chart? I know it isn't yours. But I don't understand what it is representing. If the Attacker is 40 CS and the defender is 30 CS do I subtract 30 from 40 leaving +10 and that falls in the +4 and above category so the RNG will result in a number between 5 and 25. Is that correct?  And then if the attacker is 30 and the defender is 40 then 30 - 40 = -10 so the RNG will result in a number between 43 and 58.

Who is the attacker and who is the defender is not relevant (unless there is no retaliation damage or for the purposes of defender bonus) - only the difference in CS matters, and each unit deals damage to the other based on that from its own perspective. Per the current CS formula, a 40 CS unit fighting a 30 CS unit will be dealing damage at [40-30], or +10 (43-58, or 50.5 on average), whereas the other unit will be dealing damage at [30-40], or -10 (which is the same as a -4, which is 5-25 or 15 on average).

PopsiclePrime wrote:

Sorry if I seem a bit to literal here, but I want to be certain I understand. The chart does not have +10 specifically, it has +4 and above. It does have -10 which has the min value of 43 and max of 58. But you say  "... +10 (43-58, or 50.5 on average), ...".  So I am a little confused. 

If your unit has 34 CS and the enemy has 30 CS, you will be damaging them at +4 CS (25-35), and they will be damaging you at -4 CS (5-25).
If your unit has 46 CS and the enemy has 30 CS, you will be damaging them at +16 CS, and they will be damaging you at -16 CS (which is the same as -4 CS, i.e 5-25).
If your unit has 60 CS and the enemy unit has 30 CS, you will be damaging them at +30 CS (which is the same as +16 CS), and they will be damaging you at -30 CS (which is the same as -4 CS ie 5-25).

I think the values in the screenshot are for damage taken not dealt hence the confusion.

shasho wrote:
@PopsiclePrime It's Defender - Attacker = Strength Difference, so if the defender is 30 CS and the attacker is 40 CS the attacker will do between 43 and 58 dmg with a median of 50.5. And I would assume it works the same way for Attacker - Defender, so the defender (+10) should fall into 4+ and above and take 5 - 25 dmg.

Shame it doesn't have the mean damage though, as I believe the way rolls are weighted changes for combat strength differences above 4.

The weighing does not change. You only have the range crop down from 5-25 between -3 and -30 to first 5-20 (which is around 30 to 40 CS of difference) then to 5-15 (~40 to 50 CS diff) then to 5-10 (~50 to 60 CS difference) and at last to 5-5 (Above ~60 CS of difference). It is not even gradual (like cropping the range down by 1 for every 2 CS of difference or such, but rather threshold based).

shasho wrote:

@Crunbum Really? I feel like since they introduced the "Unit Anachronism" factor I've been noticing outdated and underpowered units rolling single digits more often, but that might just be an observation bias. I can't fathom why they thought that +4 CS ought to be the lowest bracket for damage rolls.

Yes, those thresholds (5-20, 5-15, 5-10, 5-5) are the aforementioned anachronism factor. And I can't believe it either, but its just one of the many faults of the current formula - why do you think I'd write such an essay on it, otherwise? :)

The formula definitely needs to be simplified, but

1. Str Difference->Damage is good

2. a minimum Damage is good too (to prevent Technology from being Too important)

I'd Suggest

start at 15 minimum 30 maximum

for defender higher differences -1 to minimum, -2 to maximum (stopping at 10 higher, so the minimum is 5-10).... so Average Damage= 22.5-1.5/Str diff

fo defender lower differences +4 to minimum, +5 to maximum (only capping out at 100)... so at 14 its 71-100, and at 22 its 99-100..... so Average Damage=22.5+4.5/Str Diff

So the slope is slightly gentler at the extremes and stronger in the middle, and have it be a perfectly even distribution from Max to Min

And it would be MUCH simpler

The rest can be solved by reducing the Combat Strength of units that don't get retaliated against/increasing what units get retaliated against.

Krikkitone wrote:

The formula definitely needs to be simplified, but

1. Str Difference->Damage is good

2. a minimum Damage is good too (to prevent Technology from being Too important)

I'd Suggest

start at 15 minimum 30 maximum

for defender higher differences -1 to minimum, -2 to maximum (stopping at 10 higher, so the minimum is 5-10).... so 22.5-1.5/Str diff

fo defender lower differences +4 to minimum, +5 to maximum (only capping out at 100)... so at 14 its 71-100, and at 22 its 99-100..... so 22.5+4.5/Str Diff

So the slope is slightly gentler at the extremes and stronger in the middle, and have it be a perfectly even distribution from Max to Min

The rest can be solved by reducing the Combat Strength of units that don't get retaliated against. (or maybe allowing range units to not retaliate against melee or such)

That just has the exact same issues as the present formula and is in fact quite similar in design - obsolete units rule the game, terrain, fortifications etc are worthless except in edge cases, and melee units are not worthwhile to say the least. A formula based on an addition/subtraction of damage based on absolute difference between CS simply does not work - it cannot work because its impact is uneven and because its range of outcomes is by necessity kept to extremely narrow bounds, that you can exceed easily and then profit off of it.

Formulas based on relative differences / ratios in stats have worked quite well in 4X games and not just, and for years now, as have percentage-based modifiers rather than flat ones. There is no need to reinvent the wheel here, not when it breeds issues that break the game at its core. I doubt the modding tools will allow for adjusting the formulas which the game uses, and if those are flawed in the first place then there is hardly a way to make meaningful changes to the game with mere number tweaks.

shasho wrote:

I don't like the idea of a multiplicative combat strength system, I think the current one is fine principally. Firstly, because a multiplicative system has its own problem that any flat value modifiers vary in worth depending on the base that is being modified (and if you elect not to use flat values, then it pretty much works like the current system). And secondly, because it has major readability problems (currently a difference of 1 CS means the same thing in all cases, vs. having to do mental math to figure out the strength ratio between two units).

I think the current system will work fine if the damage cap scales well beyond -4 CS to the point that all attacks do 1 damage or none at all.

"Currently a difference of 1 CS means the same thing in all cases, vs. having to do mental math to figure out the strength ratio between two units."

That's the gist of the problem - it does not.

Depending on the matchup, 1 point of CS can mean +20% damage dealt and -0% damage taken. It can mean +12% damage dealt and -6% damage taken. It can mean no change at all. A difference of 1 CS (or even as much as 14 CS) can have literally no impact. It is all matchup dependant. Against Chariots, your Archers might take 20% less damage or even more because of having 1 more CS (more if we are talking hits-to-kill), and against some other unit it could be +3% damage dealt with no change to damage taken, or nothing at all.

On the other hand, a ratio-based formula does exactly what you are praising the current one for - it makes each percentage point worth exactly as much in boost, and yes with percentage modifiers. It makes each and every point matter, and always to the same, predictable and intuitive degree. Percentage modifiers in a fully granular ratio based formula are way more intuitive to understand than flat ones stapled on top of wildly differing damage ranges per point of difference based on an absolute difference in a stat that can cap out and never decrease/increase past a certain point.

In a ratio-based formula, if you see a unit with 20 CS, and a unit with 30 CS, you know how effective they will be, and how much damage they will do to one another. You can just divide one by the other and multiply by four to instantly know how many hits it will kill the other in, doing so in 1 second in your mind, or multiply the ratio by itself to figure out how many units the stronger one will be worth in an "in your face" scenario with everything retaliating. If you receive a bonus of 10%? See, your unit is that much more powerful.

Now, try doing that with the current formula. Tell me, how many CS 20 Warriors can four CS 25 Swordsmen kill at a time? How much damage will a unit with 30 CS deal to a unit with 20 CS, and how much it will take from it, without glancing at the table? How would that change if the opponent had 25 CS instead? 45 CS? How exactly would it change if they got a +4 modifer from that hill, or a -3 modifier from crossing a river? How much value would you get from three soviet districts, and how much would you get from six? Why wouldn't the latter number be twice as good at the former? I hope I got my point across.

Here is the answers by the way:

Ratio formula (done in my head):

- Each CS 25 Swordsman is worth (25/20) ^ 2 = ~1.6 Warriors, so 4 Swordsmen are worth ~6.4 Warriors.

- A unit with 30 CS will deal 3/2 * 25 = ~37.5 average damage to a unit with 20 CS, and it will take 2/3 * 25 = ~16.7 average damage. If you want to do hits to kill, then its even easier as you can just do 3/2 * 4 = 6 hits for CS 20 to kill CS 30, or 2/3 * 4 = 2.67 hits for CS 30 to kill CS 20.

- If the opponent had 25 CS instead, then its 2.5/2 * 4 = 5 hits to get killed, and 2/2.5 * 4 = just over 3 hits to kill.

- If the opponent had 45 CS instead, then its 4.5/3 * 4 = 1.5 * 4 = 6 hits to kill, and 3/4.5 * 4 = 2/3 * 4 = 2.5 hits to die.

- If you got a +4 CS modifier from a hill (which would always be +25% in a percentage formula), then you would be dealing 25% more damage and taking 1/(1+0.25) = 80%, so 20% less.

- If you got a -3 CS from river crossing (or -20%), then you would be dealing 20% more damage and taking 1/(1+0.2), so 5/6 so 16.7% less.

- From three soviet districts you would get +3, or probably +6% (+2% each) if they were to be balanced to any degree which they can only be in a percentage formula because +1 is the smallest flat increment, and its equal to 2% for up to date CS 50 units. So you would be dealing 6% more damage and talking probably 5% less. From six districts you would be dealing 12% more damage and taking 10% less.

Current formula:

- Each CS 25 Swordsman is worth... looks up table... +5 damage and -4 damage taken.... looks up values... 33.5 damage dealt by swordsman, 15 damage dealt by warrior, so 33.5/15 so.. probably somewhere like 2.2, so 4 of them are probably worth nine warriors or so? Rinse and repeat from every scenario, from simplest to complex. 3 soviet districts to a 30 CS vs 20 CS is 50.5 -> 62.5, or 23.7% more damage, as well as 0% less damage taken because its already 15. 6 Soviet districts, however, is 50.5 -> 100, or 100% more damage, as well as again 0% less damage taken. Can't do the current formula in your head unless you have every entry in damage table memorized.