Balancing the factors in melee damage

[danr]

There's been talk of a need to rebalance how stats and skills affect melee damage so I thought I'd provide a starting point for the analysis.

I used my spreadsheet to calculate average damage per turn (AD/T) vs. Sigmund for a +0 giant spiked club for a character with maxed out stats (35) and skills (27). Then I dropped each of those (stats to 10, skills to 0) in turn to see how big an effect the loss had on each.

The max AD/T was 35.4.
Dropping Dex 35->10 had almost no effect (unsurprisingly), -3%
Dropping Str 35->10: -26%
Dropping Fighting 27->0: -38%
Dropping M&F 27->0: -77%

Note though that the stats have other effects to be considered for balancing purposes, such as EV, carrying capacity, etc.

Now, the balancing discussion, on which I have no views at this point, should consider:
- how does the current balancing affect gameplay?
- is that effect on gameplay the way it should be?

To attempt a first response to those questions:

- From what I gather, Informed players tend to ignore Str unless the other stats do nothing for them, esp. as the player really only gets to control about 5 points of stat increases from level ups, but they may make further decisions re randarts with stat increases, so that is only a fraction of the 25 point Str difference I tested here.

- Fighting skill is perceived as only being good for boosting HP, though as seen here going from 0 skill to 27 skill can boost damage output by nearly 40%, which is not something to sneeze at.

- Weapon skill is by far the biggest factor, but the perception is that the main benefit from weapon skill is in reducing delay. The boost to damage and accuracy is perceived as being negligible, so informed players tend to turn off a weapon skill once they reach minimum delay for that weapon type. Doing a little extra calculation shows that in this example the speed decrease, without any decrease in damage due to loss of skill, would account for a 61% reduction in AD/T, so it is true that attack delay is by far the dominant effect of weapon skill.

Let's look at it another way - Instead of a maxed-out character, suppose that a starting HEPa finds a triple sword on D:1. His/her starting AD/T would be 2.5, let's look at boosting different things:

+12 Str (what the character can probably achieve with levelups, artifacts and potions): Increases AD/T to 3.1
+12 Dex: Also increases AD/T to 3.1
+25 Fighting: Increase AD/T to 5.4
+24 Long blades: Increase AD/T to 14.0

Again, the weapon skill is by far the dominant factor in maxing out weapon damage.

Now, I have no opinions on what SHOULD matter more - Str, fighting, weapon skill, or whatever. I just thought some number would help devs and others discuss rebalancing these.

[szanth]

If, in fact, they get around to making stats actually worth something beyond what they are now, then I'd like to have a little more control over which stats my character gains as they level up. Maybe instead of every three levels (I don't actually know) then every other level or every two levels.

[danr]

Let's try to stay on topic for at least one post. This is about balancing the factors in melee damage, not player control over stat increases.

[Moose]

You know, this may sound wacky, but why not just get rid of stats all together? They just seem to add one more annoying factor that you have to manage among others.

We already have a highly diverse skill system for managing your character.

You want to do more damage? Train a weapon school.
You want more HP? Train fighting!
You want more MP? Train spellcasting!
You want more EV? Train dodging!
etc ...

Since we're at a point in the game where skills have negligible effect on your character (except INT), why not take this opportunity to get rid of them?

Since INT is so vital to spellslingers, I think the bonuses should just be folded into the spellcasting skill.

The benefits are pretty drastic:

1) Eliminates factors you need to manage to grow your character, and lets you focus on your skills instead!
2) Gets rid of a vast swath of potions and rings!!
3) Simplifies artifact bonuses and mutation bonuses


Some might argue that stats differentiate races. That's true, but so do aptitudes, and I think the aptitude system is doing a very very good job at keeping the races distinct. Just try creating a stealthy troll!

[danr]

I can see that this would simplify the game, but is that good? The original rogue was much simpler than DCSS, but is it a better game?

Stats are wound up in so many aspects of the game and removing them would take all kinds of rebalancing. This is not going to happen.

Can we stay on topic please? If you have a radical game changing idea to discuss, please start a new thread.

This thread is about the notion of rebalancing melee damage factors, which is already being considered by the devs.

[zasvid]

I think that this discussion would be even more interesting if research like yours had higher granularity (and perhaps considered a few vastly different weapons) to catch any oddities in the system (like, for example, illustrate how the damage gains are reduced once you hit the minimum delay) and meaningfully compare the relationships between e.g. strength and dexterity (because if, as you say, Str is generally ignored by informed players, perhaps it should have a greater impact on damage than dexterity has or something).

[ohmi]

If delay is the most important component for overall damage determination, then it would seem to make sense to make delay depend more on Str and Dex, less on weapon skill. Perhaps allowing "extra" Str/Dex reduce/improve delay up to some amount short of the best/minimum that a high skill could provide.

Does each weapon or weapon class have its own delay formula already?

[danr]

Interesting idea. This could be tied to str-weighting - a dex-weighted weapon would benefit more significantly from dex, and a str-weighted weapon would benefit more significantly from Str.

What if speed was a function of dex * dex weight + 0.5* str * str weight, and damage was a function of str * str weight + 0.5 * dex * dex weight?

Using this, Str should also affect your speed with a heavy weapon, but it should not have much effect on speed with a light weapon (and it would also have less effect on damage with a light weapon?)

[szanth]

I'm very pro- making stats that weigh with weapons actually matter more. Until I looked on the forums and asked people on the IRC, I had no idea; I mean, it said in the game that certain weapons were best used by strength or dex guys, and I just had to trust that it meant that it was a significant thing. Turns out, nope.

So yes, I'm all for that, danr.

[ohmi]

Yes, that's where I was going but not with any specific kind of formula in mind yet. Hands could be considered too, as well as weapon size (i.e. dagger vs sabre). This kind of thinking leads to less/no reliance on weapon class: the delay/damage formulas and weighting of various factors for dagger could look much different than the formulas for whips or great maces.

List of possible factors:
Str
Dex
Hands
Size
Class
damage type
species <- why shouldn't Spriggans be faster than Nagas with daggers but slower with maces? Perhaps reuse aptitude instead.
Weapon class skill
Fighting
brand

Hmm, even Int could be factored in, related to damage type or class: "scientific" fighting weapons vs. "brute force" type weapons.

YA approach to balancing is to add some factors directly to the base damage of a weapon (assuming this isn't already being done). Perhaps aptitude should provide a % increase/decrease to the base damage of weapons by class or even type (i.e., merfolk like spears, but can really dish it out with tridents, particularly when wielded two-handed and STR > 15).

I can see the reluctance of the devs to introduce yet more complexities into an already pretty opaque set of calculations.

Rebalancing for the purposes of improving the player stat choice experience:

Another way to look at it is what is cool for players when they get a chance to improve a stat? Or, to put it another way, since one gets to choose a stat increase a few times a game, what makes it interesting to choose anything other than Int? For Int you can sometimes see the results in terms of hunger reduction and spell success, correct? The goal of rebalancing damage factors then might mean that Str and Dex increases should have similar levels of visible benefits for melee delay/damage calculations for the weapon/armour currently being worn. A description/extra detail page similar to all the others at stat choice time could present the various alternatives:

  Code:

     Melee        Armour        Spellcasting   <etc>...
STR      +               ++                .
INT       .                .                 +             
DEX      ++              .                  .


This very sad chart shows that at that point in time, increasing STR will help melee and armour; INT spellcasting only; Dex melee a lot (wielding a short sword for example and current Dex is low), armour not at all (wearing crystal plate), and spellcasting not at all ( ya off-topic: why not have schools/spells that incorporate Dex into the success formula, like Blade Hands?)

[danr]

I think if anything, the number of factors should be reduced. If there are 8 factors that affect melee damage, each one individually has to have a pretty small effect on the end result, and players have to work on all of them to maximise damage (depending on how it's done). Or if you let each factor have a bigger effect, the cumulative effect is the possibility of creating characters that do insane amounts of damage if they focus all their efforts on that goal.

I guess one way to limit this is to make the different factors be additive rather than multiply by each other.

[ohmi]

It's not necessarily a good thing for players to min/max, or to think they get the best out of a game only if they do so.
Having more sophisticated formulas would allow playstyle to vary depending on the series of choices the players make: species, background, stat, melee style, spells or not, etc. One-size-fits-all formulas that use just a few factors/variables led to sameness and generalizations that become dogma.

That doesn't mean your wrong - I'm suggesting rebalancing by having more formulas, not necessarily making one complex one.
For example, the delay/damage formula for Spriggans could be weighted toward Dex and give less weight for hands ; Ogres would get more benefit from higher Str and do relatively better with 1.5 and 2 hands.

That's crudely done now, but as you pointed out, it makes for much sameness given that weapon delay makes the most difference and that is determined mainly by skill.

Currently I like the idea that aptitude helps reduce delay/increase damage more directly than just by skill level, and that each weapon has stat weights -- most weapons might have Int weight = 0 but a couple might have a reasonably high Int weight, like. Staffs.

What about the idea of making base damage be a bit more variable? Aptitude again could be involved, which is nice since it itself doesn't vary. Perhaps a weapon could end up having base damage = current+ (aptitude/2)*.1, i.e., .8x to 1.2x, given that aptitudes are between -4 and +4.

What effect would that have making melee damage less reliant on delay?

[galehar]

In my opinion, those are the most important things to change:

• Increase the impact weapon skill has on damage so that it's still worth to increase it after having reached minimum delay.
• Increase the effect of dex and str on fighting. And by consequence, the importance of strength weighting.

Given the importance of weapon delay on damage output, maybe we'll have to find a way to increase the effect of weapon skill on damage after minimum delay is reached. So that the global damage output keep raising linearly once you've passed minimum delay.

For stats, I think their effect should be multiplicative instead of additive. But it also means that using a dex oriented weapon with low dex should reduce the damage.

Also, I don't think str needs to have a bigger impact on damage than dex. I like the strength weighting of weapons, let's try to make it meaningful. If dex seems more useful for now, it's because EV is overpowered, but let's not elaborate on that and try to stay on topic.

I haven't really looked at the numbers yet, so I can't provide more specific suggestions.

[szanth]

*shrugs* You could make dex and strength relatively equal by having them both affect their own respective weightedness, while dex affects EV and str affects AC. Recalculate and balance as needed, of course.

[galehar]

*shrugs* You could make dex and strength relatively equal by having them both affect their own respective weightedness, while dex affects EV and str affects AC. Recalculate and balance as needed, of course.

Well, that's exactly what's happening already, right? Except for the fact that str doesn't directly affect AC, but is necessary to wear heavy armour. Are you suggesting a more direct effect of str on AC? We're off-topic again, but it seems we can't really talk about the effect of stats on fighting without also considering their effect on the whole game.

[szanth]

Yeah I agree, it's a pretty important subject and it affects so many things.

But yes, I'm suggesting that, like dex affecting EV directly (based on dodging), str should affect AC directly - perhaps based on fighting or something. Strength and fighting would need to be recalculated to prevent it from being overpowered, of course.

[dpeg]

I'm suggesting that, like dex affecting EV directly (based on dodging), str should affect AC directly - perhaps based on fighting or something. Strength and fighting would need to be recalculated to prevent it from being overpowered, of course.

I am against this. It is good if stats work in different way.
My biggest beef is with "perhaps based on Fighting". We should try to reduce the number of subtle interdependencies: instead of having all of weapon, skill, weapon enchantment, Dex, Str, Fighting skill, Weapon skill (plus some more) affect damage and accuracy, make it clearer who does what.

[szanth]

*shrugs* I'm fine with strength just doing damage and carry weight (in addition to strength-weight weapons damage). The suggestion that it also affect AC was because I was worried that in response to dex being, currently, a method by which one can have a small increase in damage (according to danr's spreadsheet), dex would override strength and the strength-weighted weapons would seem less optimal in comparison, because dex would be a pseudo-uberstat.

[danr]

We should try to reduce the number of subtle interdependencies: instead of having all of weapon, skill, weapon enchantment, Dex, Str, Fighting skill, Weapon skill (plus some more) affect damage and accuracy, make it clearer who does what.

I agree with this. In creating my spreadsheet I found so many interdependencies, many of which did little. The main effect was to complicate things and put me through another few calculations to get an extra 2% to damage.

I can see the appeal as dev of trying to create many clever and subtle interdependencies and that this would make it feel like you've created a subtle and complex game, but I think KISS should prevail. This is not meant to be critical, I just see it as a case of getting excited about improving a detail while not seeing the big picture as much. It happens a tiny bit at a time without any critical point where someone says "Let's make melee damage really complicated!". But that's just a guess about how things got the way they are.

Some ideas that might simplify things:

- Nethack / AD&D had a simpler model IIRC - Str. of certain levels just gave a +X to damage. In Crawl, Str and Dex could be treated the same way, with the base bonus for each multiplied by the str / dex weight.

- Take fighting out of it, just call it "Battle seasoned" or something and just have it affect HP and other background things like EV penalties. Or get rid of fighting altogether.

- Cap the max damage for a weapon type at the same skill level as minimum delay, perhaps with a slight increase so that this is not a nerf. Make it so that using a dagger simply can't train short blades beyond level 10. I guess short blades as a class then would just never go above 16 or whatever it is for a sabre.

- Increase the Str. effect on damage. To counterbalance this, raise the minimum delay for large weapons from 7.

- Weapon skill should only affect speed.

That's just brainstorming, I'm not advocating any of these, just putting them out there for consideration. If you don't like these ideas I can make up a dozen more.

[ohmi]

- Weapon skill should only affect speed.

and accuracy?

- Cap the max damage for a weapon type at the same skill level as minimum delay, perhaps with a slight increase so that this is not a nerf. Make it so that using a dagger simply can't train short blades beyond level 10. I guess short blades as a class then would just never go above 16 or whatever it is for a sabre.

Would the inverse be true also?
Should weapons with large delays/large base damage not deliver bonus Str/Dex-weighted damage until trained up to a minimum level?

Is the Fighting skill supposed to be equivalent in some way to Spellcasting in that a higher skill improves other weapon skills? How does it work out for Spellcasting across the board?

[danr]

Personally I think that both fighting and spellcasting, as skills that overlap with other skills, add too much complexity, and I would rather see them go.

I'd just as soon eliminate those skills and have MP and HP gains based purely on race.

[Stormlock]

I'd like Dex to have a much bigger effect on accuracy than it currently does. It's silly that a spriggan with a dagger is hardly any more accurate than an ogre with a club. (Never mind the number of times a sleeping monster has 'dodged' being stabbed by a character with 30+ dex.) Aside from being thematically wrong, it is very unintuitive.

[dpeg]

1. Spellcasting is much better off than Fighting: it provides MP (just like Invocations do) and helps with spell hunger. It also helps with spell success and/or power, which is not necessary and rather another example for those interdependencies, which were added for realism but not gameplay (I suspect).
2. I don't think that HP should come only from species. We definitely want a pure casting human to have significantly less total health than a pure fighting human. However, that directly points towards possible solutions. Scrap the Fighting skill and hand out maxHP for xp spent in melee skills. (There are some provisions to make here, but I think you understand what I mean.) Another option would be to tie maxHP to stats, with Str giving full account, Dex half of it, and Int none. But the other approach is easier to set up as is.
3. Dex should matter for a move like sidestep. And it should directly matter with stabs, I suggested removing the Stabbing skill.

[danr]

I really like the idea of tieing HP to stats, mostly Str. That would make Str more important for melee, and would provide a much less tedious way for casters to boost their HP than grinding up fighting vs. plants and rats (which is essentially victory dancing)

Currently focusing on INT is a no-brainer for any caster. If Str was required to boost HP, then casters suddenly have an interesting choice to make: boost INT for spellcasting, or boost STR for carrying capacity and HP.

I do agree that between fighting and spellcasting, fighting would make sense to get rid of more than spellcasting.

[acvar]

I do agree that between fighting and spellcasting, fighting would make sense to get rid of more than spellcasting.

They both really need to go since they are the main impediment to getting a good replacement for victory dancing. If you can get rid of fighting and replace it with weapons skill advancement there is no reason you can not do the same for spellcasting.