Feature request: Spell-esque feedback for weapons

[kuniqs]

For spells we have:
- spell success rate
- power, which translates to expected accuracy and damage

For weapons, all we have is weapon speed.

My proposition is to add a 'Current weapon accuracy' and 'Current weapon damage' bars, similar to the MR bar and spell power bar: They don't tell exactly how much resistance/damage you have, but make it easier to guess how good you are at both of those things.
For weapons, it's kinda hard to guess how much exactly you get from Fighting/Weapon skill/Stat bonuses etc.

So, for each +3 points of maximum possible accuracy/ +5 points of maximum possible damage, the respective bar gets an + to it. For negative accuracy (I guess it's possible), the bar gets -. The maximum accuracy/damage takes all skill/stat bonuses and slaying/enchantment bonuses into account.

So, for an executioner's axe (base 20 damage IIRC), with + 5 strength/dexterity bonus and +10 weapon/fighting skill bonus, the Current weapon damage bar looks like this: '+++++++........'

EDIT: I take back adding slaying/enchantment into account.
EDIT2: I'm more in favour now to show numbers instead of a bar. If we have numbers for weapon speed, AC and EV, why not for damage and accuracy?

[duvessa]

I'm worried about how you would deal with brands, aux attacks, etc. but I'm all for the concept.

[kuniqs]

Roll the damage 100 times & take the maximum.
Seriously, if someone wants info about brands/auxillary attacks, he has the wiki. But the formula for base max damage is kinda... hard to calculate on flight.
So, the formula shouldn't take brands etc. into account.

[damiac]

Why not take brands into account? The MR bar takes all my MR+ egos into account.

Something like this might make some decisions a lot easier, like when I have a really good falchion vs a plain old scimitar, or something along those lines.

[Giant Orange Brainbow Dash]

While monster AC varies significantly, base damage is much more reliable to read than resistible damage. Heck, just think of Holy Wrath. That'd nearly double the readout, but would be completely inaccurate for non-demonic and undead enemies.

[johlstei]

Why not take brands into account? The MR bar takes all my MR+ egos into account.

Something like this might make some decisions a lot easier, like when I have a really good falchion vs a plain old scimitar, or something along those lines.

Brands are complicated. Would the bar show damage per aut or damage per swing? Some brands do less damage but trigger more often and some are proportional to base damage while others are fixed, so this would make a big difference in the display. Distortion doesn't always do damage when it's triggered, I guess you'd have to have to figure out the expected damage. Of course this is still an issue even without brands, but they are an aggravating factor and I think it's more clear if they are not included, or displayed separately in some way. (It helps that really niche but high damage brands like orc/dragon slaying are gone but it's definitely an issue for holy wrath and pain, both of which are ignored by large portions of certain branches.)

I definitely like the idea, I worry about it being misleading though if not implemented carefully.

[kuniqs]

Why not take brands into account? The MR bar takes all my MR+ egos into account.

Something like this might make some decisions a lot easier, like when I have a really good falchion vs a plain old scimitar, or something along those lines.

- Brands are already well-documented about what they do. IMHO it's impossible to guess your adjusted base accuracy & damage with a weapon without writing a script. Not to mention that the damage/accuracy formula changes once in a while.

- MR works the same for every monster - it's easy to calculate % chance of resisting/scoring a hex. With brands, we would have to recalculate the damage for every monster type in the 'x' screen or post inaccurate information. If we're going this far, we could calculate avg. damage and %toHit for that monster, with current weapon. But then, that would reward juggling weapons in the middle of combat to find the one true king of the munchkins.

[kuniqs]

1) Should armor and shields (encumbrance) affect displayed accuracy value? I think yes.
2) Should the value be displayed as a bar, or just a number? Ac/Ev are already displayed as numbers, just as action/weapon speed. So, why not?
3) Should the values be displayed for weapons only, or also for unarmed combat/transmutations?
4) I'd like to implement my idea, but don't have much of a clue how to begin. I know Lua & C. Should I download the source and compile it with my changes? Should I just write a script and post it somewhere? Can someone point me to source files that deal with calculating damage and accuracy?

[xentronium]

- MR works the same for every monster - it's easy to calculate % chance of resisting/scoring a hex. With brands, we would have to recalculate the damage for every monster type in the 'x' screen or post inaccurate information. If we're going this far, we could calculate avg. damage and %toHit for that monster, with current weapon. But then, that would reward juggling weapons in the middle of combat to find the one true king of the munchkins.

You know how diablo 2 gives you "%chance to hit X" where X is the last monster you attacked...
and "%chance to be hit by Y" where Y is the last monster to attack you.

Feedback like that would be nice in crawl.

[kuniqs]

Diablo2 has very simple attack resolution mechanic, a function of your attack value & character level VS. monster's defense value & HD.

Crawl is a can of worms with circumstances (shallow water, umbra & halo...), weapon enchantments, god bonuses (Beogh), status effects (slaying, berserk etc.) which add up in non-trivial ways. I guess it is possible to calculate an average accuracy vs average ev and call it %toHit, but it wouldn't do the player much good.
Add to this the tedium - hit the monster, Shift-@ to check the percentage every damn fight.

EDIT:
Forgot to mention that "%toHit" has been a long-time laughing stock in Diablo2 community. The damn thing has never been updated by Blizzard and never had shown the correct value anyway

[ion_frigate]

Regarding showing the damage of brands: why not just display the average brand damage (against non-resistant monsters) in a different color of + on the same line? That should be pretty intuitive, showing that this damage is of a different type, and therefore may work differently (for resistance, AC, etc) or even not apply at all, while still giving a good idea of overall damage.

EDIT:
You could even use different colors for each damage type, roughly corresponding with the displayed color of beams of that element in console (magenta for distortion). Brands that only trigger some of the time would show the average brand damage expected per swing - that's what matters, since the goal is to get an idea of relative damage.

[kuniqs]

Good idea, but a bit tricky:
- My proposal is to show the *maximum* raw damage and accuracy, to make it easier for players to choose weapons (just like ev/ac differences make it easier to judge armor)
- What is the average damage of chaos brand? Distortion brand? Elec brand activates 1/3 times - do we show average damage or average damage/3?

Brands are either flat bonus damage (so you can check the wiki if you're curious), or %increase (which is intuitive as it gets). But showing the 'easy' brands' damage is not a bad idea.
Like, flaming executioner's axe damage: 20 (+5), where +5 is displayed in red.

[ion_frigate]

Well, distortion can be calculated - it's only a few cases, and it's the same for all weapons. As for sometimes-triggering brands, I edited my post for what to do about that - show the average damage per swing. Chaos could also be calculated, though admittedly its non-damage effects wouldn't be considered (same with distortion and antimagic). Venom's a weird one, since it does a ton of damage but over time.

Actually, a better metric might be the 1/(average number of turns to kill/banish/shaft/etc a 100-hp monster). For most weapons that'll be directly proportional to damage, but this would correctly account for both damage over time (venom) and instakills (distortion, chaos).

[kuniqs]

Let's take it further:
- average damage dealt by spells
- average damage blocked by ac
- dodging percentage against currently fought monster
- etc.
I don't really see the point of showing those averages, except minmaxing.

BTW. Seriously now. Since we're showing % with hexes, could it be possible to show % to hit when aiming spells and ranged attacks?

[mps]

I just wanted to say, re: the suggestion here that one can (and imo should) go farther. The player should be able to bring up a screen that shows a histogram of total damage dealt in a customizable number of simulated melee attacks against various monsters in LoS, paged if necessary. The histogram would break it down into damage dealt on hits, misses, and effects that make the damage number moot, e.g. banishment from distortion brand. In the other direction, it would supply histograms showing the damage distribution produced by the melee attacks of monsters against the player. Since everything would be simulated using the same combat code that handles damage resolution, there would be no question of its accuracy or keeping it up to date.

I suspect this would have somewhat scandalous results, showing wild distributions and so on, but I think it would be valuable to have this kind of information readily available.

[crate]

The crawl devs explicitly don't want to give the player some of these numbers. There's no way that a non-wizmode fsim will be implemented without a major design philosophy change first.

[mps]

I suppose someone should run simulations and put a digest of this kind of information out there. It's contrary to the ideals of clarity and reducing the need for spoilers not to make it available, though. I'm surprised to hear there's philosophical opposition.

On the other hand, if your damage calculations involve adding one random number between 1 and 75, one number between 1 and 26, and one number between 1 and 24 with a 68% chance of no damage, I can see why you'd be hesitant about putting the results out there for all to see...

[dpeg]

crate is right; let me explain (I am one of the vocal advocates of number reluctance). [All of this is my private opinion, I have never suggested something like this to anyone, because I think it is still too complicated.]

First, there is a divide between new players and advanced players (arguments in favour of more numbers are generally, but not always, made by advanced players).
Any numbers we show within the game will be interpreted as crucial to playing well. So players will want to scrutinise and to optimise them. Not only can this detract from fun for some people, it also means that the threshold for new players is higher: they have to understand more about the game (or at least, justifiably believe they have to).
Example: manual skilling is the default mode because it means that players can give Crawl a go without having to bother with skills, aptitudes and the m-screen.

Second, for all numerical feedback that we do give, we have to think about how it can be processed by players.
Example: Sometimes there is demand to show the crucial numbers for monsters (AC, EV, HD etc.). However, this is misleading without additional context: player and monster numbers are not symmetric. Hence xv gives bars instead of numbers for AC and EV.

Third: The desire to be able to answer the question "Is weapon A better than weapon B?" quickly is understandable. I think it's a pity that Crawl cannot follow Brogue in this (Brogue is extremely explicit, giving hit and damage information for all weapons and all enemies using only turns and percentage, i.e. no HP values at all). However, with combat as it is in Crawl, weapon feedback is a much harder problem.

To give a positive example: trunk now shows percentages for hexes (both player and monster spells). I think that's a great new feature and was well worth showing more numbers. Here is why: (1) the information is given in percentages, i.e. there is no confusion about context; (2) it suffices to give this information locally (i.e. in xv-ing a monster, or when targeting a hex spell on a monster); (3) it is the most crucial information you could want about hexes. If you compare this with damage spells, say, then you'll see that hexes are pretty simple in this regard. In my opinion, we have a similar situation with weapons, only more complicated.


That's why the current weapon feedback restricts to base damage and weapon speed. Note that, not so long ago, more precise information about current weapon speed has been added -- a crucial piece of information. If you are about to give more detailed weapon feedback, what I could imagine is a new screen listing all carried weapons and some of their properties. (I think single screen is relevant because weapon choice is always about picking one out of some, so you always want to compare. This is different with hexes, say.) Here is some information that might be useful (and enough):

(a) weapon (e.g. "+7 eveningstar of freezing")
(b) type (e.g. Maces & Flails)
(c) aptitude
(d) current skill
(e) skill needed for mindelay
(f) base damage
(g) turns* needed on average to destroy weak dummy (no resistances)
(h) turns* needed on average to destroy strong dummy

*Instead of turns, aut is probbably better.

And this might easily be too complicated. I cannot come up with something easier to convey "average damage done".
Note that (c)-(e) could be somehow replaced by "how much xp needed to reach mindelay", but I am not sure how much that's an improvement.

[mps]

re: dpeg, I get what you're saying about putting a lot of numbers in front of people.

I surely don't quality as an advanced player, but as someone who was recently a complete beginner, I would say that the auto-skill default really only provides the illusion that you don't need to know what you're doing with skills to get anywhere. I've generally had terrible results using auto-skill myself and it's not at all obvious how one should skill manually without advice from other players, largely because the formulas involved are ridiculously complicated and fairly hard to find. Add to this that often adding skill does not seem to have any effect, so you don't even get the possibility of gradient descent heuristics (spell success rates are particularly terrible in this regard).

[Perhaps getting too far afield here, but if the intent was to make it easy for beginners to interact with the skill system, the way to do it is to enable players to create build orders that automatically schedule how your experience will be allocated as you play and set the various class/weapon combinations to have a default one that's good. I'm sure this cuts against dcss philosophy as well, but there it is.]

But if numbers are hard to understand, pictures aren't. A histogram is easy for anyone with enough curiosity to play a game like this to grasp. Especially when what you're showing is the output of a complicated formula, the likes of which you often see people puzzling over in forum discussions, it's far better than a number for understanding what's going on in the game.

[kuniqs]

(a) weapon (e.g. "+7 eveningstar of freezing")
(b) aptitude
(c) current skill
(d) skill needed for mindelay
(e) adjusted base accuracy (either number or visual representation)
(f) adjusted base damage
(g) hits needed on average to destroy the most threatening foe you encountered so far

FTFY

That would be similar to spell selection screen, and much better than manually picking weapons with i-letter.

About numbers making game harder for new players - magic users have more confusing info about their spells and are still considered easier for newbies.

You don't need to know how EV works to know 10 EV is average, 15 EV is good and 20+ EV recommended for late game.

Spells roll NdD damage, where N is quite large for most spells, so damage is consistent on average. Weapons have 1dBase_damage (and enchantments).

If I'm going to hit foes with a giant spiked club and deal no damage to them, or miss half the time, I'd like to know at least how much more worth it is to use this club instead of a demon whip.

EDIT:

Hell, the stats alone are confusing new players. Common sense tells you more strength = more damage. Turns out strength is as good as your armor.

[crate]

Common sense tells you more strength = more damage.

It does do this, but for probably historical reasons the effect is small. There doesn't seem to be a desire to change stats to, you know, actually do things, so I suspect this will remain the case.

[grisamentum]

Here is some information that might be useful (and enough):

(a) weapon (e.g. "+7 eveningstar of freezing")
(b) type (e.g. Maces & Flails)
(c) aptitude
(d) current skill
(e) skill needed for mindelay
(f) base damage
(g) turns* needed on average to destroy weak dummy (no resistances)
(h) turns* needed on average to destroy strong dummy

*Instead of turns, aut is probbably better.

And this might easily be too complicated. I cannot come up with something easier to convey "average damage done".
Note that (c)-(e) could be somehow replaced by "how much xp needed to reach mindelay", but I am not sure how much that's an improvement.

My problem has always been that without fully spoilered out-of-game spreadsheet, there's no way to answer a simple questions like "How much damage am I giving up by switching from a 2h to a 1h+shield?"

At a bare minimum it would be nice to just have two sentences:

"With your skills and stats, this weapon does Y base damage on a hit."
"With your skills and stats, it will take you X aut to (swing)/(fire)/(throw) this weapon."

To-hit chance is a different problem, admittedly. Maybe that should be displayed on monster examination based on the weapon the player is holding. But that seems like a different problem than answering something like "ideal dps in a vacuum" - which is at least a start.