The Tavern

Problem:

New player in neighbor thread wrote:The -6 accuracy dissuaded me from using the bardiche.

-6 is irrelevant because players don't see monster EV as number, unspoiled players don't know anything about basic accuracy (15), bonus from stats, bonus from fighting, bonus from weapon skill etc.

Solution:
Remove accuracy description from weapons.

Better solution:
Remove weapon accuracy bonus/penalty, making all weapons the same.

The problem with accuracy is that it does potentially matter at times, but actually figuring out when is generally not worth the hassle.

For reference, to my understanding EV distribution for a 10 EV monster looks somewhat (but not exactly!) like this: http://anydice.com/program/93b9 (change Data to "At Most", this displays the chance to roll this number or less.)

So the increase in hit chance with accuracy (and thus increase in average damage) goes something like 333%, 210%, 71%, 53%, 35%, 30%, 26%, 23%, 18%, 13%, 10%, 8%, 6% and so on.

Removing accuracy bonuses/penalties will lead to making strong weapons better for characters not training the weapon skill, you need to take this into account.

The accuracy bonus/penalty as a number is as arbitrary as the base damage to be honest. Actually, base damage is the bigger offender because new players will always think the enchantment is a flat addition to the base damage (the usual problem of "what do you mean 10+3 does not equal 13+0"). If you want to do something with one number, do the same with the other, be it changing to a written description, hiding it entirely or making it actually meaningful without reading up on Crawl's damage formulas.

Well, the obvious problem is that you're given a contextless number. Of all the time the game could give you a number, accuracy is probably the least useful, since you have nothing to compare it against.

So, change accuracy to a random number of !, and then tell us how much damage our attacks do. Win win.

Better solution would be to display what the character's accuracy is, maybe on the @ printout. Then it's not a contextless number any more. This kind of feedback makes the game much more tactical then the current guess-at-what-is-the-better-weapon.

Chicken wrote:Better solution would be to display what the character's accuracy is, maybe on the @ printout. Then it's not a contextless number any more.

Okay, say your accuracy is 13. What does that tell you?

Ideally you might have a corresponding sentence in the manual to say what it is. But even seeing the number at least gives you an idea if other factors are influencing it besides the weapon.

If monster EV would be shown as a number as well...

Chicken wrote:Ideally you might have a corresponding sentence in the manual to say what it is. But even seeing the number at least gives you an idea if other factors are influencing it besides the weapon.

The manual will say something like "Helps you hit unpleasant things (but does not affect the damage you do when you hit)", so the number is still contextless.

I agree that the number would probably be a good thing, so as to see that fighting and weapon skill and armor penalties really do contribute. I just wanted to emphasize the point that it doesn't do much more than provide a measure to say one arrangement is better than another.

(it's maybe worth noting that the ordering is, in fact not actually a total order since there are several randomized components; a higher inconsistent value might be better against some opponents, while a smaller, more consistent value might be better against others)

(as an aside, I strongly suspect the wiki page for "To hit" has inaccuracies in listing whether various effects have a fixed or randomized value)

A lot of powerful weapons are noob traps early due to the negative accuracy number and high base delay meaning you're likely to (take a long time to do your strike) and then (still miss). [For god's sake don't use a GSC at 8 M&F on D:4, are you nuts?] So it definitely has an effect, but...this is one case where I think the current situation is fine...I'm not exactly sure what the number does, because I haven't looked it up, but it does give a nice idea.

Strange, no feedback from devs so far

The accuracy penalty is irrelevant and can be very confusing even for experienced players.
For instance, this is fsim for OgBe with 6 Fighting and 8 M&F.

Vs Orc Priest

+0 mace (starting weapon)

Code:: AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam Attacking: 5,0 | 16 | 64% | 3,2 | 100 | 1,00 | 3,2

+0 GC

Code:: AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam Attacking: 14,3 | 41 | 48% | 6,9 | 120 | 0,83 | 5,8

+0 GSC

Code:: AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam Attacking: 14,6 | 41 | 46% | 6,9 | 150 | 0,67 | 4,6

Vs Adder

+0 mace

Code:: AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam Attacking: 4,8 | 15 | 48% | 2,3 | 100 | 1,00 | 2,3

+0 GC

Code:: AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam Attacking: 14,2 | 38 | 24% | 3,5 | 120 | 0,83 | 2,9

+0 GSC

Code:: AvHitDam | MaxDam | Accuracy | AvDam | AvTime | AvSpeed | AvEffDam Attacking: 15,7 | 46 | 19% | 3,1 | 150 | 0,67 | 2,1

The game displays +3 accuracy displayed for mace, -6 for GC and -7 for GSC but as we can see GC is always the best weapon and even GSC is much better than mace vs Orc Priest.
When are those accuracy numbers useful?

Hurkyl wrote:
Chicken wrote:Better solution would be to display what the character's accuracy is, maybe on the @ printout. Then it's not a contextless number any more.

Okay, say your accuracy is 13. What does that tell you?

If you're playing an Ogre and have 13 accuracy with a flail, you may compare that number to the accuracy penalty on the GSC you just found (-7!) and decide that you probably need more M&F skill before switching to it as it would cut your accuracy in half.

Whereas if you saw on the @ screen that you had (arbitrary number incoming...) 50 accuracy, you could see that the GSC's -7 penalty is a relatively smaller fraction and be confident that you could still be accurate with it.

If players don't have enough information to make an informed decision, I don't think the solution is to give them even less information. It's useful to know your accuracy if you know what accuracy means, and it's imaginable that the game could tell players what accuracy means. Sure, the bcadrens of the world make incorrect assumptions about accuracy; that doesn't mean it needs to be completely hidden from all players forever.

If you're going to remove display of base accuracy, then you need to remove base accuracy altogether. Hidden properties are terrible.

genericpseudonym wrote:If you're playing an Ogre and have 13 accuracy with a flail, you may compare that number to the accuracy penalty on the GSC you just found (-7!) and decide that you probably need more M&F skill before switching to it as it would cut your accuracy in half.

Whereas if you saw on the @ screen that you had (arbitrary number incoming...) 50 accuracy, you could see that the GSC's -7 penalty is a relatively smaller fraction and be confident that you could still be accurate with it.

Why do you think it scales lineraly? Accuracy 40 vs accuracy 60 does not mean you hit twice as often. Even if accuracy worked as you describe it (or as it works in some other games), you would see huge difference in difference between the two weapons vs different monsters depending on their EV (sorry about using "difference: so many times in one sentence).
For example, vs monster with EV 1 you would get 40/(40+1) and 60/(60+1), that's 97% and 98%, almost irrelevant difference.
Vs EV 50 you would get 40/(40+50) and 60/(60+50), that's 44% and 54%, somewhat relevant, extra 22% hits.
Vs EV 200, you would get 40/(40+200) and 60/(60+200), that's 16% and 23%, very significant, almost 50% more hits.

It should be clear at this point that devs are not going to display EV as number. In this case it is better to remove accuracy description from weapons, all it does is confuses new and even experienced players. Nobody is going to calculate to-hit in their head, it's too difficult, use fsim if you care.

duvessa wrote:If you're going to remove display of base accuracy, then you need to remove base accuracy altogether. Hidden properties are terrible.

Yes, I agree, ideal solution would be to remove base accuracy altogether.

I don't see why we have it displayed when we don't have accuracy displayed for being in shallow water or rings of dexterity, for example. If we need to display some accuracy penalty for items, it should be shields and armour. Those depend on str, armour/shield skill and size while the weapon accuracy is static.

Following much the same logic by which the success chance of MR-checking attacks is now displayed, could we just display a % to hit chance for currently equipped weapon? Would that really be beyond the pale, at this point?

I think the problem is that there's a bit too much factors involved - with enchantments it's only spellpower and monster's MR, you don't need to recalculate that, with a weapons it's base, enchantment, slaying from items, might, monster's EV, possible agility... Fsim is pretty taxing on the server, is it not?

Sar wrote:I think the problem is that there's a bit too much factors involved - with enchantments it's only spellpower and monster's MR, you don't need to recalculate that, with a weapons it's base, enchantment, slaying from items, might, monster's EV, possible agility... Fsim is pretty taxing on the server, is it not?

It's not really Fsim though to show the to-hit for each particular enemy as and_into suggests. I mean, that number is calculated every time you make a combat roll!

A single combat roll gives you either a 100% chance to hit or a 0% chance. A to-hit (adjusted accuracy modifier) is just a number, who knows what it means? I mean, it might be useful to see how stats and skills increase it...

Sar wrote:I think the problem is that there's a bit too much factors involved - with enchantments it's only spellpower and monster's MR, you don't need to recalculate that, with a weapons it's base, enchantment, slaying from items, might, monster's EV, possible agility... Fsim is pretty taxing on the server, is it not?

For the accuracy roll just checking the average is fine, and fairly cheap computationally. The EV roll is a bit trickier, but still probably cheap enough to just brute-force; another way would be to just store the distribution for all realistic EV numbers in memory/file and use as a lookup table.

(Edit: no, actually checking the average is not enough, probably need to brute-force it, or do something tricky with math. Any statisticians in here?)

Sar wrote:A single combat roll gives you either a 100% chance to hit or a 0% chance. A to-hit (adjusted accuracy modifier) is just a number, who knows what it means? I mean, it might be useful to see how stats and skills increase it...

Basically, the player has a to_hit number, which is a base value, then modified by dex, fighting, weapon skill, weapon accuracy, slaying, armour penalty, and possibly other stuff (starving, vertigo, eyeballs mut). The hit roll is a random value up to the to_hit number. If hit roll >= monster EV then it's a hit.

Knowing the to_hit number and monster EV, it should be fairly trivial to convert this into a percentage for display. This is the easy part. But then, do you want to account for shield blocking and deflect missiles too? And possibly other things I am not thinking of right now.

This is just for weapons though. Presumably you'd want show the to-hit % for the primary weapon slot (melee or ranged). But if melee weapon or unarmed, you could possibly be interested in Throwing accuracy instead. And then there are spells and wands.

EDIT: on thinking it over, maybe interface isn't such a big problem. Magic and ranged can just show the to-hit% in the targetter, similar to how hexes work. For melee, this would be reasonable to show in xv I guess.

For melee, you could display in xv based on current status, skills, and weapon. You could also show in parentheses the percentage chance to hit after an attack in the messages. (Eg, "You hit the foo (42% to hit)" "You missed the foo (42% to hit)"

Note players receive some nonspecific feedback about this already ("barely missed," etc)

Sar wrote:I think the problem is that there's a bit too much factors involved - with enchantments it's only spellpower and monster's MR, you don't need to recalculate that, with a weapons it's base, enchantment, slaying from items, might, monster's EV, possible agility... Fsim is pretty taxing on the server, is it not?

It is implemented in CA fork so I believe technically there should not be any problems.

viewtopic.php?f=8&t=20089&p=271447#p271447

It has screenshots for spells but the same functionality was done for melee attacks also (not sure about ranged, haven't played)

Is base accuracy doing anything interesting? If not, it would be good to remove it, regardless of the interface issue.

That does close off some of the design space (e.g. extra-accurate or inaccurate weapons), but this seems like a pretty boring space - spriggan's knife and sure blade were removed for a reason.

EDIT: Knife of accuracy, not spriggans.

Spriggan's knife wasn't removed.

If weapon base accuracy was removed, probably making top end weapons hit more often, then perhaps top end weapon damage could be scaled down. That would of course make player damage feel less spiky. I imagine some players would like that and others would not.

BabyRage wrote:Spriggan's knife wasn't removed.

I believe the knife of accuracy is the artefact being referred to.

I almost never read weapon base accuracy and I don't think it's really meaningful for how it is done right now. It would be better if the game displayed something like "Your skill level allows you to use this weapon with very low/low/average/high/very high accuracy" and "its enchantment level raises/lowers its accuracy moderately/to very low/low/average/high/very high". The five levels assuming a target with 10 EV and each level being the ranges between 0-20-40-60-80-100% accuracy.

Basic accuracy can be made more meaningful by making the accuracy gap wider and allowing the weapon skill to close it at appropriate levels.

ximxim wrote:(Edit: no, actually checking the average is not enough, probably need to brute-force it, or do something tricky with math. Any statisticians in here?)

I think the brute force is pretty cheap here; it's a computer science problem to do it right, not a math problem. e.g. to get the odds for rolling a number on 7d6, you don't generate all 6^7 possibilities. Instead, you generate the odds for rolling each number on 6d6 and on 1d6, and add the distributions. (or even better, generate the odds for 4d6 and 3d6, and add them)

Maybe my memory is exaggerating, but there are a few enemies that have very high EV. Swapping to higher accuracy weapons, particularly in the early game with low weapon skills, improves killing chances. Mainly I'm thinking of adders. I'd often use short blades (or a spear) over M&F/LB/axes. A player that finds an early war axe, scimitar or flail might well never hit that adder.

Alright, so... how about this? Everything stays as it is now, except that when you view your weapon in the inventory, you can see a list of how accurate your weapon is against all the unique monster types you see on the screen (assuming default stats on each monster).

For example, the weapon display screen could look like this:

"Base accuracy: -6 Base damage: 18 Mindelay: 0.7
Press [whatever button] to view effective accuracy against monsters in your line of sight."

When you press [whatever button], you are brought to the following screen:

"With this weapon and your current skill/item modifiers, your accuracy against monsters in your line of sight would be:
Adder: 22.8%
Ice statue: 56.4%
Hill giant: 43.2%
VeryAngryFelid's ghost, legendary Felid Wanderer of Dithmenos: 2.5%"

Or something like that. Displays a lot of useful extra information you could probably access via fsim anyways (in a slower and more painful manner), doesn't clutter up any other display, and eventually lets you develop an idea of how accurate various weapons tend to be in general.

or how about two simple options:

(a) just display your chance to hit a monster with your currently equipped weapon when you move the cursor over it (much like defeating MR), and/or
(b) on the character sheet or when you examine a weapon, display your chance to hit a range of representative EV's (lots of Angband variants do this).

VeryAngryFelid wrote:The game displays +3 accuracy displayed for mace, -6 for GC and -7 for GSC but as we can see GC is always the best weapon and even GSC is much better than mace vs Orc Priest.

Can we see that? fsim can be misleading when a single hit can result in massive overkill; so to speak, all the misses get averaged out with one massive hit (most of which is wasted) to produce an AvEffDam that is unrepresentative of the actual expected time-to-kill.

Attacking an adder with a GSC is such a case.

damerell wrote:Can we see that? fsim can be misleading when a single hit can result in massive overkill; so to speak, all the misses get averaged out with one massive hit (most of which is wasted) to produce an AvEffDam that is unrepresentative of the actual expected time-to-kill.

Attacking an adder with a GSC is such a case.

You are correct except AvDam shows that max damage happens very rarely so no damage is wasted usually. Yet we cannot see it, monster HP is less important than weapon accuracy bonus according to devs. So player is not expected to see how much damage is dealt or wasted, it is not important

VeryAngryFelid wrote:You are correct except AvDam shows that max damage happens very rarely so no damage is wasted usually.

I don't think so. AvDam is comparable to an adder's maximum hitpoints; many of the distribution of hits that go to make up that average must be overkill, especially if the adder had lower than maximum hitpoints. I have no idea whether the GSC or +0 mace is better (if we, say, define better as "average number of times attacked by adder").

Another factor is that high variance is undesirable; I would rather have a weapon that hit 100% of the time for X damage than one that hit 10% of the time for 10X damage, even discounting overkill. The mace user might (or might not, as above) need longer on average, but after four attacks (6 turns) there is still a 43% chance the GSC user hasn't connected _at all_. In the same time there is only a 2% chance that the mace user misses every single attack.

Oh, I confused AvHitDam with AvDam. Hitting adder with mace at least once is not useful metrics since adder still attacks with the same power as long as it is alive.

VeryAngryFelid wrote:Oh, I confused AvHitDam with AvDam. Hitting adder with mace at least once is not useful metrics since adder still attacks with the same power as long as it is alive.

I think you're missing the point in the specifics (mind you, the mace can one-shot an adder sometimes and the odds of not hitting twice in that time are pretty small); a weapon with high variance is undesirable because sometimes you'll produce a long series of misses.

damerell wrote:I think you're missing the point in the specifics (mind you, the mace can one-shot an adder sometimes and the odds of not hitting twice in that time are pretty small); a weapon with high variance is undesirable because sometimes you'll produce a long series of misses.

I can say the same about mace: you can deal 0 damage with many hits in a row and sometimes you can kill the adder with first attack of GSC

VeryAngryFelid wrote:I can say the same about mace: you can deal 0 damage with many hits in a row and sometimes you can kill the adder with first attack of GSC

Which leads us to ask the obvious question: which one has the higher overall variance? I don't know, but I strongly suspect it's the GSC because of the very low hit chance and high swing time.

(I should say, I don't disagree with the general idea that the displayed weapon accuracy numbers are very misleading to a novice. Malevolent's suggestion of being able to see the practical effects is a good one.)

The Tavern

Remove accuracy numbers from weapon description

Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description

Re: Remove accuracy numbers from weapon description