Proposal: replace spell power with spell effectiveness

[Hurkyl]

The graph of spell power is pretty much useless for all practical purposes. Even for the purposes of displaying the spell power cap, the graph isn't very useful, due to the high imprecision and that increasing spell power to make spells more effective isn't really very useful.

I propose instead to plot the effectiveness of the spell. For example, your newly minted DEFE opens up his spell menu and sees

  Code:

Spell         Effectiveness  Range   Hunger  Level
Flame tongue      73%        @-->..   None     1

because his spell power is 11, which means that Flame Tongue's damage roll is 1d10 (average 5.5), out of a maximum of 1d14 (average 7.5), and 5.5/7.5 ~ 73.

This conveys the effects of spell power in a much more useful manner: the percentage means you can see far more clearly how close you are to the spell power cap than the # graph, and being based on the actual effectiveness of the spell, it conveys the additional information of what benefit you can expect to get through further training.

Just what numerical quantity to be used as effectiveness is not obvious (e.g. it doesn't account for the fact that power also increases the accuracy of flame tongue), but there are lots of choices that are far better than spell power. Some possibilities

• The effectiveness of a conjuration is average damage
• The effectiveness of a hex is the to-hit roll
• The effectiveness of buffs is the duration
• The effectiveness of Passwall is how many tiles you can pass through

All of the individual effects could be listed on the individual spell pages: e.g. for Flame Tongue, giving the percentages for the average to hit roll and damage roll along with the plot for range, hunger, and noise.

I also propose to eliminate the alternate spell view

The alternate spell view which displays spell power, range, and hunger is not especially useful. There is not much reason to actually compare hunger and range of between different spells. And there is no reason at all to compare the power of different spells, since that doesn't tell you anything useful at all. (although it would be useful if this number was something that varied proportionally to the "average damage roll" for a conjuration or the "average to hit roll" of a hex...)

Noise might have been useful on this view to compare between spells, but alas it's not there.

Thus I propose doing away with this view entirely. Maybe move hunger up to the primary spell view, or just leave it banished to the individual spell pages. Maybe put a graph of spell level versus hunger costs someplace, like the % view.

[Bart]

I dislike the word 'effectiveness' in this place, because it suggests that if my firestorm is at 60%, it is not very effective. I would like to find somewhere information about my real spellpower, though - e.g. 73/200, which is clear, easy to read and not overly spoilery.

Regarding second screen - please, do not remove hunger bars, which are very relevant for my spellcasting training. Checking each spell separately every time would be far less convenient. Also there is good reason to compare spell power for me in hashtag format - if I see that one of bars is much longer than another, it instantly makes me think whether I should not boost the weaker spell by appropriate training. It would be inconvient to have this information hidden.
I agree on spell range comparison - I can imagine that new players might want to compare range, but it's probably not very necessary even for them.

[Hurkyl]

If effectiveness is tied to damage, then if firestorm is at 60%, then it's doing 60% of the damage it could be doing. I suggested numbers based on the percentage of max damage, since I believe the devs are strongly apathetic to putting any figures that can be used to reasonably compare different spells. A different word would be fine, though.

The only use of knowing the numerical value of spell power is so a spoiled player can look up the damage formulas and plug it in to see what their spell is doing: if you can convince the devs to give us a number, I would prefer it be a number more directly related to what we care about!

To better phrase the aspect you're worried about losing, my suggestion does hides the rate at which you would increase the percentage with training. That, I think, is really what you care about: knowing, for example, that training Conjurations and Earth will improve the percentage of max damage Stone Arrow is doing relatively quickly.

I posit that this actually isn't needed in the display: with few exceptions, the spell power cap correlates with spell level, so a person would naturally think that powering up Stone Arrow would happen a lot quicker than powering up Iron Shot.

[DracheReborn]

From your example, honestly I'd much rather that the display shows something like:

Flame Tongue Dam: 1d10 (max 1d14)

Which I'd regard as something equivalent to weapon base damage. The spell effectiveness thing you propose seems like just another intermediate step that's not all that helpful.

Edit: Hmm, looks like you have the same idea

if you can convince the devs to give us a number, I would prefer it be a number more directly related to what we care about!

[Hurkyl]

From your example, honestly I'd much rather that the display shows something like:

Flame Tongue Dam: 1d10 (max 1d14)

Which I'd regard as something equivalent to weapon base damage. The spell effectiveness thing you propose seems like just another intermediate step that's not all that helpful.

I certainly agree (I might prefer average damage roll, but that's not much different). Unfortunately that change is too radical; I'm hoping the one I proposed is conservative enough that it could get accepted. While it's not what I actually want, it's, in my opinion, somewhat better than what we currently have.

[Hurkyl]

I suppose I should add technical details: the averages can be computed by simple formula rather than requiring complicated calculations or simulation.

The average value of a 1dx roll is (x+1)/2. This is true even if n is not an integer.

The average value of xdy * mdn is the product of the averages of xdy and mdn.

The average value of a 1d(x + ydz), which I think is relevant only for Airstrike, can be computed by first finding the average (call it u) of x + ydz, and then taking the average of 1du.

And similarly for other calculations.

[Lasty]

Aside from the fact that the developers have indicated previously that they don't want to get players worrying too much the exact numbers and min/maxing of values in-game, there are some obstacles to this that prevent it from being as clear-cut as you'd like. For example, if the spell power for a spell doesn't evenly divide into a number of sides given how many dice that spell rolls, it randomly rounds the number of sides up or down weighted by the size of the decimal remainder; as such, if you display that a flame tongue "1d9", you're lying most of the time, because it's could be doing something in the range of 1d(8 or 9 or 10). This is aside from the way that Crawl deliberately does not display {x}d{y} anywhere in the game, and explicitly chooses not to display exact damage values at any point.

In terms of "effectiveness", the only reasonable metric is spell power, but presenting your "effectiveness" as a function of spell power would often give you the impression that your spells are performing well below where they "ought" to be. Any method that displays the quality of the spell in terms of its relationship to its maximum possible value will 1) not usefully apply to many spells, and 2) give the completely misleading impression that it is necessary or even desirable to get spells to their maximum power.

And that's aside from the fact that the current hash graph already gives you plenty of information about the percentage of spell power you've got (if you choose to learn the breakpoints) without presenting numbers that imply a smoothly-rising percentage scale that will distort how the average player values spell power and without convincing the player that spell power rises linearly as a result of improving skills.

If you think that a finer scale is necessary for determining the "effectiveness", I would propose that you first give an example of a case where knowing the difference between being at 76% of maximum spell power and being at 81% of maximum spell power would help the player to play Crawl better.

[TeshiAlair]

I do find it IMMENSELY frustrating and spoiler to have to look up damage values on the wiki for things like "should I use throw icicle or stone arrow against that cold resistant monster."

[Hurkyl]

For example, if the spell power for a spell doesn't evenly divide into a number of sides given how many dice that spell rolls, it randomly rounds the number of sides up or down weighted by the size of the decimal remainder; as such, if you display that a flame tongue "1d9", you're lying most of the time, because it's could be doing something in the range of 1d(8 or 9 or 10). This is aside from the way that Crawl deliberately does not display {x}d{y} anywhere in the game, and explicitly chooses not to display exact damage values at any point.

1d9.25 isn't a terrible thing to write. And this would be completely hidden if the mean damage were displayed. But I know that this won't happen an time soon.

In terms of "effectiveness", the only reasonable metric is spell power,

I strongly disagree. Spell power tells you, e.g. for a blast, where you are between the minimum and maximum damage that the spell can do. Power is thus very misleading, since the minimum is often a very significant percentage of the maximum.

Like the example I gave with flame tongue: 44% of the spell power gives 73% of the maximum damage....

but presenting your "effectiveness" as a function of spell power would often give you the impression that your spells are performing well below where they "ought" to be.

which is why this is exactly backwards. Knowing you only have ##.. spell power makes it look like your spell is much weaker than it actually is.

Any method that displays the quality of the spell in terms of its relationship to its maximum possible value will 1) not usefully apply to many spells,

Which, in part, is why my thoughts drifted to a better breakdown in the spell page; Flame Tongue can show damage and accuracy, while Regeneration could show duration, for example.

and 2) give the completely misleading impression that it is necessary or even desirable to get spells to their maximum power.

How is this any different from showing spell power, even in its current form?

And that's aside from the fact that the current hash graph already gives you plenty of information about the percentage of spell power you've got (if you choose to learn the breakpoints) without presenting numbers that imply a smoothly-rising percentage scale that will distort how the average player values spell power and without convincing the player that spell power rises linearly as a result of improving skills.

If you think that a finer scale is necessary for determining the "effectiveness", I would propose that you first give an example of a case where knowing the difference between being at 76% of maximum spell power and being at 81% of maximum spell power would help the player to play Crawl better.

The difference is more like 60% versus 96% of spell power for a ###. spell with a 25 power cap, for example. Of course, in terms of damage for Flame Tongue, that's really 80% versus 100% of the possible damage roll (if I've read correctly, you actually get full damage at 24 power for Flame Tongue).

[Hurkyl]

I do find it IMMENSELY frustrating and spoiler to have to look up damage values on the wiki for things like "should I use throw icicle or stone arrow against that cold resistant monster."

Not to mention surprises like the fact that, before you have high spell power, it's better to spam Magic Dart than to use higher level spells against low AC monsters in situations where your primary concern is mana efficiency (the other level 1 blasts are even stronger, of course). But really, this is off-topic since I don't want to get my more conservative proposal derailed by the usual argument on this topic.

[Lasty]

1d9.25 isn't a terrible thing to write. And this would be completely hidden if the mean damage were displayed. But I know that this won't happen an time soon.

I would argue that it is terrible, but worse than that, it's also just wrong. It doesn't do 1d9.25, it does either 1d9 or 1d10 with a 75:25 weighting.

I strongly disagree. Spell power tells you, e.g. for a blast, where you are between the minimum and maximum damage that the spell can do. Power is thus very misleading, since the minimum is often a very significant percentage of the maximum.

For many charms duration is all right. Damage (to the extent that it can be accurately represented) is okay, but it ignores the effect of spell power on accuracy. There are a lot of other spells where the "effectiveness" is even less clear. Here's a very brief list:

* Hexes/agony -- how does investment in hexes affect the effectiveness of hexes? That depends on target MR.
* Pain -- deals damage and also needs to overcome MR.
* Summons and anything else where duration isn't the only effect of increased power.
* Bjorgnjor's.
* Charms like Stoneskin and Ozo's.

but presenting your "effectiveness" as a function of spell power would often give you the impression that your spells are performing well below where they "ought" to be.

which is why this is exactly backwards. Knowing you only have ##.. spell power makes it look like your spell is much weaker than it actually is.

Honestly, that's pretty accurate anyway, since after factoring in AC, you probably are at something closer to half effective damage on that spell. Aside from that, those extra two #s come fast, which means that the period where you're feeling "weak" is brief. When you can never crack 80% effectiveness on your firestorm -- or even bolt of fire! -- you're going to really feel like you're doing something wrong.

and 2) give the completely misleading impression that it is necessary or even desirable to get spells to their maximum power.

How is this any different from showing spell power, even in its current form?

The sensible response to the current system is to realize that 1) the number of #s increases as various factors increase, and 2) notice that you never really fill up the bars, and thus conclude that 3) it's hard or impossible to fill up the bars, and so you shouldn't monitor them that closely. Looking at percentages makes you feel like you should keep trying to eke out additional percentage points forever.

The difference is more like 60% versus 96% of spell power for a ###. spell with a 25 power cap, for example. Of course, in terms of damage for Flame Tongue, that's really 80% versus 100% of the possible damage roll (if I've read correctly, you actually get full damage at 24 power for Flame Tongue).

Okay, then explain how knowing the difference between doing 60% and 96% damage with flame tongue (to use your not entirely accurate number) will improve your play.

[Hurkyl]

There are a lot of other spells where the "effectiveness" is even less clear. Here's a very brief list:

* Hexes/agony -- how does investment in hexes affect the effectiveness of hexes? That depends on target MR.

And damage depends on target AC, and accuracy depends on target EV. That the enemy's defenses oppose your offense is not an obscure idea.

* Pain -- deals damage and also needs to overcome MR.
* Summons and anything else where duration isn't the only effect of increased power.
* Bjorgnjor's.
* Charms like Stoneskin and Ozo's.

If we keep the alternate view field, then just pick one of the options; it won't be any worse than displaying spell power!

but presenting your "effectiveness" as a function of spell power would often give you the impression that your spells are performing well below where they "ought" to be.

which is why this is exactly backwards. Knowing you only have ##.. spell power makes it look like your spell is much weaker than it actually is.

Honestly, that's pretty accurate anyway, since after factoring in AC, you probably are at something closer to half effective damage on that spell.

Not really. Different enemies have different AC, and the relative impact enemy AC has on damage dealt varies a lot with spell. 5 AC takes a much greater huge chunk out of Magic Dart's damage than Flame Tongue's!

But that aside, a person who reasons like you do would look at ##.. and reasonably think the damage roll is only half of what it could be, and thus think that his spells are only doing a quarter of the damage they would do if they maxed it out!

But again, I posit that people understand that enemies have defenses. (although the fact it's AC that protects against magic damage rather than some magic resistance or magic defense statistic might be a surprise to newcomers)

Aside from that, those extra two #s come fast, which means that the period where you're feeling "weak" is brief.

Depends on the character. And the same goes for 50 and 100 power capped spells.

When you can never crack 80% effectiveness on your firestorm -- or even bolt of fire! -- you're going to really feel like you're doing something wrong.

People feel the same way about ########.. too.

and 2) give the completely misleading impression that it is necessary or even desirable to get spells to their maximum power.

How is this any different from showing spell power, even in its current form?

The sensible response to the current system is to realize that 1) the number of #s increases as various factors increase, and 2) notice that you never really fill up the bars, and thus conclude that 3) it's hard or impossible to fill up the bars, and so you shouldn't monitor them that closely. Looking at percentages makes you feel like you should keep trying to eke out additional percentage points forever.

You can fill up the bars for lower level spells. It seems pretty sensible to think you should be able to fill up the bars for higher level spells too. I don't see how it's any easier to give up when you fail to fill the bars than it is to give up when you never break 75%.

In my opinion, the bars are even worse in this regard. It's far easier to think "Okay, I'll train a little bit more to get to ########.." and then 10 skill levels later "what? I'm still at #######...?" due to the coarseness. You won't have that problem with the percentages.

The difference is more like 60% versus 96% of spell power for a ###. spell with a 25 power cap, for example. Of course, in terms of damage for Flame Tongue, that's really 80% versus 100% of the possible damage roll (if I've read correctly, you actually get full damage at 24 power for Flame Tongue).

Okay, then explain how knowing the difference between doing 60% and 96% damage with flame tongue (to use your not entirely accurate number) will improve your play.

In any scenario, basing the decision on the damage the spell does will be better than basing the decision on spell power.

One of my main points is that even if you believe that having a display (be it a number, the # graph or something else) based on damage is not actually useful for a player, having a display based on spell power is even worse. So even for skeptical people who believe nobody in the world could possibly find information useful for making informed decisions, making the change should still be seen as an improvement, as it makes things less bad.

You even made a the very mistake I'm suggesting people make by seeing spell power: you saw the "60% and 96% of spell power", and you mentally substituted "damage" for "power".

Anyways, there are easy examples: if I know I'm only doing 60% of the damage I could be doing with Flame Tongue, I have a pretty strong incentive to train more Fire/Conjurations for a few skill levels, and can plan for it to be a while before I train other skills. But if I'm at 90% of max damage, it's not such a pressing issue.

It's easy to see how I might play worse without these numbers: if I know I have room to make Confuse a lot better, I can reasonably consider training more Hexes. But if I think Confuse is already near its power cap, I might give up on training Hexes to improve its success rate.

This one was actually a big deal for me, as I relied a lot on confuse stabbing as I learned the game... and I gave myself a big handicap because I hadn't realized there was a lot more room to improve Confuse.

[duvessa]

I am assuming the response to that will be to remove the spell power display.

[Ghoul King]

I'd like a better idea of spell performance in general. The game only occasionally clarifies specifics (eg Lee's Rapid Deconstruction specifies what number of #es is required to break X, Y, and Z types of walls), and doesn't really explain anything about what it means otherwise. If I have two, is that actually roughly twice the performance? I don't know, and the game doesn't indicate anything about it. By the same token I honestly can never tell if a spell is close to max power, at max power, nowhere near max power... I've never figured out what the game's display is supposed to be conveying. It's so bad I often feel like I'd be better off with zero information, so that I would at least not have an illusion of knowledge.

In particular, number of #es is not a helpful barometer for working out roughly how powerful X spell is compared against Y spell, because a # is not some base unit of performance, it's specific to the spell. If Flametongue has ###, and Stone Arrow has ####, that doesn't tell me that Stone Arrow is roughly 33% stronger than Flametongue, it just tells me I have more Earth training than Fire training... which I can already tell by looking in the Skills screen.

Or put more simply: so far Lee's Rapid Deconstruction is the sole spell where I have ever paid any attention to the Power bar, because it is 100% useless to pay attention to normally.

On the topic of spells being affected by AC: honestly, I've never gotten a good sense of how this works. Do enemies get GDR? Does GDR affect spell damage? Or is it just AC? Resistances existing just further confuses the issue, and intuition doesn't tell me anything: there's no Earth damage so it makes perfect logical sense to me for Earth spells to behave basically like a physical attack, except for mana costs et al, but if everything is affected by AC anyway, what makes Earth damage spells anything other than flatly better than roughly equivalent elemental damage spells? (eg Stone Arrow vs Throw Frost and Throw Flame, all three being low-level projectiles which stop at the first target they hit, can miss and pass through to other targets, etc)

Digging around on the wiki, the forums, and other sites entirely has never clarified some of these questions either (DO enemies get GDR from armor??), so it's not a matter of being unspoiled.

[Hurkyl]

I am assuming the response to that will be to remove the spell power display.

That would be unfortunate, but mostly better than the current situation anyways. After giving some thought, I realize that spell power doesn't even correlate linearly to your skill training due to spell power stepdowns, so pretty much the only information content in the bar is what the power cap is and if you're there yet.

I argue something needs to be displayed, though. For a wizard spamming Magic Dart as his primary weapon, the difference between 1d7 and 1d8 is really significant, and knowing when you've maxed it out is very useful. For an Enchanter, knowing that Ensorcelled Hibernation won't become any better with further Hex/Ice training is something that is good to know. There are probably other intermittent examples; e.g. Ghoul King mentions Lee's Rapid Deconstruction above.

[Siegurt]

Digging around on the wiki, the forums, and other sites entirely has never clarified some of these questions either (DO enemies get GDR from armor??), so it's not a matter of being unspoiled.

No, monsters don't get GDR, and GDR doesn't effect spells (or ranged attacks) it's only for melee damage, and only for the player.

[duvessa]

https://loom.shalott.org/learndb.html#gdr

[Lasty]

there's no Earth damage so it makes perfect logical sense to me for Earth spells to behave basically like a physical attack, except for mana costs et al, but if everything is affected by AC anyway, what makes Earth damage spells anything other than flatly better than roughly equivalent elemental damage spells? (eg Stone Arrow vs Throw Frost and Throw Flame, all three being low-level projectiles which stop at the first target they hit, can miss and pass through to other targets, etc)

You're right that damaging spells generally are resisted by AC, with a few exceptions (Freeze/Ozo's Refrigeration, Pain, and Agony are not, some lightning damage is affected by half AC). What sets Stone Arrow apart from Throw Frost/Flame is that 1) it has a shorter range, 2) it is higher level, and 3) it does significantly more and irresistible damage.

Throw Frost and Throw Flame are quite weak, but they're rare among conjurations in having a full-LOS range. Earth spells in general pay for their wide applicability by having significantly shorter range than other conjurations, hitting only single targets, and having poor accuracy at low power.

Digging around on the wiki, the forums, and other sites entirely has never clarified some of these questions either (DO enemies get GDR from armor??), so it's not a matter of being unspoiled.

GDR is actually one of the things that works better if you're unspoiled. Knowing about GDR at best leaves you making the same decisions you would if you didn't know about it at all. The only way that GDR can get you into trouble is if you change your behavior based on knowing that it exists.

[Ghoul King]

Knowing GDR exists influences my perception of the game, if nothing else, since it's one of the rare cases of the game being something other than 100% screw-you-whenever-I-feel-like-random. A big frustration with the game that looking into the maths has confirmed is not simply perceptual or unusually bad luck on my part is how the game can always roll: you miss twenty times in a row, you do a big fat 0 out of your arbitrarily large damage number twenty times in a row, you fail to dodge twenty times in a row in spite of godly dodging, etc. The fact that GDR exists means good armor can't roll 0 AC twenty times in a row (Or once at all, in fact) is a relief. Sure, the general rule of thumb is always going to be 'wear the heaviest armor you can get away with' regardless, but it makes a difference in how I view my experience of the game.

My point about Earth spells was not 'they are unfair in design', my point is that intuition suggests a line of reasoning like: Earth spells do non-elemental damage, and therefore are logically enough influenced by armor, so it makes sense they don't have a resist. The inverse is that resist must obviously be the only thing that influences elemental damage, since otherwise I would expect Earth to have a resist in addition to being affected by AC. The fact that exceptions to being influenced by AC do exist can just exacerbate this misunderstanding. (When I first played I mostly cut my teeth on Mummy Necromancers ie I spammed Pain as my damage spell and didn't have reason to believe it was highly unusual as a damaging spell) In short: the game's construction is not intuitive on this topic, and it doesn't bother to explicitly inform you. It wasn't until I noticed a pattern of difficulty hurting Orcs in Chain Mail and similar with Flametongue, even if the armor was unenchanted or proved to be non-Fire Resistant when I picked it up, that I became aware of the possibility that elements that could be resisted might also be affected by AC. It's just not intuitive that some spell realms are double-punished and others aren't, and honestly most enemies don't have noticeable AC and don't use elemental spells so you mostly don't get a chance to see your Crystal Plate Mail laugh off Throw Flames or enemy Crystal Plate Mail laugh off your own.

Though while we're commenting on Throw Flame and Throw Ice I will say I never bother to memorize Throw Flame and only memorize Throw Frost as an Ice Elementalist because Freeze is point-blank and Throw Icicle is even more mana and hunger-intensive and prone to spell failure for quite some time. Throw Flame's added range basically never adds value because everything I'm fighting before Fireball is a practical option is: immune to Fire entirely, unwilling to go through a flame cloud, so pathetic I have no need to snipe it before it 'threatens' me, so dangerous I really should just be running instead of sniping, willing to go through a flame cloud but not actually immune and therefore Conjure Flame can be used to do obscene damage reliably, or is a Centaur, which I would rather duck around a corner and Flametongue once it gets within reach than burn mana at twice the rate as Flametongue and run out before it's dead. If I was playing a straight warrior dabbling in magic, I suspect I'd rather dabble in Throwing and hurl Javelins/Tomahawks/whatever in situations Throw Flame is theoretically useful in.

[TeshiAlair]

So it isn't obvious that wearing armor makes you get hurt less by a blast of frost than wearing a robe?

I agree with the throw flame/throw frost thing, there was a thread on it that died from lack of implementation as I recall. (I actually think I started it O_O )

[duvessa]

So it isn't obvious that wearing armor makes you get hurt less by a blast of frost than wearing a robe?

I've actually seen a lot of people who think that ranged attacks ignore AC.

[Sar]

I think I had 3-5 wins before somebody told me that elemental attacks check AC. I was just stacking as much elemental resists as possible.

[roctavian]

I noticed a pattern of difficulty hurting Orcs in Chain Mail and similar with Flametongue, even if the armor was unenchanted or proved to be non-Fire Resistant when I picked it up, that I became aware of the possibility that elements that could be resisted might also be affected by AC

Recognizing patterns and changing how you play the game is not only a basic conceit of roguelikes, but of most games in general. I just want to point out that you learned how this worked through gameplay and that's the system working as intended, as far as I'm concerned.

Please do not extend this sentiment of mine to all parts of the game; I'm not arguing against transparency. However, "how much damage will I do against X type of enemy with Y offensive option" is highly conditional in an interesting way, something you can experiment with during the course of a game, and isn't something I've ever felt I had to look up spoilers for.

[Ghoul King]

Digging around on the wiki, the forums, and other sites entirely has never clarified some of these questions either (DO enemies get GDR from armor??), so it's not a matter of being unspoiled.

No, monsters don't get GDR, and GDR doesn't effect spells (or ranged attacks) it's only for melee damage, and only for the player.

I didn't register the significance of this before, and having noticed it: what. The reason Centaur and Cyclops are disproportionately threatening, and so too for any random Goblin or Orc that spawns with or picks up a ranged attack option, isn't just because ranged attacks are vastly more difficult to deal with in general but because they ignore a major benefit of armor? Why? What possible design motivation could drive this?

Also that would explain why some people think...

I've actually seen a lot of people who think that ranged attacks ignore AC.

... this is true.

So it isn't obvious that wearing armor makes you get hurt less by a blast of frost than wearing a robe?

Somebody is conjuring raw anti-heat from the ether by waving their arms and gibbering nonsense, and I'm supposed to expect the resulting anti-physics to be basically the same thing as an arrow? And in what universe does setting someone on fire work less well by stuffing them into a sweltering tin can to cook them inside? (Crawl's universe, apparently) Or how about the stuff that unlife is made of being zapped at you, is that just like an arrow too? And of course everybody knows that lightning can't pass through highly conductive metallic substances with greater ease than non-conductive substances. (Of course in actual fact the way electricity hurts you is by heating you up when it meets resistance ie a lack of conductivity!)

Poison is honestly the only resistable damage type where I find it entirely unsurprising that say, Sting, cares about AC. Poisons normally need to get into your bloodstream, which means penetrating your skin, which means reaching your skin in the first place. Fair enough.

I agree with the throw flame/throw frost thing, there was a thread on it that died from lack of implementation as I recall. (I actually think I started it O_O )

I think it's actually one of the top Google results for 'Throw Flame Dungeon Crawl'.

Recognizing patterns and changing how you play the game is not only a basic conceit of roguelikes, but of most games in general. I just want to point out that you learned how this worked through gameplay and that's the system working as intended, as far as I'm concerned.

Please do not extend this sentiment of mine to all parts of the game; I'm not arguing against transparency. However, "how much damage will I do against X type of enemy with Y offensive option" is highly conditional in an interesting way, something you can experiment with during the course of a game, and isn't something I've ever felt I had to look up spoilers for.

I'm not pushing for it to be changed, just commenting that it is legitimately unintuitive and unclear, which wasn't even a primary point/agenda to push/whatever of mine, just something I said that drew a strong response. (Which in turn suggests I was on the mark!)

In particular, the game uses so much randomness that it isn't actually all that readily learned through playing the game. I have had lv10+ characters fail to hurt a random unarmored, unarmed Goblin for 10+ turns in a row through sheer randomness. Am I to suspect the game randomly generates unlabeled uberGoblins? By the same token, when Flametongue repeatedly fails to get damage through on one random Orc, the explanation is usually 'RNG screw' not 'Plate Mail has high AC, and that's relevant'.

[Lasty]

I didn't register the significance of this before, and having noticed it: what. The reason Centaur and Cyclops are disproportionately threatening, and so too for any random Goblin or Orc that spawns with or picks up a ranged attack option, isn't just because ranged attacks are vastly more difficult to deal with in general but because they ignore a major benefit of armor? Why? What possible design motivation could drive this?

This is actually the wrong lesson to draw here, as GDR really isn't particularly significant in this case either. The reason that monsters with ranged weapons are so absurdly dangerous is because for some reason the monster code is set to get absurd bonuses to thrown/launched attacks. This was extremely visible in previous builds of Dithmenos shadow mimic where the shadow was generated a monster with level equal to yours. Mimicking your long sword attacks would be a modest damage boost, but mimicking your unskilled thrown stone fucking murdered everything.

[and into]

Ghoul_King: There are definitely things Crawl can be more transparent about while still avoiding info dumps and the like, but I think a lot of the stuff you are asking for simply reflects your own preconceptions from other games. Of course everyone will do that to some extent, unavoidably. But you seem to be saying "unintuitive" in a fairly subjective way. To you it is intuitive that poison and earth magic spells check AC, but not fire, ice, or air. But to others, it will probably vary, perhaps even by spell. To some it makes sense that explosions are mitigated by AC, and this is true in Crawl (and in other games); to others, they might initially assume fireball bypasses AC.

FWIW I think it would be fine to say in a spell description, "Damage from this spell is mitigated by AC" or "Damage from this spell bypasses AC" or "Damage from this spell bypasses half of the target's AC" etc. But I don't think many people would exactly have the same intuitive sense of "this made up spell is blocked by armor, this one isn't."

As for GDR it really shouldn't be displayed because it is extremely difficult to describe meaningfully and succinctly what it does, especially because when people hear "GDR" their mind jumps to things that are actually wrong and do not reflect how it works in Crawl. Just do a search on the Tavern for GDR for the long exhaustive discussions about this, if you are curious. For reals.

The long and short of it is that high GDR will make melee damage less prone to variance/spikes than otherwise, assuming your total AC value is not too low. However even apart from GDR, a character with good AC and bad EV will generally have less damage spikes even in a robe (no GDR) than a character with bad AC and good EV, even if the average damage mitigation over time is the same for both.* So in this sense GDR is emphasizing a difference between AC and EV that already somewhat exists. Because high AC and high GDR generally (though not always) correlate, a lot of what people think is the awesomeness of really high GDR is oftentimes just the awesomeness of having high AC.

*(Of course in practice the two aren't at all mutually exclusive and most characters want, and will eventually get, good AC *and* good EV, even if one is higher than the other, but I'm just illustrating a point here, not positing game scenarios that would realistically occur.)

So the way I think about it, and I think this is a good way to think about it, is this: GDR exists in order to make AC in heavy armor "feel" different from EV, even in situations in which the average damage mitigation is (close to) the same. It should be thought of primarily in terms of differentiation: GDR helps differentiate the way one thing feels in combat, versus another, and that differentiation is good for the game design of Crawl.

[Ghoul King]

I didn't register the significance of this before, and having noticed it: what. The reason Centaur and Cyclops are disproportionately threatening, and so too for any random Goblin or Orc that spawns with or picks up a ranged attack option, isn't just because ranged attacks are vastly more difficult to deal with in general but because they ignore a major benefit of armor? Why? What possible design motivation could drive this?

This is actually the wrong lesson to draw here, as GDR really isn't particularly significant in this case either. The reason that monsters with ranged weapons are so absurdly dangerous is because for some reason the monster code is set to get absurd bonuses to thrown/launched attacks. This was extremely visible in previous builds of Dithmenos shadow mimic where the shadow was generated a monster with level equal to yours. Mimicking your long sword attacks would be a modest damage boost, but mimicking your unskilled thrown stone fucking murdered everything.

Well that's pretty inexplicable either way. I've wondered why ranged attacks are always vastly more threatening, whether it's a Cyclops throwing Big Rocks (fair enough) or a Goblin throwing Stones rather than hitting me with a +2/+3 Scimitar of Flaming. (Buh)

and into: I was actually intending to add to my previous post anyway after some thought (But my internet crapped out), and part of what I have failed to allude to is this: when a monster has a resistance of relevance to the attack and gets its resist roll, you are explicitly informed by the game that the monster got its resist roll, and that's why you did no damage. But the game provides no feedback on AC vs low damage roll: if you do no damage, there's no "The Orc's armor completely absorbs the blow!" when you get a high damage roll that is blocked by a high AC roll vs "Your feeble blow fails to move the Orc" when you just flatly roll a 0 on the hit. All you know is that you did no damage, for some reason. This is part of what makes the whole question of whether a given spell is affected by AC unclear: when I do no damage with Magic Dart, is it because I got a bad damage roll, or because their armor blocked it? No clue.

As far as intuition goes: no, I honestly am not drawing from other games. Crawl has a lot of its logic explicit and still more of it implicit (Silence shuts down spellcasting, so spellcasting must have some sound-production component. In fact, I seem to recall Silence will also block your own shouts?) and extrapolating from that logic suggests things that don't necessarily apply. AC is a direct representation of armor: armor in the real world protects you in two ways. Either the armor aborbs the energy (eg bulletproof vests) or it redirects it. (This is the main way most medieval armor provides most of its protection, and is why modern tanks have lots of slopes in their design) A suit of plate mail can, naturally enough, readily absorb/redirect the energy involved in trying to jam a dagger into someone, and can deflect oncoming arrows. (But if it doesn't, it punches through and does severe damage. Crossbows made the armored knight substantially less valuable, because suddenly any untrained peasant could bypass your armor with a much cheaper piece of gear that took considerably less time to equip) But absorbing a rush of heat is in no way helping the problem (Again: cook you in your tin can), and the idea of it re-directing heat/flame seems strikingly odd. How exactly does it do this?

If Dungeon Crawl made less of an effort to make itself more realistic eg patching Felids in specific to have a huge stealth penalty from Blade Hands and then re-patching it so this doesn't apply when they're flying, I wouldn't really pay that much attention to its explicit or implicit logics because they would obviously be thin justifications of little bearing on actual gameplay. But because it does deliberately hold to a kind of 'realism' when it can, even when no game logic reason presents itself for doing so, this suggests that real-world knowledge is a good starting point for how to think about the game. Poison is meant to resemble real poisons: it is not simply a term for damage-over-time as in most games, including that many poisonous creatures have poisoned flesh. (But not snakes and the like who, in the real world, are perfectly edible if you're not eating the venom sac contents)

So no, I'm not talking 'oh, every other game works this way, so Crawl must' and calling it 'intuitive' to be like other games. (Which is a sentiment I honestly find puzzling)

I will also add that I honestly largely don't care about the topic of GDR. At this point I'm more curious as to why my talking about GDR has drawn more of a response than anything else I've said.

I will also also add: I was not entirely correct in my earlier post disparaging Throw Flame. I'd forgotten Giant Spores. On the other hand, I usually have a wand to handle them, or more likely three, at least one of which is relevant.

[basil]

If Dungeon Crawl made less of an effort to make itself more realistic eg patching Felids in specific to have a huge stealth penalty from Blade Hands and then re-patching it so this doesn't apply when they're flying, I wouldn't really pay that much attention to its explicit or implicit logics because they would obviously be thin justifications of little bearing on actual gameplay.

That's more an argument for removing that Fe special case than anything else.

But because it does deliberately hold to a kind of 'realism' when it can, even when no game logic reason presents itself for doing so, this suggests that real-world knowledge is a good starting point for how to think about the game.

realism of crawl[1/11]: whacking KILLER FUCKING BEES with a halberd and throwing javelins at them

That aside, perhaps the tutorial could have a line or two saying that AC affects ranged and most damaging spells and abilities if it doesn't already.

[duvessa]

As far as I know, this is the entirety of what Crawl currently says about AC:

  Code:

Armour Class
  Abbreviated to "AC". When something injures you, your AC reduces the
  amount of damage you suffer.

[Ghoul King]

If Dungeon Crawl made less of an effort to make itself more realistic eg patching Felids in specific to have a huge stealth penalty from Blade Hands and then re-patching it so this doesn't apply when they're flying, I wouldn't really pay that much attention to its explicit or implicit logics because they would obviously be thin justifications of little bearing on actual gameplay.

That's more an argument for removing that Fe special case than anything else.

It's not an argument for anything. It's a statement about how the game constructs itself, and what this says about developer intent and lines of reasoning.

But because it does deliberately hold to a kind of 'realism' when it can, even when no game logic reason presents itself for doing so, this suggests that real-world knowledge is a good starting point for how to think about the game.

realism of crawl[1/11]: whacking KILLER FUCKING BEES with a halberd and throwing javelins at them

That aside, perhaps the tutorial could have a line or two saying that AC affects ranged and most damaging spells and abilities if it doesn't already.

I've always been puzzled that they're not labeled 'giant killer bees' or anything, honestly. Why Giant Ants but not Giant Killer Bees?

I have doubts many people really play through the tutorial particularly. I played through it a little, because I prefer to see as much content as I can if possible, but honestly the tutorial was largely tiresome and boring and all I really remember was the 'hidden door' mechanic striking me as so resoundingly dumb that I couldn't believe it existed. (I am thus exceedingly glad it's been removed)

I'd rather see it placed somewhere in the Help menu, or made more explicit in general message stuff. (eg 'the Orc's armor completely absorbs your attack' when anything affected by AC does no damage in part due to AC)

A better description than what is apparently the current one, anyway, which among other points led me to believe that AC was a flat damage reducer eg 10 AC is 10 damage that didn't get through every time AC applies.

[basil]

It's not an argument for anything. It's a statement about how the game constructs itself, and what this says about developer intent and lines of reasoning.

Aside from "BHands reducing Fe stealth" doing approximately nothing in a hidden way, it says the wrong things about developer intent (or instead, dev intent shouldn't tend towards those lines). So instead of attempting to simulate how plate and magical fire interact in real life, Crawl attempts to be clear with its mechanics, as well as consistent-ish. Which is why I'm going to edit the tutorial and manual to clarify what AC does "soon enough."

[Lasty]

Well that's pretty inexplicable either way.

I don't get it either.

At this point I'm more curious as to why my talking about GDR has drawn more of a response than anything else I've said.

I'm interested in helping people play better and understand the game better, and I've found that when people bring up GDR, it tends to distort their play badly, so I try to address the topic ASAP. The rest of your posts I didn't feel I had a useful perspective on, so I didn't reply.

I do think I'd rather get hit by a blast of flame while wearing plate armour than while wearing no armour. Plate usually has layers of other materials under it that would ameliorate the heat -- and heck, even just the metal heating would probably be better than contacting my skin directly. If the flame heats the metal enough that contact with it would seriously damage me, then having it impact me directly would probably inflict even worse damage. Ditto cold. But at the end of the day, Crawl doesn't go nearly as far into realism and interaction-modeling as Nethack or other roguelikes, so it's probably not that useful to appeal to those things.

There might well be value to the idea of displaying different messages depending on how much armour mitigates damage, and if you feel strongly about it, I'd encourage you to start a new GDD thread that lays out the proposal in more detail.

[TeshiAlair]

More on topic: I do think still that being able to compare spells better, just in terms of raw potential, is a good thing, particularly in cases where that isn't clear. It is easy to tell that lightning bolt has very variable damage, but harder to gauge power levels of bolt of magma vs bolt of fire.

[Ghoul King]

It's not an argument for anything. It's a statement about how the game constructs itself, and what this says about developer intent and lines of reasoning.

Aside from "BHands reducing Fe stealth" doing approximately nothing in a hidden way, it says the wrong things about developer intent (or instead, dev intent shouldn't tend towards those lines). So instead of attempting to simulate how plate and magical fire interact in real life, Crawl attempts to be clear with its mechanics, as well as consistent-ish. Which is why I'm going to edit the tutorial and manual to clarify what AC does "soon enough."

Blade Hands Felid interaction is maybe a bad example, but my point was more that this kind of decision is all over Crawl. To be entirely fair the game has become markedly more 'game-like' with development switching from the original creator to being a community project, but that's part of why I selected the Blade Hand Felid interaction as an example: it's not a legacy feature from before the game became a community project.

In general Crawl's modern design philosophy seems to be: the game should be as realistic as it can be without substantially compromising the gameplay. (eg cooking, salting, or otherwise preserving meat chunks will never happen because it breaks a major gameplay dynamic) That was what I was trying to convey.

I will also add: note that Killer Bees have sky-high evasion such that poking them to death with a spear, though possible, is an exercise in frustration. This is basically what would actually happen if you tried to stab real Killer Bees to death. The main difference is honestly that Crawl Killer Bees come in hordes of 2-3 dozen at most, not hundreds.

I do think I'd rather get hit by a blast of flame while wearing plate armour than while wearing no armour. Plate usually has layers of other materials under it that would ameliorate the heat -- and heck, even just the metal heating would probably be better than contacting my skin directly. If the flame heats the metal enough that contact with it would seriously damage me, then having it impact me directly would probably inflict even worse damage. Ditto cold. But at the end of the day, Crawl doesn't go nearly as far into realism and interaction-modeling as Nethack or other roguelikes, so it's probably not that useful to appeal to those things.

The point I was conveying poorly wasn't "I'd rather be naked than be in a tin can when a fireball is slung my way", but rather was meant to be "I am unconvinced plate mail would offer greater protection than robes against fire, and have enormous difficulty imagining ring mail offering better protection than leather against flames".

And again, I'm not calling for change and saying the game should be more realistic. I am saying: this aspect of the system is unintuitive, here is my line of reasoning for how it qualifies as unintuitive, and Crawl is sufficiently 'life-like' in its strivings that a player is not going to assume the game is arbitrary on the topic, they are going to draw from their knowledge of the real world to guess at how the game works.

There might well be value to the idea of displaying different messages depending on how much armour mitigates damage, and if you feel strongly about it, I'd encourage you to start a new GDD thread that lays out the proposal in more detail.

I'm not sure how much detail there is to get into, but I'll work on that.

(I would like to be on-topic myself but can't think of anything to add)