Spellcasting vs. spell skills?

[wolloloo]

When good players give advice or write guides on spell casting characters, they often say things like "train spellcasting if you need more spell levels". I read that as "do not put skill points into spellcasting unless you are short of spell levels". Why do people consider spellcasting not to be a good investment in general? Spending XP on it has the same effect on spell success as training a skill of a dual school spell, and even a bigger effect on triple school spells. And it also increases the power of spells.

Example (assuming identical aptitudes for the skills):

1) Spellcasting 15, Ice Magic 10, Conjurations 10
vs.
2) Spellcasting 10, Ice Magic 15, Conjurations 15

(1) Costs fewer XP than (2), yields the same success rate for Ice/Conjuration spells and gives you an additional ten spell levels and more MP.

So, why do people often recommend to go for (2) and not (1)?

[eeviac]

Spell power is more heavily weighted on spell schools than it is on spellcasting and int. Your bolts of cold will be hitting harder in scenario (2).

[BlackSheep]

You're misreading the formula for spell success. Improving Spellcasting has the same effect as training one skill of a theoretical 4-school spell. Spellcasting also is more expensive to raise compared to a school skill of equal aptitude.

[galehar]

Blacksheep is right, spellcasting is worth 1/4 of spell skills. So even for a dual school spell, training spellcasting has half the effect that training one spell skill does. And spellcasting is a bit more expensive yes. You can reduce its aptitude by 1 to compare it to other skills.
eeviac is wrong, the weighting is the same for spell success and for spell power (still 1/4).
spellcasting also gives MP which can be quite important sometimes.

[eeviac]

I don't understand: "for a dual school spell, training spellcasting has half the effect that training one spell skill does" - doesn't this mean that spell schools contribute more to a spell's power and success than spellcasting?

[mageykun]

Spellcasting also contributes to cutting down on the hunger cost of spells (elemental schools do not). Getting the hunger on critical spells low enough that you can spam the spells you want without burning through your permafood is nice, especially if you're playing a race with food restrictions.

Of course, like everything else, it's a balancing act. Points you put into mp and cutting hunger take away from points you could have put into damage and spell success.

[BlackSheep]

FR: reduce spellcasting aptitudes by 1 instead of having a hidden penalty

Ditto Fighting. Anti-ditto Stealth.

[Southpaw]

It's not "exactly" a hidden penalty when the descriptions on the aptitudes page say "Please note that many things affect how quickly a character will actually learn a skill. Thus, the numbers below are good enough for comparisons among species, but not necessarily among skills."

I took that to mean "some skills take more or less experience to learn than others"

[BlackSheep]

Last I'd heard, a level of Fighting or Spellcasting cost 130% the XP of a standard skill, and Stealth was changed from 50% to 80%.

[eeviac]

isn't evo a cheap skill too?

and yes, I'm all for making all skills cost the same amount of XP, and adjusting aptitudes accordingly.

[zogre]

I have been laboring under the impression for years now that the reason spellcasting and related skills cost more to raise was one of balance - the utility of spellcasting is partially offset by a slightly increased experience cost - would that be readily replicated by simply adjusting all Spc and related apts down 1 or 2 steps?

[wolloloo]

Okay, I get it. I did not have the factor 0.5 that is applied to spellcasting before the calculations on my radar. This mistake made me play spellcasters badly for half a year.

By the way, I really _hate_ the secrecy the devs make about important game mechanisms and formulas. It would be so easy to add this information to the ingame help, along with weapon, spell damage, spell power and armour data etc. That wouldn't hurt or confuse anybody - actually the secrecy makes a lot of people play suboptimally. Sometimes I think this information is hidded from the players so that the devs who know all this can always play better than anybody else.

Note on skill costs: In 0.10.3 spellcasting costs 130%, stealth, invocations and evocations cost 80% and everything else costs 100% (skills2.c, line 497).

[mikee]

Last I'd heard, a level of Fighting or Spellcasting cost 130% the XP of a standard skill, and Stealth was changed from 50% to 80%.

Fighting is 100% as far as I know.

[wolloloo]

I have been laboring under the impression for years now that the reason spellcasting and related skills cost more to raise was one of balance - the utility of spellcasting is partially offset by a slightly increased experience cost - would that be readily replicated by simply adjusting all Spc and related apts down 1 or 2 steps?

One step is equivalent to 118.9% cost.
Two steps are equivalent to 141.4% cost.

So the actual 130% are somewhere in the middle.

[Kate]

I don't understand: "for a dual school spell, training spellcasting has half the effect that training one spell skill does" - doesn't this mean that spell schools contribute more to a spell's power and success than spellcasting?

Exactly, yes.

[XuaXua]

I put a proposal on the dev wiki long ago to remove Spellcasting by making it a generated value based on a factor of all known spell skill levels plus Evocations.

[galehar]

FR: reduce spellcasting aptitudes by 1 instead of having a hidden penalty

I agree with that. Special costs should go away IMO, even if it means that humans won't have +0 everywhere. Last time I brought it up, I think I met some opposition (from kilobyte maybe). Team has changed since then, maybe I should try lobbying for it again.

Sometimes I think this information is hidded from the players so that the devs who know all this can always play better than anybody else.

Don't be paranoid. Most best players aren't devs and most devs aren't even good players. And anybody can read the source. The reason some information is hidden is that we don't want to clutter the game with tons of irrelevant numbers and formulae. We try to show only what's really relevant, but of course, some may disagree about what is.
Also, when this kind of thing is brought up, we think about why this information is needed and how we could make it more intuitive instead of just showing it.

[wolloloo]

Don't be paranoid.

I'm not paranoid, I'm frustrated that it takes sooo much effort to get decent information about game mechanics.

And anybody can read the source.

Nope, the vast majority of people (i.e. players) can _not_ read source code. And of the relatively few people who can read code, most would be unable to understand the most important calculations, because - no insult meant, but it's true - the code is a big mess. Okay, it's probably historical, but I think a third of the code would vanish instantly if _all_ calculations were done in floating point arithmetic instead of integers (is there a _real_ reason why you cannot have 17.5 HP or 3.14159 MP?).

The reason some information is hidden is that we don't want to clutter the game with tons of irrelevant numbers and formulae. We try to show only what's really relevant, but of course, some may disagree about what is.

Hey, what would be so confusing about showing the actual values of a spell/weapon/armour when you use it?

Fireball, 25% failure, 75 power of a mximum of 200, range 7
Glaive +2/+3, 16-27 damage, 10-30 to hit roll, attack delay 120%
Plate Mail +1, absorbs at least 39% or up to 30HP, affects Evasion by -123 and reduces to hit rolls by -12

That does not mean you have to display each and every roll and clutter the whole display or explain the details of the formulas. Just a summary page for weapons and armour (+evasion, +shields) would be fine. It could list some details to help compare weapons and armour with each other so that people don't have to come to the tavern all the time and ask "should I prefer this +2/+3 Short Sword of flaming (SInv) a Sabre +1/1, my Short Blades skill is 10 and Fighting is 8?

Also, when this kind of thing is brought up, we think about why this information is needed and how we could make it more intuitive instead of just showing it.

I'd find it very intuitive if the game allowed me to see that on average I'd need three hits to kill that ogre with a given weapon. I had the Captain's cutlass in a game lately, and I think it's the one hand weapon with the highest potential damage in the whole game. But actually it did much less damage than a highly enchanted plain demon whip, and the game gives me no hint why that is. It probably has something to do with armour, but the documentation of that in the Wiki is practically nigh impossible to understand.

[galehar]

FR: reduce spellcasting aptitudes by 1 instead of having a hidden penalty

Done.

[Southpaw]

FR: reduce spellcasting aptitudes by 1 instead of having a hidden penalty

Done.

So Humans are now better/worse than Humans at very specific skills? Doesn't really change gameplay, but you should also consider changing how the aptitudes are defined (in other words, "0" can't be the Human standard if Humans aren't 0 in everything)

[Galefury]

That does not mean you have to display each and every roll and clutter the whole display or explain the details of the formulas. Just a summary page for weapons and armour (+evasion, +shields) would be fine. It could list some details to help compare weapons and armour with each other so that people don't have to come to the tavern all the time and ask "should I prefer this +2/+3 Short Sword of flaming (SInv) a Sabre +1/1, my Short Blades skill is 10 and Fighting is 8?

Diablo 3 does this quite well by simply displaying average damage per second. If you view the stats of an item, you are also shown how it will affect your DPS. You can expand the stats panel to display a huge list of stuff with a single click, but usually you don't need to see that list, so it is hidden by default.

Unfortunately, due to how armor works in crawl, damage per time can vary significantly between different enemies, so there is no single informative number to display. A weapon that is better against stone giants might be worse against blink frogs. Same for defensive stats: AC will be excellent against a pack of blink frogs, but against the stone giant EV and SH will suit you better (note that one stone giant gives twice the exp of a pack of four blink frogs, so the comparison between the two might be a little off).

For weapons at least delay, damage and to hit would need to be shown in crawl, which would quickly get impractical.

[wolloloo]

Well, a comparison page monster vs. character would be really cool: You press ctrl-x and select some monster in view, and from there you can switch to the details page. This page would list all known combat information about the character and the monster and show equipped weapon/bullet of the character vs. the worn armour of the monster and vice versa. This information is usually slightly incorrect because it does not include unknown information like equipment enchantments, brands, egos etc. And if you want to know if a different weapon is better, you'd have to switch weapons, so comparing weapons includes a risk.

[Galefury]

This is kind of available via fsim in wizmode. Can take a while, though, so it's not practical to use for anything other than a information gathering/debugging tool. It is rather difficult to calculate such numbers without simulation (which would be much much faster), and would be hard to maintain (would have to be updated whenever some combat formula is changed).

[eeviac]

Doesn't this mean that spellcasting is slightly cheaper to raise now, and stealth/invo/evo slightly more expensive? Because the old hidden penalties didn't quite match up with an even aptitude level.

Either way, I dig this change, and hope it makes it into .11.

[wolloloo]

Doesn't this mean that spellcasting is slightly cheaper to raise now, and stealth/invo/evo slightly more expensive?

Spellcasting is 8.52% cheaper now and stealth/evo/invo are 5.11% more expensive.

[zogre]

FR: reduce spellcasting aptitudes by 1 instead of having a hidden penalty

Done.

What's the harm of keeping humans at 0 and adjusting everyone else? Humans get a little better at spellcasting, a little worse at evo/stealth...

I guess it comes down to, which is more important - Humans with 0 apt across the board which keeps them as the metric to which all other races are compared, or keeping the aptitude balance between the races where it was before the change that removed hidden penalties/benefits?

I think I like humans as yardstick better

[sardonica]

the ONE time that the forums are used to make a change to crawl, and it happens in the Advice forums. Hahaha.

[evilmike]

I guess it comes down to, which is more important - Humans with 0 apt across the board which keeps them as the metric to which all other races are compared, or keeping the aptitude balance between the races where it was before the change that removed hidden penalties/benefits?

I think I like humans as yardstick better

Back when aptitudes used the finer-grained, percentage based system, humans didn't have completely "flat" aptitudes. Their spellcasting was 130, while their invoc/evoc were 75 (everything else was 100, although I believe stealth had a hidden bonus). Overall, most of the "hidden" stuff didn't become hidden until the current system was put in place.

Humans are still a "yardstick", if you want to look at them that way. It just happens that -1 spellcasting is considered average, rather than 0 (as it is for most other skills).

[sanka]

Or you can buff human spellcasting to 0 (previous +1). Humans are not weak, but this would not buff them to extremes, and few players play them anyway.

[BlackSheep]

This was not a change to balance the species but to make the skill training rates visible across the board. There's nothing thematic or helpful about having every Human apt be "0". They can be a baseline no matter what the numbers in the aptitude table.

[Southpaw]

I don't really care what the aptitudes for Humans are. I just don't think things like

"0 standard aptitude (Humans)"
" +/-4 learning twice/half as fast as Humans"
"0 is the Human standard"

should be littered around the aptitudes help page if it's no longer true in all cases. Truth be told there really isn't a reason for it to be there anyways because it doesn't mean anything. Hell, removing the word "Human" from the aforementioned lines still tells the reader exactly the same thing, but doesn't imply that any one specific species is being used as the meter stick.

[eeviac]

^that's true, it could be worded better. 0 is the normal learning rate, and just so happens to be (for the most part) the human learning rate. It makes more sense to compare aptitudes to themselves, than to compare them to humans.

[sanka]

I myself like these things that "humans have 0 apt for everything". Of course you do not even need a meter stick, but it looks nice.

Also it would not suggest newbies ideas like "humans are below average at spellcasting, so maybe I choose fighter instead of wizard".

[wolloloo]

Removing references to humans could even make playing them more attractive. Stressing that they are not particular good or bad at anything, but raise their levels fastest among all races might be an incentive to play them.

[XuaXua]

Mountain Dwarves never had these problems.