Acceptable Spell Failure Rates?

[MainiacJoe]

I'd like to solicit discussion of when a spell's failure rate is "good enough" to stop training it. There is of course no one answer. Here are some of my thoughts to get things started. I'm confident these are mostly correct but feel much less confident that I could evaluate them in a given game.

1. Every casting attempt has a 10 aut* opportunity cost. (* !Haste)
2. Every casting attempt has an MP opportunity cost.
3. Emergency spells have higher opportunity costs than buff or killdude spells because of when you attempt to cast them
4. Every casting attempt risks* a miscast effect that varies in severity by school on top of the obvious spell level severity (* Sif/Kiku)
5. Having consumables on hand reduces the risk of miscasts because you can better respond to them.

So I feel that spell failure is more important early (E & B) but that is precisely when XP to reduce spell failure is at a premium. And this situation is why I'm making this post, since I don't really know what to do and not coincidentally, I rarely make it to Lair with casters.

[VeryAngryFelid]

I think the best approach is to compare magic skills with others. You should be asking questions what is more useful: more EV, more HP, more damage (UC for Tm, ShBl for electro dagger, M&F for plain flail when you run out of MP), more Evo for wands, more spellcasting for MP, more Invo for your god etc. and train what seems the best option.

[VeryAngryFelid]

Also early game gives great returns on damage for conjurations spell, I would still contniue training magic schools if I found 3 rings of wizardry on D:1 as OpCj

[4Hooves2Appendages]

Probably the biggest factor for me is weather spell power matters or not. If it matters, then I usually train it the spell to negligible fail rate. If it doesn't matter, then it's a relatively complicated judgement call. Lower for emergency spells like Blink and Swiftness and higher for utility spells like Ozo or DMsl.

[crawlnoob]

Deflect Missiles almost always drops while you are taking enemy fire, so C is not quite correct and with respect to the last post, I dont like to leave it floating around with for example > 10 or 15% fail rate if I can afford the XP.

All of this comes down to where you have spent XP, what you have to gain spending it on X vs Y, and what you can afford to do at the time.

Myself, if I am going pure blasty conjurations, I just pump spell schools until they are around 10 or so, then I'll give dodge and maybe fighting a few easy points until I get some gains, then back to pumping schools to the next plateau (maybe 14 or so depends), before I turn around and really take a look at alternative areas for XP. Basically, if you can dish out the damage and play cautiously, you dont need a lot of defenses.

[watertreatmentRL]

It depends on what the spell does. If it's direct damage, failrate comes out of your damage, so lower is better. These spells are generally spellpower dependent too, which provides some additional incentive to skill to a low failrate.

Duration spells often have little or no spellpower dependence (here I include tmuts, anything that produces allies, or other effects that persist over a number of turns). They can also be cast outside of combat or at least prior to engagement, which means there is often little or no penalty from failing to cast a few times. These spells can have pretty high failrates. I'm often happy with 20 or even 30%. (Or I was when I played dcss on a regular basis.)

That said, you don't want to have a realistic chance of getting serious miscast effects. Luckily, there aren't very many of these in the game and you won't run into them casting spells under level 6 with failrates around 20%.

[TonberryJam]

In my experiences with casters Getting it low as possible is the only exceptable goal. I've had issue where 10%-12% failure rate will fail two to three times in a row, but I barely ever see it happen at 5% or less and almost never twice in a row.

As far as getting past Lair. I've found that just about every starting book is siffucient to clear Dungeon, Lair and Orc. Wizard; for example; seems to have a more difficult time but not an impossible time at it.

Most casters can just train spellcasting and schools of choice and be fine without anything else if you lean on the spells effectively.

If you want more practice with casters, try an EE.

1. Sandblast. Rely on it until you run out of stones. Conserve some ammo if need be, by using melee on weaponless foes without a branded attack. Or just general units you know to be popcorn level threats. It's great for tough things like bees as an example.

2. Passwall. A great spell to escape stuff like packs if you like breaking them up. Safe play means you can learn it last.

3. Stone arrow. Replaces sandblast once it's strong enough or if you're low on rocks.
Note: Sandblast is incredibily useful for a long time. Lean on it when MP is half or less. Early-Mid game.

4. Petrify. This is a magnificient spell. Petrify one unit in a pack then use LRD to hit the pack. Use it on lone unit that is threatening, then stab it with a dagger rapidly or smite it with LRD. Double cast it to make it fully stoned as a useful escape option like blocking a path or a step into stabbing it.

5. LRD. The key spell to doing well as an EE. Using it every time an enemy unit is next to a wall should be like second nature. Don't worry about the noise. LRD can easily be use for smite targeting.

I find it most effective to focus spellcasting and earth for a long while to get the best out of this starting background. When you change this depends on race/class combo and what the game throws at you.

Getting the other earth spells like statue form and iron shot are a great idea. And this background works well with learning other transmutations like irradiate; which I will write is a key skill and will make your caster life a lot more simple.