One of the things that I've always wanted to introduce in D&D are spell failure mechanics.

Why?

• I feel one of the reasons people bitch about the magic-user's near-invincibility at higher levels is the infallibility of magic.


• Magic being infallible does not reflect any fantasy fiction or real-life occult system that I know of. And even if it did, I still feel magic feels more magical (mysterious, unknowable, inescrutable) if there's an element of unpredictability and danger involved in its casting. I don't want to do away with so-called Vancian casting. If anything, I want it to feel really Vancian. And in both Dying Earth and Lyonesse, magic is a fickle thing, that can rub off on those who depend on it, and not always in a good way.
  
  
• Having a spellcasting roll allows you to insert all sorts of interest modifiers. In fiction and occultism, magical efforts are influenced by all sorts opf conditions: astronomical events, locations, material components, etc.

I am particularly interested in adapting DCC's mechanics for Spellburn (temporary sacrifice of ability score points for a spellcasting bonus) and Corruption (backlash and permanent consequences from botched castings).

As far as I know, of all the retro-D&D games out there, only Goodman's DCC and Jason Vey's Spellcraft & Swordplay use a spellcasting roll. And both do it similarly: the spell only fades from memory when the magic-user fails a roll. I think, but I'm not sure, that this is derived from Chainmail (which also makes me leery. If the rule's good, and it was on Chainmail, why didn't it make the cut to OD&D?). The cleric is even more radical: clerics never lose a prepared spell, only accrue a penalty of -1 for each previous casting (e.g. a cleric who's already cast 5 spells is -5 until he rests).

In theory it sounds like a fair balance against the now ever-present possibility of failing (or worse, catastrophically botching) a spell. But I am afraid it'll imbalance the game even further, especially as level goes up and the odds of successful casting increases; theoretically the magic-user can keep casting that spell until the cows come home, as long as he doesn't roll a failure. This would worsen, rather than fix, the issue with high-level magic-users being unbeatable.

For example, DCC uses 1d20 + Int modifier + magic-user level vs. 10 + (spell level x 2). This means a 1st level, Int 16 Wizard has a 50% chance of succesfully casting magic missile, which is fine and dandy... but it also means that the same magic-user at 10th level will auto-succeed except with a natural 1, effectively (95%) turning the spell into an "at-will" ability. At 14th level, the same happens with fireball and lightning bolt and God forgve me, haste. I'm not sure that's what I want for the game. Or rather, I'm sure I don't want this for the game.

I don't want to punish PC magic-users for solving problems with their trademark ability, but I sure as hell don't want them to have infinite spells at high levels either. I just want magic to feel a bit more unpredictable and dangerous.

One idea I've had is to use spell slots for memorisation only, and implement some "mana point" system in its place as casting units (with each level's worth of slots being converted into a point). For example: a 5th level magic user gets 2/2/1. Under this system, he can still memorize 2 1st level and 2 2nd level spells, plus 1 3rd level spell; it also means he'd have 2 * 1 + 2 * 2 + 1 * 3 = 9 "spell points", with each spell costing the same as its level (so fireball costs 3 points, invisibility costs 2 and color spray costs 1). A failed spell check means that he's lost the memorized spell, and needs to study it again, with or without rest. Depleted spell points are recovered with rest.

Thoughts?

I'm sure this could be fun in a number of situation, particularly if using some kind of DCC variant where the failure of the spell may have some weird, unintended effects, but effects nonetheless.

Now that said, my basic issue with a system of spell failure is that (1) it basically ruins the MU's resource management, or twists it in a way that I find counter-productive to the strategic intent of spell memorization. Basically, since you know that you've got say 30% chance of failing your fireball, you are all of a sudden tempted to memorize two or three fireballs that you spam until one succeeds. The net effect is that it alters memorization habits at best, or it makes the slot allocations tactically pointless, at worse. It's like playing Chess with random attack rolls when a piece tries to destroy another. Does that make the game strategically better? I don't think so.

(2) Spell failure already exists for the spells that actually require an anti-auto-win rule. It's called a Saving Throw.

I've tired a number of spell failure systems in TSR D&D and never found one that actually worked -- unless one defined worked as "totally gimped the Magic-User". Worse, they usually gimped the MU much worse at lower levels where the MU had few spell slots available.

The only one that I tried that came close to working well effectively replaced the Vancian system.  MUs no longer memorized spells, they could cast any spell they knew (had in their spell books). Spells cost spell points to cast. When a MU tried to cast a spell, he had to roll for casting success (inverse of spell failure). If the casting was successful, the spell cost twice its level in spell points. If the spell fails, its level in spell points were burned anyway. A magic-user had enough spell points to cast each of the spells he could have cast by his level in the D&D books once (assuming 100% casting success). A 1e first level MU had 2 spell points, a 1e second level mage had 4 spell points, a 1e third level had 8 spell points, a 4th level had 14, etc. IQ bonus (if any) added to spell points. Spell points recovered with a night's full rest.

Yep you may have gimping problems, at least unless you give PC wizards some compensating advantage as well - faster xp progression, slightly bigger hit die, or somesuch. TSR era D&D did have spell failure for priests (at least those with Wis <14 or so) but clerics also had better armour, weapons and hit dice.
Wild mages in 2E Tome of Magic had just a 5% chance of having their spells botch on them, not including odd variations in caster level, but got an extra spell of each level (being specialist wizards) and had their own unique spells.
 
As far as auto-casting goes, you could use something like d20+level+an Int adjustment vs. [10+effective caster of spell]. So 1st level magic missile (1 missile) would be only a -1 penalty and almost automatic for a 10th level wizard, while 9th level magic missile would be a -9 penalty and difficult to cast.
Presumably doing that, you would need to use a minimum caster level to cast higher level spells e.g. -5 to cast fireball since that's acquired at level 5.
 
If you wanted to prevent boning wizards entirely you could have their normal spells be reliable, but give PCs the option of casting extra spells with the casting roll, that they hadn't memorized - emergency spellcasting as it were. That actually makes wizards stronger, though.

I like the idea, but because I may be tempted to make magic risky & such on lower levels (to give an idea of wizard learning how to master his talent blah blah blah), not to "gimp" him so that Mr Fighter can show off. Mr Fighter doesn't get random sword failure on early levels, when he's better then wizard strictly mechanically, does he.

Quote from: Benoist;511790I'm sure this could be fun in a number of situation, particularly if using some kind of DCC variant where the failure of the spell may have some weird, unintended effects, but effects nonetheless.

Now that said, my basic issue with a system of spell failure is that (1) it basically ruins the MU's resource management, or twists it in a way that I find counter-productive to the strategic intent of spell memorization. Basically, since you know that you've got say 30% chance of failing your fireball, you are all of a sudden tempted to memorize two or three fireballs that you spam until one succeeds. The net effect is that it alters memorization habits at best, or it makes the slot allocations tactically pointless, at worse. It's like playing Chess with random attack rolls when a piece tries to destroy another. Does that make the game strategically better? I don't think so.

I am aware of this on a theoretical level. My explicit goals are (1) make magic more interesting, and (2) not gimp the magic-user while we're at it.

DCC counterbalances the chance for spell afilure by rewarding success with enduring memorization of the spell just cast. Which is cool, but has the (I have to belive) unintended side effect of making high-level casters potentially infinite sources of low-level spells, which might not be cool at all.

I was trying to reach a compromise when I came up with this:

Quote from: The ButcherOne idea I've had is to use spell slots for memorisation only, and implement some "mana point" system in its place as casting units (with each level's worth of slots being converted into a point). For example: a 5th level magic user gets 2/2/1. Under this system, he can still memorize 2 1st level and 2 2nd level spells, plus 1 3rd level spell; it also means he'd have 2 * 1 + 2 * 2 + 1 * 3 = 9 "spell points", with each spell costing the same as its level (so fireball costs 3 points, invisibility costs 2 and color spray costs 1). A failed spell check means that he's lost the memorized spell, and needs to study it again, with or without rest. Depleted spell points are recovered with rest.

Since this proposal is an attempt to address some of the same points Benoist (mantaining spell memorization as a resource management mini-game) and Randall (using spell points and flexible casting as a trade-off for the failure mechanic) raised, I'll try to do a better job of explaining it.

This system dissociates spell memorization from actual casting capabilities ("payload", as it were). 2/2/1 means you can memorize 2 unique 1st-level spells, 2 unique 2nd-level spells and 1 unique 3rd-level spells.

Casting capability or "payload" is defined by a "spell point" system, in which each level's worth of spell slots is converted to "spell points" on a 1:1 bases. And the point cost of each casting is defined as each spell's level. The 5th-level magic-user in the example has 9 spell points (2 * 1 + 2 * 2 + 1 * 3 = 9).

Now, once you've memorized a spell, you can cast it as often as you like... as long as you have enough spell points. The 5th-level magic-user in the example can use his 9 spell points to cast 9 magic missiles, or 4 webs and 1 color spray, or 3 lightning bolts.

This way, repeat memorization makes no sense whatsoever, and the magic-user gains in flexibility without losing actual casting power, and most importantly, without doing away entirely with the spell memorization resource management mini-game, and adding a new complication in the form of spell failure, complete with critical fumbles (and critical hits, while we're at it! What should a charm person critical hit look like?).

Now, my question: is this enough of a trade-off to accept a chance of failure? I'm inclined to say "no" (again, purely on theoretical grounds), and wonder whether an increase in the sheer number of spell points (perhaps approaching the casting capability of a 3.x wizard) would make it more balanced.

Of course, the only way I'll ever get it right is by playtesting, but still, I'd like to pick the board's collective brain on this.

Quote from: Rincewind1;511829Mr Fighter doesn't get random sword failure on early levels, when he's better then wizard strictly mechanically, does he.

He does actually, it's called "missing"

Quote from: Bloody Stupid Johnson;511827As far as auto-casting goes, you could use something like d20+level+an Int adjustment vs. [10+effective caster of spell]. So 1st level magic missile (1 missile) would be only a -1 penalty and almost automatic for a 10th level wizard, while 9th level magic missile would be a -9 penalty and difficult to cast.
Presumably doing that, you would need to use a minimum caster level to cast higher level spells e.g. -5 to cast fireball since that's acquired at level 5.
 
If you wanted to prevent boning wizards entirely you could have their normal spells be reliable, but give PCs the option of casting extra spells with the casting roll, that they hadn't memorized - emergency spellcasting as it were. That actually makes wizards stronger, though.

Both interesting ideas; I particularly like the first one.

Quote from: The Butcher;511841He does actually, it's called "missing"

And there are also Saves for the MU . Then again, as I said - I am not against spell failure. As long as you do it with a purpose of making the magic dangerous/risky, especially at lower levels, rather then gimping MU. The philosophy of subject approach, so to speak.

This is a thing where most of what you need are partial failures:

1 - Sharp pain: Lose 1 hit point. Spell works as normal.
2 - Sharp, lasting pain: Lose 1-4 hit points. Spell works as normal.
3 - Dizzy spell: Caster is stunned for 1-4 rounds. Spell is normal.
4 - Weakness: Caster suffers -4 to strength for the day. Spell = normal.
5 - Fainting: Caster is unconscious for the remainder of the round. Spell = normal.
6 - Spell veers off target. Use grenade-like weapons table to determine where it strikes.
7 - Spell operates in a different manner than expected. A fireball may be replaced by a lightning storm or a gush of water; a charm spell may enrage or confuse the target.
8 - Spell is an illusion and is dealt with accordingly.
9 - Spell operates at half strength, half duration, or +4 to save.
10 - 1 in 6 chance of attracting an extra-planar entity which arrives in 1d10 days. Spell has double the power, duration, or area of effect, whichever is least convenient.

And make it more likely as spells increase in level, so as not to further gimp low-level MUs.

Arcane Burn

When casting the spell (as well as when struck or damaged in combat) the caster must make a Concentration (or Willpower Check) against a DC.

Success means the spell goes off. Congrats.

Failure means you take Arcane Burn. That's 1d4 per level of the spell, no save (or a save, depends on how you want to do it) and you lose not only the spell in memory (for a wizard) or the ability to cast the spell for 12-72 hours (for a sorcerer), but you now have to make an additional save with the same DC. Fail that one, and you lose the ability to cast that spell forever. If you succeed, congrats. Now, if you failed that one, you make a third. If you fail again, you lose the spell slot itself. It's now filled with Arcane Burn, raw magical energy.

Now that you cast the spell, you get to make another Arcane Burn check, just like above.

If you are struck or damaged or otherwise distracted or inhibited during casting, you make another check. If you fail that check, you make an Arcane Burn check. If you succeed, well, you are still casting.

That's worked for me for over 20 years. I've even had players that deliberately failed a check so they had at least one spell slot with Arcane Burn in it, since Arcane Burn can be useful.

I played with many versions of this.

It might be one of the prime reasons I moved away from Vancian casting.  But when I was using it and spell failure, the point of the spell failure rules was to make higher power/complexity spells a little more challenging to cast (when the mage's power starts to tilt the balance a bit more).

no one wants a weaker low level mage.  

I think we finally came up with a rule in AD&D that used the (spell memorization % +10%) as the base success % to cast a spell.
that was modified down by 5% per level of the spell (3rd level spell=-15%, 9th level spell -45%)
modified up 5% for every class level above the minimum casting level of that spell.

we also had ritual rules that allowed the player to take more time or components (where appliable) for a stackable +5% each.

it was a decent compromise that allowed us to accomplish what we wanted, plus it allowed brighter mages a better success%, didn't gimp lower levels too badly, and simulated higher level mages getting better at lower level spells)
We also had a few funky slot rules where a caster could choose to memorize a lower level spell in a slot, which makes more sense in a spell success system.

So a starting mage at level one with a 16 IN had a ((65%+10%)-5%)=70% chance of success, +5% if they doubled casting time, +5% if they doubled components.

by level 5 for that same mage,
 first level spells are now the same 70%+(5%*4 levels)=90% chance of success
 second level spells are now ((65%+10%)-10%)+(5%*2)=75% chance of success
 third level spells are now ((65%+10%)-15%)+(5%*0)=60% chance of spell success

and all can have those 2 stackable 5% time and component add ons.

I really, really, really like your system.

I'll go look into the 1e spell learning percentages, and think it over.

Quote from: The Butcher;512757I really, really, really like your system.

I'll go look into the 1e spell learning percentages, and think it over.

http://www.mjyoung.net/dungeon/char/step018.html
well the top is a 19 INT with 95%, and it really does make a high INT mage a better caster.  There are secondary effects of this, such as the whole group combat dynamic changes when the 'artillery' is less dependable, but the 'preparation bonuses' allow for the careful mage to actually cast better.

We added faith-specific cleric components asn failure as well.  Clerics with lots of ritual items had a nearly zero chance of failure, but without them, they had as high a failure chance as mages.

we had other 'grittifying' houseules, as most games do.