Miscast risks as a balancing system for magic

[deaddmwalking]

There an implementation of Wild Magic that's not dumb then?

No, there's not.

Wild Magic is supposed to be unpredictable, but no matter how many d100 tables you create the effects always fall into a few categories:

1) The spell is not at all effective
2) The spell is somewhat effective
3) The spell is normally effective
4) The spell is more effective than expected
5) The spell is significantly more effective than expected

What descriptive effects of 'butterflies appear and fill the spell area' do is provide an explanation for why it falls into one of those categories, but using them straight up and then adding appropriate flavor would probably be better - possibly along with two axis of choice.

Ie, even if the spell is normally effective, it could be because a different spell is cast - likewise a replacement for disrupt undead might not target a undead creature so fall into the 'not at all effective category'. So in addition to determining how well the spell works you can add a semi-random variable about determining what makes it look that way, but choosing an appropriate explanation within the guidelines would work better than randomly determining an effect and then figuring out how that interacts with the game.

[Foxwarrior]

Your mathematical reductivism reasonably accurately models the effectiveness of wild magic. I hate it deeply, because it entirely misses out on the actual point Omegonthesane correctly identified. Wild magic is about the dice telling you jokes, the new effect it gives you is the joke, and figuring out whether it's better or worse than what you intended is the punchline. Obviously dice aren't that good at telling jokes, so rolling on those tables had better be fast so we can gloss over the duds.

So yes, wild magic is dumb, it's supposed to be dumb, and removing the dumbness just makes it boring and bad instead.

[jt]

You could significantly improve wild magic by moving it from output randomness to input randomness. Instead of pointing at a target and getting a random spell effect, you get a random spell effect then point to a target. Or have a "hand" of three random spell effects that you're stuck with until you use one, then it's replaced with another random spell effect.

I don't think this really appeals to the kind of player who wants wild magic in the first place though.

[Krusk]

You can tell wild magic is universally terrible, because without fail, there are always rules to remove the randomness and let players pick their outcome or negate bad ones.

I think you could genuinely replace wild mages with a sorcerer style class with refluffed spells. He doesn't cast fireball, he casts "Disco Ball. He launches a small pea sized disco ball at his targets, and it explodes like a pinata. Everyone in 30ft has to make a reflex save or take 1d6/level flaming debris damage. roll 1d6 to determine if the pinata erupts with flaming candy, toys, paint, glitter, glowing lights, or roll twice!"

Mage armor becomes "Chaos distortion field. You gain a +4 armor bonus to your AC for 1 hour per level, as the you surround yourself with a chaos distortion field. Attacks that miss you by 4 are intercepted by a random coincidence. The attacker stumbles, a branch falls from a tree, you bend over casually and the attack goes overhead, or whatever similar effect you can come up with".

[Dean]

I actually really like DM and Krusks take. A wild magic system that told people "you can't plan your turn, you'll get a spell and then throw it and sometimes it will rule and sometimes it will suck but it will be a real spell" would actually be the only wild magic system that anyone would ever like.

I disagree that shit like wild magic is "a joke". A joke requires a subversion of expectation. By being the guy who rolls on the d100 chart full of gender switching and fart sound spells you aren't surprising anyone. You're the dumb asshole laughing about the Knights who say Ni who came to the game with a character intended to make fart butterflies.

If you made Wild Magic something that felt like wild magic in any piece of fiction, like chaos magic from the Vlad Taltos books maybe, where it could theoretically do anything but fuck knows what it's gonna do right now you could set a base tone that would enable you to tell a joke. You could make a character who's grabbing wild arcane power they don't really comprehend or control who's throwing Acid Fireballs and Divine Powering themselves. That guy could look like someone and when 3 session in they roll a couple nat 1's and make some fart butterflies you would actually laugh. It's funny when the Pope slips and falls, he has status. If a grown man in a caked adult diaper falls down it's not funny. He has no status and therefore he cannot fall.

I like the idea of wild magic that gives you some random but consistently powerful shit that you make due with as best you can. Bonus spell power for the inability to plan your turn would be a huge draw to the sort of person who wasn't gonna plan shit anyway. If that person tries to Enlarge someone and makes them grow tits instead 1 of 100 times then I could see people liking that.
I know no one likes it now because, as krusk said, I've never seen anyone use it without making every possible effort to remove as much of the wacky randomness out of it as possible.

[Omegonthesane]

Vidjagame example!

Many summoning effects in Dungeon Crawl Stone Soup summon a random thing from a list determined by the effect. It's never as wild as a traditional Wild Magic table, but you won't necessarily know which kind of imp you're summoning for example, or which variety of ghosts will appear to haunt a target.

The example that made this enter my head was the Chaos god Makhleb from the same game, whose defining abilities are to call randomly determined destruction upon your enemies (you pick a direction, but don't get to pick if it's acid or fire or cold or poison or a necromantic spell) and random demonic summons. In both instances you know enough to know the effect is "hurt target" and "summon ally" while still introducing a degree of randomness.

From the same game there's the other god Nemelex Xobeh, which gives you the deck of destruction, the deck of escape, and the deck of summoning. So you know vaguely what you're in for but not the specific form it'll take.

Both of these are ofc intended to supplement or be supplemented by other, less random means of problem solving, but that's partly because it's a single player roguelike instead of a party TTRPG.

[The Adventurer's Almanac]

As someone who frequently rolls Makhleb and Nemelex characters, I can vouch for Omegon's claims of them being fun and tolerable 'random' mechanics. Nemelex in particular is interesting, because you can really get yourself into and out of some shit. The effects are actually useful, too. In a sense.

For example, the Deck of Escape. This sounds good, but here's what it actually includes:
[*]A smokescreen
[*]A random heal
[*]Sending a monster and/or you into the fucking Abyss with no way out
[*]Opening a hole beneath your feet to drop you several levels lower where stronger monsters live
[*]Surrounding you with impenetrable, temporary rock
[*]Makes you run faster

You may note that, at best, only two of these are really something you want most of the time, but the others are technically useful enough to save you sometimes. The Abyss and hole ones are particularly undesirable, but even then if you're a high-level character or have other methods of escape, then even those aren't too bad. The other decks are like this too - they're all generally useful, while some are just better than others. It appeals to my inner gambling addiction.

The important part is that the player has methods of manipulating their draws, too: Eventually they can draw 3 cards and pick 1 they want, deal 4 cards immediately as an "oh shit button, and draw 5 cards from any deck and rearrange them into a new, smaller deck of their choosing, called 'stacking' the deck. Given that the player knows they'll get cards eventually, but don't have a guarantee of when, they'll not want to spam cards in every situation, but they're not so stingy that you'll avoid using Deal 4 when shit hits the fan. Since each deck has a limit to the number of cards in it, you can't horde them too much, either. All-in-all, it's pretty neat and probably something you'd want to emulate for some sort of chaos magician or whatever.

[OgreBattle]

My vague memory of Chrono Cross is that casting spells of certain colors powers up that element in the battle.

I could see a kind of chaos magic with some control happen where certain color levels are hit, or if two colors clash like casting fire after lots of water spells has a chance of causing mist, while casting fire after fire after fire increases chances of pillars of flame staying around or fireball size increasing unpredictably

[deaddmwalking]

Reviewing the 2nd edition Tome of Magic, when casting a spell, the caster rolled a d20, which determined the effective caster level for the spell. At each level, there was a 5% chance that the result included a 'wild surge' random effect. If you were level 1 and you rolled a +1 to your effective level, there was a chance you had a wild surge. If you were 9th level (or higher), a wild surge only occurred if you rolled a '-1' to your effective level.

As a result, you'd get a wild surge 5% of the time - sometimes when you were casting at above your level, sometimes when you were casting below your level but only when you rolled 'the magic number' based on your current level. Technically, you only roll if there is a variable level-dependent effect, but any 'range' spell actually qualifies (even if the range is immaterial).

In that case, you then roll a d100 and you get an entirely random effect, most of which are pointless or annoying, but they're in addition to the spell's normal effects (adjusted for the modified level).

Examples:
10 - Caster's face is blackened by small explosion
41 - One magical item within 30' of caster (randomly chosen) is permanently drained

Riffing off myself, putting the 5 categories on a 2d6 where 2-3 is ineffective, 4-5 is less effective, 6-8 is normally effective, 9-10 is increased effectiveness and 11-12 is greatly increased effectiveness seems approximately correct in terms of probability distribution

Odds of Spell Power Effect - rounded to the nearest whole percent
8% 1) The spell is not at all effective
19% 2) The spell is somewhat effective
44% 3) The spell is normally effective
19% 4) The spell is more effective than expected
8% 5) The spell is significantly more effective than expected

Unlike the 2nd edition table, I think it would be appropriate if there was a chance for a 'random effect' whether the spell was a higher level or lower level. The easiest way to do that is to package it in with the proposed roll for spell level (just as they did in 2nd edition) but not require a table lookup. In my mind, the easiest way to make that happen is to have an easy rule that a player can keep in their head - I propose that 'doubles' are special - that is rolling a 2 (1, 1) or an 8 (4,4) would result in a wild-magic effect but rolling a 3 (1, 2) or an 8 (5,3) would not.

So using the categories above, if you rolled a 2-3 (not at all effective) there are 3 possible combinations on 2d6 (1,1/1,2/2,1) and only the result of 2 is a double. This would get us the following chance of a surge based on each category:

Odds of Wild Surge - rounded to the nearest whole percent
33% 1) The spell is not at all effective
17% 2) The spell is somewhat effective
13% 3) The spell is normally effective
17% 4) The spell is more effective than expected
33% 5) The spell is significantly more effective than expected

I'm pretty satisfied with that - more extreme results are more likely to have an unpredictable effect.

Diving in in more detail, imagine that you rolled a 'normally effective spell' with no additional effects (which happens when you roll a 6-8 without rolling doubles - which is 14/36 possible rolls or about 39% of the time - in that case it really is just the normal rules - you get the spell you tried to cast with nothing else happening.

If you rolled 'normally effective spell', but you had a wild magic effect (which happens in 2/36 possible rolls or about 5% of the time) something 'amazing' happens that doesn't increase/reduce the effectiveness. A short table of possible effects might be [roll 1d6]:

1) A harmless cosmetic effect is applied to [roll 1d6] (1, the target of the spell, 2, a random target within the spell area, 3, the caster, 4, the caster and allies, 5, all creatures in the spell range, 6, the area of the spell)
2) A different spell from the same school/level that would be approximately as effective is cast instead (DM choice)
3) A different spell from the same school/level that would be approximately as effective is cast instead (Player's choice)
4) A different spell from a different school of the same level that would be approximately as effective is cast instead (DM choice)
5) A different spell from a different school of the same level that would be approximately as effective is cast instead (Player's Choice)
6) In addition to the spell effect, a minor detrimental environmental effect is applied to the area of the spell (ie, a layer of grease coats the floor)

The same table could be used with the other categories (ineffective/very effective) with the spell effective level being decreased/increased by -2/-1/+1/+2. That is, if you cast a fireball as a 5th level caster and rolled a 12 (6/6, significantly more effective) you would be treated as a 7th level caster (increasing the numerical effects of the fireball), and the spell would be treated as a 5th level spell for wild magic effects (ie, if you rolled a 5 on the surge table you would pick a non-evocation spell from the 5th level spell list that would actually be MORE effective than the spell you intended. Ie, instead of choosing to cast Fireball, the player chooses a 5th level spell they want instead (like Dominate Person).

Having a random table of cosmetic effects and area effects would be helpful options. Some of those might have consequences - ie butterflies filling the area would at least temporarily create concealment, but those types of effects would generally be random enough and infrequent enough to give the flavor of wild magic in a consistent and fun way.

Recap
You're a wizard just like a normal wizard. When you cast a spell, you roll 2d6; there's a 30% chance that your spell is less effective, 30% it's more effective, and 40% chance that it's normally effective, plus a small chance of an additional effect (including potentially getting an entirely different spell of a higher or lower level).

[merxa]

one issue with wild magic is speed of resolution, besides the joke d100 tables, just rolling the various requirements and possibly adjudicating contextual problems, adding in additional spells etc, tended to cause the game to crawl.

rolling randomly to determine power seems ok, but it might be more interesting if the character has some input on what they are trying to achieve.

odds of wild surge table, I'd probably want a different probability threshold
50%) spell not effective
20%) spell semi effective
0%) spell normal effective
20%) spell more effectively
50%) spell extremely effective

replicating rando d100 tables is troublesome, they are arguably the draw for why someone might want to play a wild mage, and while we can argue d100 tables are terrible, yet capturing what is exciting about them seems important.

Instead of introducing different spells, the randomness should apply to the spell being cast. Maybe a roll to determine what about the spell changes and how, things like duration, range, area, maybe even things like school. Maybe that evocation firewall spell suddenly came from the divination school, look at the first table -- spell is more or less effective, and decide what it does.

But adjudicating all of these, or rolling on various tables is just too slow for reasonable resolution.
~

How about:
Player announces intended effect, trying to cast x normally, or at +1, -1 etc.
make a control check, this control check should be calibrated and bonuses tightly controlled, but basically make it a character level check vs DC, and adjust that DC up or down depending if caster is trying to over or under cast.

For example, using 3d6 we could set the default control dc for a level 1 spell at 11, and someone would be rolling 3d6 + level (3d6+1)


~9% 4-7 Spell very ineffective, major mishap
~28% 8-10 Spell less effective, minor mishap
~25% 11-12 Spell normal, no mishap/surge
~28% 13-15 Spell more effective, minor surge
%9 16-19 Spell very effective, major surge

attempting to upcast would move the DC up by 1 per level, down casting down by one per level. Such a check and table could be written:
DC = 9 + spell level +/- over/under cast
DC -4 or greater, spell very ineffective, major mishap
DC -1 to DC -3, spell less effective, minor mishap
DC, DC+1 spell normal, no mishap/surge
DC +2 to DC +4, spell more effective, minor surge
DC +5 or greater, spell very effective, major surge
~

With just one table, now it is just to adjudicate the surge/mishaps. This is where the rando d100 tables would show up, but I'd rather have something that lets people sort of simulate d100 tables on the fly.

This could be either a table with outcomes, that lets DMs and PCs fill in how that occurs, or a procedural table that makes certain changes to a spell, and leaves the DM and PCs to adjudicate how that plays out. The first one, rolling to determine an outcome seems both easier to balance and faster to resolve while allowing people to provide impromptu explanations for how that occurs, and of course examples would be provided.