D&D General Why is the multi-classing spell slot math so weird?

[Silam]

Getting to round up as a single class character versus down as a multiclass character is likely intentional. The point that it makes multi classing weird with certain combinations is a side effect that was likely considered a small trade off for favoring single class characters.

Personally, while I used to see multiclassing as essential to the D&D experience (especially pre-3e), I've come around to still loving the concept, but wanting to pretty much eliminate the mechanical implementation. My players almost never use it anyway. We had a rogue/monk once in a 20th level one-shot. We had a character that started as a 0th level fighter plus 0th level wizard before switching to some homebrew gestalt multiclassing I didn't end up liking. In our main campaign our rogue had taken a level of sorcerer. And I had actually suggested to my friend that his bard should take 1 level if wizard to do what he wanted to do.

I eventually decided that anything that someone wanted to do conceptually with multuclassing (as opposed to purely mechanical intetactions) either could already be done with the rules, or I would make rules to allow it.

Many classic combos work more or less out of the box or with feats in 5e. Anything + rogue works with just skills, tools, and proficiencies unless you see sneak attack as essential to the concept. Fighter + cleric is as easy as War Domain (or another heavily armed option) + cleric. A cleric/wizard can be one of the classes + magic initiate and ritual caster for the other (doesn't work for all concepts).

The rogue player just decided she wasn't satisfied with the dip and retconned it to the Magic Initiate feat (which I buffed to also allow usage of magic items as spellcasting foci as the class).

I made an Arcanist's Spellbook feature for the bard that lets arcane casters trade two of their known spells for flexible spell slots they prepare from a book like a wizard. That killed two birds with one stone, since it seems hardly anyone plays prepared casters in my group, and I miss seeing spell preparation.

Fighter + wizard works with Eldritch Knight or Bladesinger if you want to lean strongly to one class, but not so well as a more evenly mixed hybrid. After the unsatisfying experiments over the years I finally came up with Warrior-Mage class that I'm really happy with. Full caster without the extra features of a wizard, d8 HD with medium armor and martial weapons. Extra attack, War Magic, and then mostly subclass features. The three subclasses are Elven Fighter-Wizard (for the most iconic MC combo), Gish (for actual githyanki gish or githzerai zerth feel), and Planetouched Champion (for that Planescape fighter-mage aasimar, tiefling, and genasi feel, and a dash of the 3e planar champion prestige class).

If someone ends up with a multiclass concept they like that isn't well supported in the rules (and feels appropriate to me) I'll just make a feat(s), a subclass, or if absolutely needed a full class (though I can't think of any others that are really necessary for classic combos).

These are interesting thoughts...

I do agree that Magic Initiate goes a long way. It's a pretty powerful feat. And especially with some spells being augmentable, such as Cure Wounds, it can go a long way if you have slots from another class. And on top of it, it allows you to get those MI spells keyed off of whatever mental stat, so in that sense it's actually even more powerful than multi-classing...

Regarding the invention of new feats and classes, I personally feel like that's a bit dicey. Not to say that everything in the splatbooks is perfectly well-balanced. In fact, there is clearly a commercial incentive for any additional player option to be gradually more overpowered the longer after the start of a given edition the splatbook comes out. But... even if the splatbook rules aren't perfectly balanced, they are nonetheless often not too bad, and it's pretty difficult to balance homebrew stuff correctly. But it certainly is feasible and if the DM/table likes it this way, then it's totally fine, of course.

Overall, I feel like 3.5e had the best multi-class rules (except the oddity of 1st class levels giving big boosts to saves, which was a bit too powerful, but besides that the rest worked pretty well I think). 5e is pretty good overall, in my opinion, but its multi-classing rules are a weak spot. They don't seem very well-thought out to me.

But oh well... it's fine

 

[TwoSix]

Don’t they all round up now (and therefore why rangers and paladins have spells at level 1)? The artificer used to be the only one that rounded up on multiclass, but don’t they all do so in 5.24?

I believe that to be the case for those using the specific 2024 rule set. Couldn't say for sure.

 

[Charlaquin]

I believe that to be the case for those using the specific 2024 rule set. Couldn't say for sure.

I can, I just looked it up. Half-casters round up and 1/3 casters round down like I said.

 

[Sorcerers Apprentice]

I think the most precise solution would be keep fractions when adding the caster levels, and then finish by rounding to the nearest whole number.

 

[tetrasodium]

Consider Alice, a character who is an Eldritch Knight 7. They get 4 level 1 slots, and 2 level 2 slots. If that character then becomes a Wizard 1, the multi-classing rules say that their slots are now determined by their full Wizard level + a third of their EK level (rounded down). 1 + 7/3 rounded down = 3; which provides 4 level 1 slots, and 2 level 2 slots. And therefore, their Wizard level, which is a full caster class, gives them exactly 0 extra slots.

Consider Bob, another character who is an Eldritch Knight 9. They also get 4 level 1 slots and 2 level 2 slots. If that character becomes a Wizard 1, then the formula gives us 1 + 9/3 = 4; which provides 4 level 1 slots and 3 level 2 slots.

I'm not looking for an explanation of what rounding down means. Obviously I know that. I just find it weird that if we considered the "1/3 rounded down math" on the pure EK7 (Alice, before multi-classing) in isolation, we would arrive at the result that they are a 2nd level caster, and therefore get 3 level 1 slots and 0 higher level slots. But instead of that, the pure EK effectively rounds up their caster level to 3 for the sake of their slots (and spells known). So going from the pure EK to the multi-classed EK/Wizard, the contribution to spell slots from the EK transitions from rounding up to rounding down.

It's not a big deal per say. I don't think it breaks balance or anything. It's just weird that taking a level of a full caster class does not actually improve the character's ability to cast (beyond extra spells known/prepared) for Alice, but it does for Bob. It's also weird that, in a sense, becoming a wizard results in a slight nerf to what the EK in isolation contributes to the build (due to the rounding direction getting flipped).

The phenomenon can be explained away, e.g., "The years of learning magic as a hobby made Alice sloppy, and some of the shortcuts and assumptions accumulated along the way had to be unlearned when wizardry became their main pursuit. Bob on the other hand had no such trouble and the transition into wizardry as a focused endeavour came quite naturally."... but like, why is Alice getting shafted while Bob is just fine? It's just weird, IMHO

Cruft leftover from trying to bolt a couple big changes onto spellcasting and multiclassing as if they were smaller changes than they really were. In the past there were two big things that came into play with all of the situations you note
* Firstly was the fact that spells known spellcasters (ie bard sorcerer etc) got quite a bit more spell slots but fewer spells and could cast flexibly from those spells while prepared casters (ie wizard cleric druid etc) got more spellslots than 5e but had to prepare each spell for slot. That resulted in both more spells to cast but a high likelihood for some of those spells going unused at day's end.

* Secondly was that multiclassing spellcasting classes kept the spell selection and spell slot pools separate rather than stacking like 5e does(although I think there were some feats & PrCs that allowed limited mixing at the cost of taking them instead of something more laser focused.

Having them prepared slot by slot and keeping the selection separate avoided a lot of the multiclass cheese that 5e creates

 

[cbwjm]

I remember seeing this back in early 5e and didn't like it, if we weren't all using dndbeyond I'd have just stacedk caster levels so a level 7 EK with caster level 3 adding a level of wizard would have caster level 4.