3E/3.5 Multiclassing in D&D 3rd Edition

My best friend Rob Heinsoo was the lead designer on 4th Ed, and one of his jobs was to fix things that 3rd Ed hadn’t fixed. Multiclassing was on that list of systems that needed work. At one point when playing 3rd Ed, Rob was running a 3rd level barbarian-fighter-ranger. Given the way multiclassing worked, why not?

3ephp.jpg

Meanwhile, the barbarian-cleric I ran in the RPGA never gained a 2nd level in barbarian. Giving up cleric spells would have been too high a price to pay, and in fact the one level of barbarian that I had given this character was a nod to style and a tactical mistake. (Arguably playing anything other than a full-on cleric in 3rd Ed RPGA games was a mistake.) The Third Ed version of multiclassing “worked” in that you could mix and match as you pleased, but it didn’t really work in that most combinations were a mess. Multiclassing rules are a bitch.

When we started design on 3rd Ed, we knew that multiclassing would be an issue. The earliest takes were basically classes that combined the traits of two base classes, with a slightly steeper XP curve for leveling up. Theoretically, this system is like the Elf class in Red Box. The approach was solid in that it would have let us balance each “multiclasses” like we balanced the base classes. But this system seemed too limited for our purposes. Third Ed was about busting open limits, and combo class system seemed to make multiclassing more restricted than before. Today, after seeing the “mix-and-match” system in play for 20 years, I wonder whether we might have done better by developing that original system.

As it is, we got pretty far in the design process without solving the multiclass problem. In the end, I proposed more or less the current system, with levels from different classes stacking benefits on top of each other. The best thing about the system, I figure, was the concept of prestige classes. They were basically “multiclass only” classes. The prestige class concept was pretty exciting and made all sorts of interesting designs possible. And the beauty of the “libertarian” approach is that it required almost no work to balance. It wasn’t balanced.

One of the guiding tenets of the 3rd Ed design was “consequence, not restrictions.” It meant that we wouldn’t tell you that you can’t play a halforc paladin. Now halforcs have a Charisma penalty, so there will be consequences, but you can do what you want. This approach can be something of a disaster when it comes to making permanent choices about your character. And with the “anything goes” rules for multiclassing, there were more ways to build a weak character than to build a strong one.

On some level, balanced, anything-goes multiclassing rules are systemically impossible, and here’s a thought experiment to help you see what I mean. Suppose that the game designers hand-balance the base classes so that they play well next to each other. These base classes have the right power level and that right number of options: not too many or too few. That’s where you want the classes to be. Now imagine that you add on an algorithmic system for taking any two of those classes and combining them in any combination of levels. Maybe throw in a couple extra classes, up to as many classes as you have levels. What sort of “class” are you going to end up with when you combine different classes into one? The ideal result is that the character has more options balanced against less overall power. In addition, the increase in the number of options has to be modest enough that the player doesn’t get burdened by having too many. If you hit that ideal sweet spot that balances power with options, you’re lucky. Most combinations, especially with spellcasters, come with too harsh a penalty for the benefit. For others, like the fighter-ranger-barbarians, there was an increase not only versatility but also in effectiveness.

The multiclass rules are a dramatic example of how treating things the same is a mistake if those things are different. The rules allow players to mix and match classes in virtually any combination, as if the Nth level of any class is the equivalent of the 1st level (or Nth level) of any other class, even when combined. With this “wild west” or “libertarian” approach to multiclassing, combinations are bound to vary from weaker to stronger depending on how well the classes line up. Two classes that rely on Strength and Dexterity, like fighter and ranger combo up pretty well. But what about a Strength-based, heavily armored class with an Intelligence-based class with spellcaster that’s penalized for wearing armor? Any system that makes the fighter-ranger OK is going to be hard going for the fighter-wizard. If the game designers balance the system to makes the fighter-wizard OK, then the fighter-ranger is too strong. Those two combinations are quite different, so using the same rules for both of them leads to imbalance somewhere in the system.

To complicate things further, there were countless ways to combine two classes. If the fighter-1/wizard-9 is balanced, can the fighter-5/wizard-5 be balanced, and the fighter-9/wizard-1? Not really. There are so many multiclass options that inevitably most of them are going to be too strong or, more likely, too weak.

One problem with multiclassing is that classes came front-loaded with lots of great stuff at 1st level. If you’re a barbarian, the reasoning went, you want to be able to rage at 1st level. We toyed with the idea of giving each class a special feature that only single-class characters would get, but it was a new idea and it would have taken lots of work to get right, and we passed.

For 4th Edition, an overarching goal was to prevent players from making choices that led to them being disappointed. They headed off the problem of multiclass characters by not allowing regular multiclassing. A fighter could pick up some bits from the wizard class, and you could play a class built from scratch to be an arcane spellcasting warrior, but you couldn’t give yourself a bad experience by building a fighter-5/wizard-5.

For 13th Age, Rob and I forced a solution. For one thing, the rules support only an even split between two classes, reducing the complexity by at least two-thirds. The rules ended up somewhat resembling the AD&D multiclass rules, combining reduced-power versions of two classes. We also force every class/class combination to care equally about two different abilities. That way there’s no natural advantage for a combination of two classes with the same main ability, such as the bard-sorcerer, who needs Dex as much as Cha. Each class-class combination also got hand-balanced with power possibly adjusted up or down and special rules provided when necessary.

Fifth edition gets a lot of things right. It has some forms of “multiclassing” built into the classes, such as the fighter’s eldritch knight option, which is a nice touch and easy to balance. Fifth Ed also returns to the mix-and-match system, but they plug a lot of holes when they do. Many rules contribute to a smoother multiclassing system: ability minimums, limited proficiencies, more generous spellcasting, classes getting cool stuff at 2nd level, and the universal proficiency bonus. These concise, useful rules obviously come from people who played the hell out of 3rd Ed and knew exactly what was wrong with multiclassing. Even so, the various combinations all are going to work more or less well, and only some of those combinations can be balanced right. Spellcasters still lose out on their most powerful spellcasting levels, making it painful to multiclass with a non-casting class. Multiclass rules are a pain to design.
 
Last edited by a moderator:
Jonathan Tweet

Comments

TwoSix

The hero you deserve
In my 'stricter niche class system' the Eldritch Blast+Agonizing Blast combo would be exclusive to a 'Eldritch Sharpshooter' lv 1-10 class and be essentially a magical version of an Archer Champion. They'd have a small selection of other spells (and probably a utility cantrip to pick from) but most would be to support a 'magical sniper' play style and not to grant them more damage on top of the basic.
You should write some of this up...a system of more specific niche classes is right in my wheelhouse.
 

dave2008

Legend
Haha, yeah I don't like random stats, but I guess I'm willing to compromise a way to have both...
I personally don't care for rolled stats, but my players do. So for me, any stat generation would have to include a rolled and point-buy base.
Heh, I'm still not convinced we need the full scores. The only thing in 5e is the distance in feet you can jump and I think the ammount of weight you can carry is calculated on your Strength score? The biggest difference is when you end up with two odd scores and then bumped both up a Modifier point by assigining your ASI to the two of them.
Not for 5e, I was suggesting for a theoretical new edition. This would be designed from the ground up with the understanding that every point matters. It would be no different really from using just modifiers, it is just bigger modifiers an uses the traditional range.

How about we make it so, if you want to used roll stats you keep the ability score, but if you don't you just go straight for the modifier bonuses?

Basically you roll your stats a certain way (probably 3D6 but you reroll the 1s?) THEN all the +1s and +2 build up your stats on top of your lower score so their impact is still there but less impactful?

Let's imagine a combination of Race/Culture/Background/Class that ends up with my non-random character with those modifiers (with technical score in parenthesis) :

STR +4 (18), CON +1 (12), INT +1 (12), DEX 0 (10), WIS +2 (14), CHA -1 (8), Total: +7

Then I rolled stats a few times as I suggested above and applied the same bonuses and I got

STR 17 (+3), CON 12 (+1), INT 8 (-1), DEX 10 (0), WIS 13 (+1), CHA 8 (-1), Total: +5
STR 19 (+4), CON 12 (+1), INT 8 (-1), DEX 10 (0), WIS 12 (+1), CHA +8 (-1), Total: +6
STR 18 (+4), CON 16 (+3), INT 8 (-1), DEX 12 (+1), WIS 16 (+3), CHA 9 (-1), Total: +9
STR 16 (+3), CON 14 (+2), INT 14 (+2), DEX 10 (0), WIS 14 (+2), CHA 12 (+1), Total: +10
STR 19 (+4), CON 14 (+2), INT 12 (+1), DEX 12 (+1), WIS 12 (+1), CHA 10 (0), Total: +9
STR 20 (+5), CON 14 (+2), INT 10 (0), DEX 10 (0), WIS 16 (+3), CHA 10 (0), Total: +10

So that's a tendency to be TOO strong compared to generated, so I went and rolled without rerolling the 1s and I got

STR 18 (+4), CON 12 (+1), INT 10 (0), DEX 12 (+1), WIS 14 (+2), CHA 8 (-1), Total: +7
STR 18 (+4), CON 14 (+2), INT 14 (+2), DEX 10 (0), WIS 14 (+2), CHA 8 (-1), Total: +9
STR 20 (+5), CON 14 (+2), INT 15 (+2), DEX 14 (+2), WIS 7 (-2), CHA 7 (-2), Total: +7

Better but more swingy... Still work to be done but I think a little tweaking it could work.
I prefer a simpler approach. Just roll for a base score (3d8, drop one die, for a max of 16 or something similar) and then you add / subtract your race, class, background, & culture modifiers.
 

Undrave

Hero
I prefer a simpler approach. Just roll for a base score (3d8, drop one die, for a max of 16 or something similar) and then you add / subtract your race, class, background, & culture modifiers.
Okay so keeping the +4 to STR, +1 to CON, +1 to INT, 0 to DES, +2 to WIS and -1 to CHA, I went and rolled a couple of 3d8s, dropping the lowest, and I got the followng results:

9, 9, 10, 6, 14, 15
8, 12, 6, 13, 15, 11
3, 9, 11, 8, 12, 14
15, 7, 15, 15, 14, 14
13, 14, 16, 11, 9, 5

Not a big fan of this much swinginess. I didn't bother calculating the bonuses as a result. The third guy is pitiful and the fourth array is over powered.
 

FaerieGodfather

Aberrant Druid
Every D&D-like game has to provide a random option and a non-random option. Most players have a strong preference for one or the other, and there's no real compromise option that can be reached.
I'm fond of hybrid systems. For instance, combine a moderate to moderate-low point buy (25 in 3.X, 15 in PF) with the roll-in-order method of your choice: 4d6k3, or Xd6k3 with 9-12 dice to assign. You get reliable performance in your class-essential abilities, but the surprise scores that fans of random generation love.

If you want, you can make the reliable performance less reliable by including a mechanism for lowering the purchased stats-- say, -1 for every 1 on the d6s.

As for random ASIs... same thing. You use your random rolling method, with some kind of bonus for your class and level, and if you roll higher than your ability: +1. Has the unfortunate tendency of normalizing your abilities over time unless the random method really accounts for your primes.
 

TwoSix

The hero you deserve
I'm fond of hybrid systems. For instance, combine a moderate to moderate-low point buy (25 in 3.X, 15 in PF) with the roll-in-order method of your choice: 4d6k3, or Xd6k3 with 9-12 dice to assign. You get reliable performance in your class-essential abilities, but the surprise scores that fans of random generation love.
Hmmm....how would that work?
 

Lanefan

Victoria Rules
Given that you can self-regulate and that not every quick dip is for "munchkin" or "min/max" purposes, that line should be drawn where it is. If you don't want to "munchkin" or "min/max," then don't. Others enjoy that style of play and it should be open to them.
The problems arise when you've got those options available, and some use them to the fullest and min-max to the nines while others at the same table - for any of a bunch of reasons - don't.

There's three solutions (one of which really isn't), which boil down to:

a) find a way of boosting up those who don't min-max so they more or less match those who do
b) find a way of chopping down those who do min-max so they more or less match those who don't; the easiest means of which is to remove the options that allow for min-maxing
c) live with it.

I'll take option b) here every time. Option a) is the expressway to unbridled power creep. Option c) is just a punt.
 

Lanefan

Victoria Rules
Probably the biggest question for any random stat generation method is: Do you want the method to drive your character generation (like I rolled a 17 in Intelligence, I should play a wizard) or just to provide some organic changes to your already determined concept?
Depends if the DM or system allows the rolls to be rearranged such that the 17 doesn't have to stay in Int.

If there's no rearranging then yes, the rolls are largely going to determine in broad terms what you play. If rearranging is allowed, then the distribution will still have a say (you've less options with 13s across the board than you do with, say, 17-15-13-13-11-9 even though the average is exactly the same) but not as mich; and this middle ground is kinda where I like it.
 

Maxperson

Morkus from Orkus
The problems arise when you've got those options available, and some use them to the fullest and min-max to the nines while others at the same table - for any of a bunch of reasons - don't.
I can see that happening in an AL game or convention game. At my home games, though, I find like minded people to play with.

There's three solutions (one of which really isn't), which boil down to:

a) find a way of boosting up those who don't min-max so they more or less match those who do
b) find a way of chopping down those who do min-max so they more or less match those who don't; the easiest means of which is to remove the options that allow for min-maxing
c) live with it.
Or just don't care. I've never really cared if others at the table are combat beasts. That just meant that I would generally be better at the non-combat portions of the game and those were the most fun anyway.

I'll take option b) here every time. Option a) is the expressway to unbridled power creep. Option c) is just a punt.
It's not generally a punt unless you have envy of those who do better or get irritated at people who don't match up to you. Both of those attitudes aren't welcome at my table. Enjoy your character and let others enjoy theirs.
 

TwoSix

The hero you deserve
Depends if the DM or system allows the rolls to be rearranged such that the 17 doesn't have to stay in Int.

If there's no rearranging then yes, the rolls are largely going to determine in broad terms what you play. If rearranging is allowed, then the distribution will still have a say (you've less options with 13s across the board than you do with, say, 17-15-13-13-11-9 even though the average is exactly the same) but not as mich; and this middle ground is kinda where I like it.
Of course. Ultimately, you choose a generation method based on what play priority you're trying to cater to, the important thing is to recognize first and foremost what that desired play priority is.
 

Lanefan

Victoria Rules
I can see that happening in an AL game or convention game. At my home games, though, I find like minded people to play with.
I don't see it so muich in 1e-like games, as the min-maxing options are fewer; but I really did see it when I played 3e - a couple of players were quite dedicated min-maxers while the rest of us, not so much.

Which meant one or two characters became the keys to the party while the rest were, at times and despite our best attempts, glorified sidekicks.

Or just don't care. I've never really cared if others at the table are combat beasts. That just meant that I would generally be better at the non-combat portions of the game and those were the most fun anyway.
Fair enough, but if your PC is in theory supposed to be the party combat beast yet you're constantly being outshone in combat by someone else, it ain't working.

It's not generally a punt unless you have envy of those who do better or get irritated at people who don't match up to you. Both of those attitudes aren't welcome at my table. Enjoy your character and let others enjoy theirs.
I'm competitive by nature. I don't get irritated when others don't match up to me; instead I'm often the one who doesn't match up and I do get irritated sometimes by this.
 

FaerieGodfather

Aberrant Druid
Hmmm....how would that work?
First method is straightforward: you run your point buy first, assign your stats. Then you roll 4d6k3 for each ability in order, and you use the higher result for each ability between your point buy and your roll.

No swapping, no dumping-- you take what you bought, or you take what you rolled.

Second method, you have a number of d6s to assign-- suggest between 21 and 30-- and you assign each ability score between 3d6 and 6d6, keeping the best 3 as normal. (I cacked my math in the first example because I didn't include the base 3d6. If that's all you wanted clarified, I apologize.)
 

Maxperson

Morkus from Orkus
I don't see it so muich in 1e-like games, as the min-maxing options are fewer; but I really did see it when I played 3e - a couple of players were quite dedicated min-maxers while the rest of us, not so much.
Yes. I saw it much more in 3e than in any other edition. It just never bothered me.

Which meant one or two characters became the keys to the party while the rest were, at times and despite our best attempts, glorified sidekicks.
If combat is the goal, then sure. If not, not.

Fair enough, but if your PC is in theory supposed to be the party combat beast yet you're constantly being outshone in combat by someone else, it ain't working.
Sure. If you aren't realizing your character concept, whatever the concept, it aint working.

I'm competitive by nature. I don't get irritated when others don't match up to me; instead I'm often the one who doesn't match up and I do get irritated sometimes by this.
For me it's a matter of perspective. I have a character concept in mind when I make a character. If I am excelling at that concept, even if I don't match up in combat(not my concept), I am ahead of the rest of the party. Combat is only one metric that I measure by and it's rarely the central portion of my PCs.
 

TwoSix

The hero you deserve
First method is straightforward: you run your point buy first, assign your stats. Then you roll 4d6k3 for each ability in order, and you use the higher result for each ability between your point buy and your roll.

No swapping, no dumping-- you take what you bought, or you take what you rolled.

Second method, you have a number of d6s to assign-- suggest between 21 and 30-- and you assign each ability score between 3d6 and 6d6, keeping the best 3 as normal. (I cacked my math in the first example because I didn't include the base 3d6. If that's all you wanted clarified, I apologize.)
No, the Xd6 to each stat method I've seen. Do point buy, plus roll 4d6k3 and keep better I hadn't seen, that's pretty interesting.
 

TwoSix

The hero you deserve
I don't see it so muich in 1e-like games, as the min-maxing options are fewer; but I really did see it when I played 3e - a couple of players were quite dedicated min-maxers while the rest of us, not so much.
Hmm...so, if I combine this with your advocacy for random determination of stats above, does that mean you're advocating against system mastery as a virtue, in favor of random distribution of character effectiveness?

I'm assuming that's part of a broader pro-verisimilitude play agenda?
 

Lanefan

Victoria Rules
Yes. I saw it much more in 3e than in any other edition. It just never bothered me.

If combat is the goal, then sure. If not, not.

Sure. If you aren't realizing your character concept, whatever the concept, it aint working.

For me it's a matter of perspective. I have a character concept in mind when I make a character. If I am excelling at that concept, even if I don't match up in combat(not my concept), I am ahead of the rest of the party. Combat is only one metric that I measure by and it's rarely the central portion of my PCs.
I was using combat as an example, and could just as well have used "being the party face", or "best Wizard on the block", or whatever.

In that game I had three significant characters (along with far too many one-hit wonders!). The first was in fact intended to be the party fighting machine; that went out the window real fast when a) 3e's garbage multiclass rules reared up and bit me (I wanted him to have a tiny bit of Wizard to him for character reasons) and b) when the min-maxed party Cleric started putting me to shame at about 2nd level.

The second was intended to be a spinny fluffy attempt to make a 1e Illusionist in 3e; by sheer luck she worked out the best as I blundered into 3e's strong prefernce for hyper-specialization. She wasn't nearly as effective as many, and had the worst stats in the party (we rolled them all), but boy was she fun to play!

The third was a deliberate attempt to build another of my favourite concepts from 1e: a "heavy" warrior Con-based Ranger. Again the 3e system, which had leaned into the Drizz't style light Dex-based Ranger, fought me tooth and nail; and he ended up being thoroughly sub-optimal. Later branching into Cleric didn't help matters any.

Funny story: when I left that game I was replaced by a player who was a serious hardcore min-maxer; and as the party was in mid-adventure with no way for his PC to come in he was (with my OK) given my Ranger to play until that adventure was done and the characters could swap out. His reaction on reading through the character sheet, I'm told, was priceless; involving many "wtf?!"s and "what was he thinking?!"s; but on talking to me later and realizing what I'd been trying to do, even he grudgingly admitted I'd done the best I could with it.
 

FaerieGodfather

Aberrant Druid
No, the Xd6 to each stat method I've seen. Do point buy, plus roll 4d6k3 and keep better I hadn't seen, that's pretty interesting.
I don't know whose idea it was originally. Might even have been mine, but I'm not trying to take credit.

I am altogether fond of "use the better result from two different approaches", especially in Alternity where there are optional "random by species" and "random by profession" options-- which are each guaranteed to meet their respective minimums, but require some accommodation for species maximums.

I have been thinking a lot about trying to incorporate something like that into D&D...
 

Advertisement

Latest threads

In Our Store!

Advertisement

Top