Class Design Philosophy

This is an essay on how I prefer to design classes and subclasses for 5e. Do I follow it 100%? No. Is it perfect? Hah no. But I think it's tremendously better than most philosophies I've seen.

When I'm homebrewing elements (classes, spells, etc.) for my D&D games or even just trying to understand rules better, I look for underlying principles. And I try to draw from other aspects of life that have faced similar issues and apply the principles (or the analogous ones) they come up with where they make sense. When reading rules, I reach for the canons of construction of legal analysis, where contrary to common belief, "rules lawyers" and "RAW" don't actually triumph based on clever word parsing. But this essay isn't about that. It's about class design. As well as subclass design. And for that, I reach to software design, my day job. Here are a few of the principles I've pulled from there, as long as others I've devised myself (or stolen shamelessly from others--I'm a magpie by nature). None of this is truly original, but it works for me.

Thesis Statement (aka TL;DR)

Homebrew classes and good 3rd-party classes and subclasses share common principles.

1. They have a clear thematic vision. An elevator pitch, if you will. A short statement of the core aesthetic, why they adventure, and how.
2. They have thematic, unique mechanics. Not everything has to be unique, but they should have at least one thing that screams "I'm a <class name here>" every time they use it.
3. Their mechanics and thematics should be tightly linked--in computer speak, this is "high coherence". Everything in the class should play towards the theme, without feeling bloated or there just to fill space.
4. Their mechanics and thematics should be self-contained to the degree feasible. In computer-speak, this is low coupling. You shouldn't need 57 splats or cobble together a multi-class/feat-dependent build to have a competent character (mechanically). You shouldn't need specific item support just to make it work. Similarly, the thematics should be explainable with minimum out-of-world references--"Just like <superhero>" means you're now coupled to that character's concept and discrepancies will cause jarring dissonance.

Vision Statement

This is probably the most important thing. If, when writing a class, you can't come up with a one-paragraph, 3-5 sentence description that encapsulates the class fiction and provides a hook for someone to go "hey, yeah, that sounds cool. I want to play that," go back to the drawing board.

An example of a class just bursting with vision is the 2014 paladin. You can sum up their vision statement as follows:

A paladin is an expression of the Knight in Shining Armor archetype. Empowered by their unwavering confidence in the rightness of their cause, they wield holy power to defeat evils face to face, weathering their blows with holy magic and steel. If they falter in their confidence, their power wanes.

This, to me, provides a clear mental image. This is a Holy Warrior, an up-close magic-and-metal warrior whose life revolves around a Cause (with a capital C).

On the other hand, try coming up with one for the 2014 ranger or fighter. Do so without referencing mechanical elements like "makes a lot of attacks" or "can take two actions." And without referencing another specific fictional character (*cough* Aragorn *cough*). It's not impossible, but it's a whole lot harder. And it shows in their class design. Especially the ranger, who has been in search of a good class vision this entire edition.

One other key aspect here is that it shouldn't step on other classes toes. So saying that the wizard is "the quintessential arcane magic user" cuts into, well, just about every other arcane magic user's territory. Being specific is better than being vague--how am I to know I want to try that class (without reading through all the abilities and parsing all the options) if you just say "casts spells out of a book and is smart?" That covers a whole lot of ground.

So, point number one. Have a specific vision for what the class is that you can explain in a single short paragraph, without referencing mechanics or outside fiction.

Unique Cool Things

Once you have a vision, you need mechanical support. You need abilities. Abilities that reinforce that vision. If the paladin is a holy warrior powered by supernatural-strength confidence, giving them powers involving deception, lies, and betrayal probably isn't the best idea. Giving your tree-hugging nature-lover class abilities that call up ravening demons is probably not the right thing, but giving them ones that grow flowers is probably good.

But not just any abilities, even thematic ones. One of the key ways I depart from the current WotC vision is that, to them, everything is a spell. Everything except the most mundane stuff. And spells are generic. They're external--they're not in the class entry, they're in the back mixed in with all the other not-directly-class stuff. So giving your class spell slots and spells doesn't make it an interesting one, even if those spells are powerful or even thematic. Because spells, intrinsically, are a fairly generic, bland thing.

You want abilities that just scream to everyone around that this is a <class name>. Rage is a great example. If someone rages, they're a barbarian. The name and mechanics are thematic, they play into the class's vision, and they're fairly unique (even if they could be more so). Same with a paladin's auras. Or (in 2014 pre-hexblade), their smites. Sure, mechanically, they're just "trade spell slot for damage." But as a package, it's something no one else really did.

I call this having a Unique Cool Thing (UCT). Each class should have some kind of central mechanical ability that is unique to them, at least as a whole package, and is cool. Boring stuff just doesn't fit here. This is a place where rangers really suffer--their supposedly-unique things (in 2014) were Favored Enemies and Favored Terrain. But...that came down to a passive, mostly forgettable feature that at best simply turns off one aspect of the game. Yay. Unique-ish...but supremely not cool. At least as presented.

Druids have a UCT--wild shape. Pity that it's stapled onto a full caster chassis and as a result ends up being a side-light at best for everyone but moon druids. Moon druids fail hard at being loosely coupled, but at least they get value out of wild shape. And the 2024 version is somehow even worse, basically making it "you cast polymorph, but only on yourself, and slightly differently." Yay.

So now you have a Vision for the class and one or two Unique Cool Things that really exemplify that vision. Now it's time for the rest of the class.

High Coherence

Once you have a vision and one or two UCTs that reinforce this vision, go all in. The rest of your abilities, your subclasses, your spell choices (if it's a spell-caster), your proficiencies--all of this should go to reinforce that theme. If you're making a Huggy Bear class, everything should be about being the huggiest darn bear that ever hugged or beared. You shouldn't simply wear the label, you should be actively huggy-bearing your way through the game. As I heard it said, the 4e racial abilities did a good job of this--by giving active abilities tied to your race, it let you not just simply be a guy wearing a label that said "dwarf", but let you actively dwarf your way through problems.

Again, the 2014 paladin is a great example of highly-coherent class design. Everything either supports the Holy part or the Warrior part. The Oaths have tenets, and those tenets matter. You have Lay on Hands, Divine Sense, Divine Smite, and your Auras. Even the spell list goes all-in on the Holy Warrior angle. And since damage is provided by Divine Smite, the spell list doesn't need to have a bunch of redundant direct damage spells. Etc.

The wizard and the fighter, on the other hand, are the exact opposite of high coherence. Basically nothing in either class really plays into their themes, except by accident and as ribbons. The one fighter subclass that actually tried (the Banneret/Purple Dragon Knight) actually made using the core ability in question (Second Wind) counter-productive for its main job, since you needed to save it to actually use your subclass abilities. Wizards...well...don't really have a theme to play from, so their only stuff is "I cast more spells." And that's the opposite of coherent.

In general, "build-a-bear" (ie pick N from list of M) features tend to go against high coherence. Warlocks are like this--they're a grab bag of bits and pieces you can assemble into a coherent character, if you're willing to ignore all the rough edges and mis-fit pieces. Pick-N things can work, but in my strong opinion are best left to secondary or parallel progressions, where they're not part of the core class at all.

Low Coupling

I should preface this by making it clear that not all coupling is bad. A lot of coupling, such as to the core mechanics of the game system, is absolutely necessary. But coupling should be considered carefully and done intentionally, knowing what you're signing up for.

There are two forms of negative coupling I see frequently--coupling to outside fiction for your Vision/power set and coupling to optional elements for mechanical viability. The first is routinely seen as the shorthand in a vision statement like "it's just like <fictional character>, but as a class". You're setting yourself up for disappointment when that other character's core design, who isn't adapted for life in a cooperative table top roleplay and is often a Main Character (with all that entails), founders on the shoals of D&D reality. D&D isn't a generic game. It's not a fiction emulator. D&D emulates, well, D&D. Even a lot of things that originally were D&D clones (such as Final Fantasy) diverge enough that D&D doesn't really do them justice. You're also stuck trying to slice this character, who often has crap-tons of various incarnations, depictions, etc. into a single balanced class. It rarely works well, and never fits a D&D world's aesthetics comfortably. If I never hear "how can I play Captain IP Violation in D&D" again, it'll be too soon.

The second, more subtle form of pernicious coupling in D&D class design is basically the class design version of "we'll fix it in post". Your class is fundamentally mechanically inadequate, but works really well as a 2-level dip (or with a 2-level dip in something else). Your class really really needs that one specific magic item or spell or feat (or two or three) to function. Don't be afraid to just import those things. Yes, you're giving part of a feat as a class feature. GOOD. At least if it's what the class needs to function and fit its theme. Building a pirate who fights with sword and hand crossbow (*cough* pistol if firearms weren't wack *cough*)? Flat out give them the necessary feature(s) from Crossbow Expert. Building a fire mage class who does nothing but fire? Give them Elemental Adept (Fire) as a feature. Don't make the player have to realize that they need to pick that feat or suck. Similarly, don't predicate your class on something like needing medium/heavy armor without giving it to them. Would doing that make them imbalanced? Then re-think your design.

As a bad example, I'm going to harp on the wizard (again) <snip 42 page rant on why the wizard is the worst-designed class and a net drag on the game ;)>. Wizards have no real features, except in a couple of the subclasses. Instead, they're an empty chassis waiting to be filled with spell choices. Your spell choices define your wizard. Which means a few (bad) things--

• If you picked the wrong spells, you simply suck and failed at character creation. Yay.
• If you picked the broken-strong spells, you obviate a lot of challenges. Effectively, you "won the game" at character creation. Yay.
• Unless you know the spell lists inside and out and all of the quirks therein, it's a pain in the hind end to make a viable, not-broken character. And you and your DM will be spending way too much time trying to figure out how to resolves stuff.
• There's nothing pushing you to be thematic--except for very small things, a Divination wizard is only marginally better at Divination spells than any other wizard. In fact, they're encouraged not to pick up divination spells as their free spells on level up if they can get spell scrolls--if they do pick them up for free, they've wasted a subclass feature. Etc. A few sub-classes break this trend, but since they're stapled on a super-high-ceiling class, this tends to just make them badly out-of-balance.
• Worst of all, their performance is completely unpredictable to the DM. It can swing on a long-rest-to-long-rest basis from breaking the challenges over one knee effortlessly to being almost useless.

All in all, the wizard is basically "gigantic spell list, the class". And that doesn't make for good design (even if it does make for really strong characters).

Wrapping Up

So by the end of this process, you've written a class. One with a strong vision, one or two Unique Cool Things, mechanics and thematics that tightly cohere to each other and to themselves while being loosely coupled both to the outside fiction and to other optional elements.

Note that you can have good class design and suck at actually making good numerical implementations. This was 2014's cardinal sin for a lot of classes/features/etc--decent ideas, crappy attention to detail. I can appreciate good design while only knocking off small numbers of points for rough edges. Too many rough edges mechanically and the design needs more work.

Examples of what I consider good design by this standard (from 2014 5e):

• Paladin, hands down. Every aspect is firing on all cylinders.
• Cleric is decent at this. Too spell-heavy for my taste, but not as bad as others, and their spell lists are thematically cohesive. Decent vision in 2014 (partially ruined IMO by 2024's move of the subclass to level 3, which means you're just generically "divine" until then), a decent (for the most part) UCT. Could be better, but still not so bad.
• Monks are pretty darn good. Implementation wasn't as good as the paladin, but the overall design works.

Decent, but flawed examples:

• Barbarians. They have the vision and UCTs down. But in trying to make them cohesive, they ended up making them mostly one-note with very little width. Rage and smash, that's all.
• Sorcerers. Thematically great (for the most part), but the execution wasn't all there. And their UCT (metamagic) was too half-hearted to truly be Cool. Again, too much dependence/coupling to good spell choices.
• Rogues. A good steady workhorse. But again, too timid of execution.
• Bards. A good solid vision with a few UCTs...and then there's Magical Secrets and the decision to make them full-casters. Which means that they're tightly coupled to every single other major spell-caster out there. And end up struggling to not just be a bag of spells.
• Druids. They've got a good vision and UCT...and then totally forgot about it for most of the classes, because it's way too specific and extending it makes it broken mechanically. Very very low coherence. Most of the subclasses (and even moon druids in late T2 and T3) are just nature-flavored wizards. Again, being a full-caster eats up so much of the design budget that it means you can't actually explore what it means to be a shape-shifter or guardian of nature.

Bad examples:

• Fighters. Compared to 3e, at least they have something like a UCT in Action Surge (and Second Wind). But taking another action is pretty bland when all you can do with it is make a bunch of attacks. And nothing really buys in and builds on these things. Worst of all, there's very little vision here. "He's the guy who fights". Yay. Everyone in D&D fights. That tells me nothing.
• Rangers. Poor rangers. They have no idea who they really wanted to be, and it shows 100%. Absolute lack of vision, coupled with very tight coupling to a very specific external character...that shares only a name. Oof. Big oof.
• Wizards. I've ranted enough about them.

• Artificers. If they're not "insert Marvel Superhero Here", they're "pick options from this big meta-list" (lists that contain lists inside of them). All without really doing justice to the vision they put forth.