Why "Sea Witch = Sorcerer" Is the Wrong Answer

When I started designing playable witch classes for D&D 5e, the first thing I did was look at what already existed. Homebrew wikis, Reddit threads, old Unearthed Arcana content. I wanted to see where the bar was set.

The bar was on the floor.

The standard approach goes like this: find a witch, match it to the closest existing class, slap on a few flavour features, done. Sea Witch? Sorcerer, obviously. Green Witch? Warlock, because of the illusions. Night Witch? Maybe Warlock again, maybe a spooky Wizard. Iron Witch? Barbarian, she's big and hits things.

I understand why people do this. It's fast. The math is already solved. And if you squint hard enough, you can argue that a Sea Witch's manipulation of people isn't so different from a Sorcerer's charm spells.

But here's the problem. A Sea Witch isn't a Sorcerer. She isn't a class of person. She's a philosophy. She is the logic of the deep ocean made into a creature that can look you in the eye and watch you drown on dry land. Giving her a Sorcerer's spell list and calling it done is like giving a wolf a dog tag and saying that counts as character design.

What actually makes a witch a witch

Before you can design a playable witch class for D&D 5e, you have to actually understand what a witch is. Not what she can do. What she is.

Witches are the original dark mirrors. Every type of witch is a corruption of something people find comforting or powerful. The Green Witch lives in the forest that's supposed to be magical and good. The Night Witch invades your sleep, the one place you're supposed to be safe. The Sea Witch is the ocean, which people already fear, wearing the face of the one thing more frightening: a person who knows you're afraid.

That's the design key. A witch class should feel like playing a corrupted archetype. Not a wizard who's ugly. Not a sorcerer with a bad attitude. Something that specifically inverts something the player finds familiar.

Once I understood that, the approach changed completely.

The problem with mapping to existing classes

The mechanical problem with "Sea Witch = Sorcerer" is that Sorcerer as a class is about innate magical power radiating outward. You have too much magic inside you and it comes out. That's not what a Sea Witch does.

A Sea Witch's power is about pressure. The crushing weight of deep water. The pull of a current you can't fight. The understanding that the ocean does not hate you; it just doesn't care whether you live. Her magic should feel like that. Not like fire erupting from your hands. Like something inevitable, dragging you down.

Here's what the comparison looks like in practice:

Approach	Sea Witch mechanics	What it actually feels like
Lazy mapping	Sorcerer, Draconic bloodline (water), reskinned fire spells as cold damage	A cold damage wizard. Nothing specific to witches. Nothing specific to the sea.
Roll Evil approach	Custom class. Spell list built around force and cold, themed around tides and drowning. Signature mechanic: Undertow, a persistent drag that relocates targets without their consent. Hideous Visage as a once-per-rest fear tool that scales with coven size.	Playing as something the ocean would send. Enemies are pulled, repositioned, slowed. The damage feels like pressure, not explosion.

The difference isn't just flavour text. When someone plays the Sea Witch in our campaign, they lean across the table and say I want to drag the paladin into the flooded room. They're thinking like the class, not just using the class.

That's what good design does. It shapes how you think at the table.

How we built each witch from the ground up

The design process for each witch started with a single question: what does this witch's power feel like to be on the receiving end of?

Not "what spells does she use." What is it like to be the person in a room with her, and how do you translate that into mechanics a player can control?

For each of the six witches in Daughters of the Black Forest, the answer was different:

• Cinder Witch: She doesn't burn you. She leaves ash. The damage lingers. The environment changes. Her coven spell, Burning Ash, creates a blind-and-burn cloud that scales from nuisance to room-wiper by level 8. Playing her feels like playing a slow, inevitable fire.
• Green Witch: Her power is that you never quite know what's real. Illusion and mimicry as the core of the class, but built around social control rather than battlefield control. She wins encounters before combat starts.
• Night Witch: Soul-based mechanics. She accumulates leverage over enemies through their dreams, over multiple sessions. She's the only class in the book designed to reward patience across a campaign rather than just within a single fight.
• Sea Witch: Force and cold damage, movement control, Undertow. The Hideous Visage scales with coven members in range. She's strongest when she's not alone, which feeds into the coven mechanics at the campaign level.
• Iron Witch: Pure brutality, but not the Barbarian kind. She doesn't rage. She calculates. Her iron grip grapple chain is the most mechanically complex part of the book: you grab, you squeeze, you break, you use the body.
• Snow Witch: Patience as a mechanic. Her power accumulates over rounds. She's slow to ramp up, nearly unkillable once she does. Playing her feels like a chess player who knows she's already won.

The coven magic system

Individual class design was only half the problem. The other half was making the witch classes feel like a group in a way no other D&D 5e party composition does.

In official D&D, hags get coven magic when three of them are together. It's a passive bonus. You show up with the right group and you get extra spells. That's fine for an NPC, but it's not interesting for players.

The Roll Evil coven magic system works differently. Each witch initiates coven spells in a different way, and the effect changes based on which witches are present. A Cinder Witch's coven spell is Burning Ash, but if a Sea Witch is also in range when she casts it, the ash becomes wet, the damage type changes, and the area gets a difficult terrain component from condensation. It's a small thing but it makes the coven feel like a coven, not just three strangers with a buff.

On top of that, the system scales by level. At level 4, the coven spell gets an extended effect. At level 8, it gets a second mechanic triggered by hitting the initial cloud with elemental damage. You're rewarded for learning your coven's synergies.

In 20 sessions of the Daughters of the Black Forest campaign, the moment players figured out their coven combinations was always the same: they stopped thinking individually and started thinking as a unit. That's the goal. That's what makes the witch campaign different from a regular D&D villain arc.

What this means for playable monster classes in general

The lesson from designing these six classes isn't specific to witches. It applies to any playable monster class for D&D 5e.

The question is never "which existing class is closest to this monster?" The question is "what does it feel like to be this monster, and what mechanics make a player feel that?"

That's a much harder design problem. It takes longer. It requires more playtesting. You can't just reassign a Draconic bloodline and move on.

But when it works, players stop asking "what does my class let me do in this situation" and start asking "what would my witch actually do here." They think in character before they think in mechanics. That's the gold standard for tabletop RPG class design, and it doesn't happen by accident.

Volume 1 covers the six witch classes in full. Volume 2 is where things get strange. If designing a Sea Witch class requires you to think about what pressure feels like from the inside, imagine what it takes to design a playable Gelatinous Cube.

I'm looking forward to that problem.

---