Design Notes: Spells and Niches (Patreon)

Today I want to talk in little more depth about spells and how they fit. And I want to focus on two concepts that have come up a few times as I work on the necromancer: spell overlap and DM-only spells. This is more of a pure design-discussion post; there aren't absolute right or wrong answers on this one, though obviously I've got opinions.

Spell Overlap

Overlap is when you have multiple spells that basically all serve the same purpose, whether mechanical or thematic (or both). Consider the classic fireball: 3rd level, 8d6 damage, 20' radius. It's the most iconic and basic area nuke. So the question is: how jealously should we guard fireball's niche? How okay is it to have other spells that do similar things?

Scenario A: No Niche Protection. For the sake of argument, let's pretend the game includes coldball, acidball, poisonball, and lightningball — identical spells with different damage types. This would serve a couple of useful purposes: it would let players differentiate their spellcasters, it would make it easier to have a preferred damage type, and it would spice up builds a tiny bit. But obviously it would also do a pretty poor job on all those things. Choosing between identical spells with the same damage type isn't very interesting, and the choices don't really feel different because, of course, they aren't (meaningfully). This would obviously be too much overlap.

Scenario B: Absolute Niche Protection. So take the opposite hypothetical: let's guard fireball's niche super tightly. Let's pretend that it's the only 3rd-level area nuke, and any spellcaster who wants to blast groups of enemies has to take it. (This is actually somewhat true in practice: because fireball is intentionally overpowered, it does elbow out the other options for 3rd- and 4th-level area nukes.) The impact of this is that any spellcaster who doesn't get fireball or intentionally chooses to pass it up would feel very distinct: fireball in this case isn't just occupying thematic space, but it's completely filling up a mechanical space as well. Distinctive and interesting, but super limiting. Your acid-specialist sorcerer will wind up having to throw fireballs to be effective, and characters without fireball just wouldn't be able to handle the area-nuke function.

Those are both extreme examples, to illustrate the endpoints of the idea. But let's ditch the hypotheticals and talk in actuals, for a bit. Here's fireball and a few comparable spells.

Fireball: 3rd level, 28 (8d6) fire damage, 20' radius. Open to sorcerers and wizards only.

Shatter: 2nd level, 13.5 (3d8) thunder damage, 10' sphere.

Ice Storm: 4th level, 23 (2d8+4d6) bludgeoning/cold damage, 20' radius, creates 1 round of difficult terrain.

From Xanathar's:

Erupting Earth: 3rd level, 19.5 (3d12) bludgeoning damage, 20' cube, creates lasting difficult terrain. From Xanathar's.

Snilloc's Snowball Swarm: 2nd level, 10.5 (3d6) cold damage, 5' sphere. From Xanathar's.

Vitriolic Sphere: 4th level, 25 (10d4) acid damage, 20' radius. Deals another 12.5 (5d4) damage after a turn. From Xanathar's.

Some of these spells have distinctive uses and niches: shatter in particular gets some use purely by being available a level earlier. Snilloc's, by comparison, sees virtually no use, since it's almost strictly worse than shatter in every way. Erupting earth and ice storm see a little use, mostly by virtue of being available to druids, but not much otherwise — they're also significantly worse than fireball. Vitriolic sphere makes maybe the best case for its existence, with higher damage but a significant downside (since 1/3 of the damage is delayed and a successful save results in 1/3 damage rather than the typical 1/2). But we're honestly a lot closer to Scenario B than Scenario A. Trying to build an AoE blaster without taking fireball is absolutely an uphill climb. What if we pull in some more spell sources, though?

From KibblesTasty:

Cold Snap: 2nd level, 13.5 (3d8) cold damage, 5' radius. Applies a slow and creates 1 round of difficult terrain.

Earth Ripple: 2nd level, 18 (4d8) piercing damage, 10' radius. Can sacrifice some damage to apply control effects instead.

Geyser: 4th level, 35 (4d6+6d6) bludgeoning damage, 10' radius. Since 60% of the damage comes from a knock-upwards and subsequent fall, doesn't really function in areas with ceilings.

Illusionary Fireball: 3rd level, 28 (6d6) psychic damage, 20' radius. No damage on a save, but doesn't affect allies.

Pillar of Fire: 4th level, 35 (10d6) fire damage, 5' radius.

From Spells That Don't Suck:

Burst of Flame: 2nd level, 10.5 (3d6) fire damage, 10' radius. Can sacrifice all damage for control effects, or situationally get a extra 3.5 (1d6) damage.

Corrosive Burst: 4th level, 20 (8d4) acid damage, 20' radius. Deals 10 (4d4) damage every round until an action is spent to clear it.

Fireblast: 3rd level, 24.5 (7d6) fire damage, 20' radius.

Hailstorm: 4th level, 25 (10d4) bludgeoning/cold damage, 20' radius. Upcasts at 2d4 and gets extra radius.

Riptide: 3rd level, 18 (4d8) bludgeoning damage, 20' square. Knocks prone and shoves.

From So Many Spells:

Coruscating Flash: 2nd level, 7 (2d6) radiant damage, 15' radius. Blinds for 1 round.

Crystal Blossom: 4th level, 18 (4d8) cold, 10' radius. 36 (8d8) in 30' after a round, unless creatures destroy the blossom or get away.

That's a lot of spells to go through all at once, but there's a general trend there: more interesting options for mid-level AoE nukes. (I opted to leave out damage-over-time and cone spells, which would both sweep in a ton more). Fireblast deserves a special mention as a straight nerf to fireball; it's still above the damage curve, but less, which actually opens up quite a bit of space for other spells to play around it. But almost all of these spells occupy slightly different niches than fireball, usually getting extra damage in certain situations or sacrificing damage to get interesting riders. They're still all broadly functional as AoE nukes, though. Nothing here really looks like Snillloc's, erupting earth, or ice storm — strictly inferior spells that see use only when you're forced into them.

Going back up to our broader design point: we've definitely shifted our spell-design philosophy here, if unintentionally. We've departed quite a bit from the Scenario B "Absolute Niche Protection" approach. I personally think it's a happy medium here, where fireball stays on top in its specific niche ("how to do the most area damage, full stop") but is no longer the unrivaled winner in the entire area-damage theater.

DM-Only Spells

This one's more directly relevant to the necromancer work. D&D's always had the concept of "DM-only spells" — spells that players literally aren't meant to use. It's not an inherently awful concept, but it's got some issues for sure. You might look at spells like geas, finger of death, or skywrite.

In general, a DM-only spell is a spell a player would never use. They might be too evil, too underpowered, or too niche. (In fact, WotC often intentionally aims for two of those at once: gratuitously evil spells are usually underpowered, to aggressively steer players away from them.) For the examples above, geas is so wildly niche a PC will almost never have a use for it. Skywrite isn't pointless, exactly, but you'd never pick it over another option, not in a million years. Finger of death is usable, but very evil (immediately block resurrection and create a zombie) — and WotC steers players away from it hard by making it strictly worse than disintegrate.

To say some nice things first: DM-only spells are effectively a worldbuilding tool. Even if they're effects you don't intend players to have access to, writing them through the same magic system helps the world feel more coherent. That evil wizard isn't just some bespoke obstacle; he's a character, with motivations and feelings and with powers constructed using the same rules and spell corpus you use. (Seems like he's kinda bad at spell selection, though.)

The alternative here is part of what I call "videogamey" design, where enemies are constructed explicitly around their narrative needs and freely use mechanics players can't relate to. It reinforces the sense that you're playing a game, that the entire world is just a stage for the PCs to stride around on. This isn't explicitly wrong; there are people who really enjoy this style of game, especially folks whose gaming habits came of age in the MMO era. But for me, I think videogamey design sacrifices the one thing TTRPGs legitimately do better than any other medium: provide a reactive and breathing simulation to the players' actions. Let's say the party tangles with our evil wizard above and kicks his ass, but he gets away to terrorize them again. If he's modeled entirely with bespoke features, then either he shows up again exactly the same, or the DM rewrites his stat block and he's now different in unpredictable ways. But if he's using the same spellcasting system and spell corpus, there's a much cleaner mechanical answer: maybe he's leveled up a bit, swapped out some of his spell selections, and retooled to better fit the fight. (Obviously he doesn't mechanically level up, though you can do that if you're willing to give an NPC class levels; some people do, but it's pretty clunky and not recommended.) It reinforces the sense that he's playing in the same world as the PCs and operating under the same rules.

Now to the bad about DM-only spells: they're kind of awful. Oftentimes, they function as trap options, choices that a player can theoretically choose but make their character strictly worse. A player might pick one because it sounds cool, and only later do they realize it underperforms or is kind of useless. (We've got some trap options up above, too; woe to the player who takes Snilloc's because it sounds funny and then realizes it does crap damage.) Finger of death is a classic trap option; it's totally metal, but unfortunately also a pretty bad choice. And while DM-only spells help the rules feel consistent, they also force their own suspension of disbelief: you might say "Really? An archmage wasted his time learning magic mouth?" And when an evil spell is also mediocre, it undercuts the entire fiction about why characters resort to evil — the dark side is meant to be a shortcut to power, not a shortcut to being weaker than a good-aligned wizard would be.

The Necromancer

Where these intersect with the necromancer is that the class requires a lot of new spells, and inevitably, they run into issues related to spell overlap and DM-only spells.

For overlap: when you write a lot of spells, you're obviously going to hit overlap issues. Sometimes it's mild; maybe two spells overlap, but the necromancer doesn't deserve access to the original and the original's classes don't deserve access to the necromancer's. Sometimes it's major; I gave the necromancer undead familiar, which is 90% identical to find familiar but has just enough twists on it that it needs to be its own spell. Shadow phase is explicitly a shadow-magic answer to invisibility, covering a bunch of the same narrative ground using different rules. If I wanted to stick to WotC's Philosophy B, though, "Absolute Niche Protection", I don't think I could write a lot of these, and the necromancer would really be hurting for spells.

For DM-only spells: Historically, WotC has implemented a lot of necromantic functions via DM-only spells. As with finger of death, they intentionally write evil options as trap options, to try and nudge PCs away from being evil. That's not a concern for the necromancer: if you're allowing a player necromancer, you're automatically on board with darker-and-edgier PCs. But it does mean I have to do some rewriting, replacing WotC's DM-only spells with alternatives suitable for PCs. Of course, Spells That Don't Suck handles some of that for me already, but there's still a bit more to do.

Edit: caveat that all the above is couched around PHB and XGtE, when WotC had more design intent going on. As you get into later books and especially 2024E, they seem to have increasingly poor knowledge of their own design ideas.