Spell Research: Knock and Wizard Lock

The knock spell first appears in original D&D's Book I: Men & Magic. It is a 2nd-level magic-user spell. The description is as follows:

A few things about this are immediately interesting to me. First, the spell opens secret doors, obviating the need to puzzle out how the secret door is intended to be opened. It also means that secret doors which are designed to open from only one side can be opened from the unintended side using this spell. I had always intuited that the purpose of this spell was to unlock doors, not open them, so this is a bit surprising to me on a close read.

I am also surprised to see that knock is a ranged spell. I had always assumed that the magic-user had to physically knock on the door for the magic to work. Since the spell doesn't simply unlock but instead opens such doors, that means doors can be opened at range, allowing any traps triggered by opening the door to be triggered while the magic-user is at a distance.

Interestingly, if we read the spell literally, it does not open doors that are not secret, held, locked by magic, barred, or otherwise secured, so the PCs will still have to risk opening those doors themselves.

No discussion of knock is complete without looking at its counterpart, the other 2nd-level magic-user spell, wizard lock:

Wizard lock has another counterpart alluded to here: hold portal. This post will not cover hold portal because it is not consistently represented in all editions of D&D. It is very similar to wizard lock except that its duration is shorter (think of a magic-user magically warding a dungeon door against an ogre attempting to batter it down, versus a magic-user permanently securing an important door in their inner sanctum).

Knock suppresses wizard lock only temporarily, and does not remove it. The magic-user must cast knock each time they wish to pass through the wizard lock. This makes me want to make the occasional wizard lock door trapped such that it automatically swings shut after a certain number of individuals pass through it, separating the party unless the wizard can open it again.

Interestingly, the magic-user seemingly cannot open their own wizard lock without a knock spell. That seems horribly inconvenient!

Since this is our introduction to the spell, I feel this is a good time to link to Designing Obstacles for OSR Play, which contains some interesting discussion about "Hard Locks" and "Soft Locks" and about knock and the problems posed by "skeleton key" spells like it and dispel magic. The post criticizes the use of wizard lock in published adventures and the tendency of adventure authors to take the ubiquity of knock for granted, rather than including more interesting ways of bypassing wizard lock and related obstacles. I don't see this as being an inherent problem with either spell so much as a flaw in the design culture surrounding these particular obstacles. 

As the post explains, there are much more interesting ways to integrate locked doors and their keys into adventures. The existence and implementation of knock does not offend my sensibilities. If I've designed a truly interesting magically locked door with an ingenious puzzle that needs solving to open it, but the player characters happen to have knock and can bypass it with a single spell, more power to them. It is less interesting, but I'm not always prioritizing what's interesting above all other considerations, such as rewarding players for being well-prepared.

I'm also not entirely sure whether wizard lock would be considered a Hard or Soft Lock. Is it a Hard Lock because it identifies specific means of bypassing it (the implication being that there are no other means of doing so)? Or is it a Soft Lock because there are multiple ways of bypassing it (knock, dispel magic, or the service of a higher-level magic-user)? This issue will only become muddier as we dig into different iterations of the spell.

Another common criticism of knock in particular is that it allows the magic-user to be a better thief than the thief. I will point out that at the time of original D&D's publication, the thief did not exist. The magic-user is not yet stepping on those particular toes. That is about to change.

Knock appears again in AD&D:

This being AD&D, the spell has been complicated quite a bit! It is classified as an alteration spell for the first time. Alteration (later called transmutation) is the school of magic concerned with the alteration of a thing's properties. In this case, the locked object is being magically transmuted into an unlocked object. Sure. That works I guess.

Knock now has an area of effect of 10 square feet/level of the caster. What exactly this means isn't at all made clear. Perhaps it will be clarified in the future, in some sort of updated version of the AD&D rules?

AD&D's version of knock also opens stuck doors (the original D&D version of the spell isn't explicit about this - my interpretation is that it doesn't affect stuck doors, unless a stuck door qualifies as "otherwise secured") and locked or trick-opening boxes or chests (in hindsight, it's interesting that the original spell only works on doors), and loosens shackles or chains. The description also specifies that knock does not open portcullises (the original version's "otherwise secured gates" language may or may not include portcullises). Lastly, each casting of the spell removes only two impediments to opening the door (i.e., a lock and a bar, but not a lock, a bar, and a spell like hold portal or wizard lock).

The thief is now a class. They are pretty bad at the things they're supposed to be good at. Knock, among other spells, allows the magic-user to automatically succeed at the things the thief is supposed to be good at, albeit a limited number of times per day. Similarly, the magic-user also automatically succeeds - again, just a few times a day - at opening stuck doors, which is one of the things the fighter is supposed to be good at. At least the magic-user can't bend bars/lift gates!

I personally am not really bothered by the toe-stepping. The magic-user only has so many spells per day, and if they want to cast knock multiple times they have to use multiple spell slots to prepare it. The magic-user is sacrificing a significant amount of utility just to open the occasional door. There is no on-demand magic (i.e., cantrips) in AD&D like would come in later editions, so the magic-user is making a choice to specialize their loadout for the day and then be limited to throwing their darts or whatever.

I think it's more likely that the party explores the areas of the dungeon that can easily be explored, the fighter and thief try their luck at any stuck or locked doors they come upon, take note of those they can't open, and the magic-user then prepares accordingly for the next delve into the dungeon to open any doors that continue to confound the rest of the party. 

That too comes with tradeoffs. The party may have random encounters as they retreat from the dungeon. If they make camp outside the dungeon - which they would be wise to do - they may have to fend off a dangerous wilderness encounter. While they're gone, the rooms they explored may be reoccupied. They may have more random encounters while traveling back to where the offending doors are. The party can make short excursions into the dungeon to achieve a straightforward goal such as this, but it's simply not an efficient way to explore.

In a world where any magic-user of 3rd-level (which is considered low-level in all versions of the game) can magically open almost any door, you also have to imagine that most important things are going to be behind several secured doors. Or, maybe, just one really secure one.

Each casting of knock only foils two "locks" on a given door. Imagine a huge vault door with multiple complex locking mechanisms and magic wards. The magic-user can get through two at a time. A 3rd-level magic-user can get through two locks per day. At 4th- through 6th-level, they can get through four locks per day. At 7th- through 9th-level they can get through six locks per day. And on and on. Suddenly the thief doesn't seem so superfluous (but you should probably make them more competent somehow - there are far more blog posts on the subject than I care to wrangle right now).

I also like how the spell suppresses a wizard lock for 1 turn specifically. In my world, doors with wizard locks are on a timer - after the turn is up, the door swings closed! I really like the idea of the party breaking into a magic-user's vault with their one use of knock, only to neglect to spike open the door and end up trapped inside.

The AD&D version of wizard lock similarly adds a number of specifications:

Like knock, wizard lock is an alteration spell, has the same unexplained area of effect, and specifies the types of objects it affects. The wizard lock can now be bypassed by breaking the affected object (the Lock is becoming Softer). However, magic-users now need to be one additional level higher than the original caster (i.e., 4 or more levels higher compared to the original spell's 3 or more levels higher).

The bit at the end is referring to the fact that hold portal can be shattered at will by any extra-dimensional creature (demons, devils, elementals, etc.). In other words, wizard lock is a permanent version of hold portal that also effectively keeps out such creatures.

AD&D 2e's version of knock cleans up the 1e version:

First, I love that the spell is now reversible. OD&D and 1e both had reversible spells, but knock wasn't one of them. We don't need a knock spell and a lock spell. They're the same spell. You just flip 'em! 

...Is what I would say, if it weren't for the fact that 2e still has a wizard lock spell! We'll get to that in a moment.

This version of knock makes the odd specification that the spell does not affect "ropes, vines, and the like." Okay. Why would it? It also specifies that the location of a secret door must be known in order for knock to open it. This sounds like common sense to me, but in hindsight one could read the original D&D and 1e descriptions as allowing the spell to open undiscovered secret doors (the 1e version in particular "causes secret doors to open", which is vague wording).

The 2e version also elaborates on the area of effect - delightfully, in my opinion. The intention is, apparently, for more powerful wizards to be able to open larger doors! A 20th-level wizard can knock a 200-square-foot door. How big are a giant's doors? If a standard human-sized door is 4' x 7', a 26-foot-tall storm giant (assuming similar proportions) would use a door that is roughly 15' x 26' (though probably even bigger than this). That would require a 39th-level wizard to open! Does the spell assume the caster is human-sized? Can giant wizards easily knock open their own doors? Much to think about!

2e's version of wizard lock is much the same as 1e's, albeit with an important quality of life change:

The caster can now open their own wizard lock! This feels like it should have always been the case. It also introduces an interesting possibility: If the wizard who originally cast wizard lock is at hand, the party could take the wizard hostage and force them to open all of their locks! I could imagine this happening if the party is raiding the wizard's stronghold and is able to neutralize the wizard, but still needs to gain entry into the vault. The Lock gets Softer yet again!

The description does not clarify the area of effect, but we can intuit that it is similar to knock. Interestingly, a 20th-level wizard can knock a 200-square-foot door, but wizard lock a 600-square-foot door. Such a door would require a 60th-level wizard to knock it open!

So what is the point of the reverse of knock, lock? Well, it only affects doors that have a physical locking mechanism. Because of knock's smaller area of effect, is also affects smaller doors than wizard lock. I have no idea why one would prepare lock rather than wizard lock. Making knock a reversible spell was a good idea, but it ultimately feels like a missed opportunity.

Surprisingly, the 3e version of knock is largely unchanged:

This version specifies that the spell only loosens welds, shackles, or chains which "serve to hold closures shut." I think what it means is that it loosens, for example, chains that block a door or bind a chest, but not shackles restraining an individual. There's a very minor change to its range (most likely accounted for by D&D switching from yards to feet as the standard measurement at the time). Very powerful wizards (and sorcerers) can now unlock very large doors from very far away!

Annoyingly, the description for arcane lock is split onto two pages:

First, the name change. Locking doors with magic isn't just for wizards anymore! Second, it's an abjuration now. This makes sense, as abjuration is the school of warding, protection, and negation. I like it better than magically altering the properties of a door from unlocked to locked. The spell also now requires a material component in the form of gold dust worth 25 gp. 

I like how the spell is integrated into 3e's skill system by specifying that it adds 10 to the normal DC to break open the door or portal. In AD&D, a mundane lock and a wizard lock are equally difficult to force. In 3e, each type of door has its own DC, mundane locks add to the DC, and arcane lock adds onto that, making it possible to create a door that is particularly hard to muscle open.

Importantly, this version no longer allows a higher-level caster to open the arcane lock at will. The Lock has become Harder. 

4e, unsurprisingly, changes knock more than any edition yet:

I don't care for the glowing blue key that floats into the door causing it to glow amber. It's weirdly specific and feels like a video game (a comparison no one has ever made with 4e). Otherwise this is interesting. 

In past editions, knock was a 2nd-level spell, which could be cast by a character of 3rd-level or higher. In 4e, knock is a 4th-level ritual, which means the character must be at least 4th-level to cast it (i.e., it is not comparable in power to a 4th-level spell in other editions). Knock is unlocked (haha) for player characters slightly later in 4e than in other editions.

Since knock is now a ritual, it's not tied to spell slots or to 4e's system of at will/encounter/daily powers. If the party has a knock scroll, they can cast the ritual once. If they have the ritual in a book, they can cast it as much as they want, if they have the necessary components.

This is the first time knock has a component cost. I think this is just because all rituals in 4e have a component cost (have to limit them somehow I suppose - 5e would change this, IMO for the worse). It also makes sense considering that 3e introduced a component cost for arcane lock.

I like the idea of the spell requiring a key carved from bone and etched with arcane symbols (a literal skeleton key) which crumbles into dust after the spell is cast. I suppose the ritual requires a healing surge because it's draining for the caster or their assistants (either can provide the necessary healing surge). I really like 4e's use of healing surges as meta currency and wish 5e did more with its hit dice.

I also like how opening a lock is now a skill check, with multiple locks or closures requiring multiple checks. If the rogue has to do it, the ritual caster should too! The description points to the Thievery skill for its DCs, which are 20 (Heroic tier), 30 (Paragon tier), and 35 (Epic tier). It doesn't seem fair that the ritual caster then gets a bonus to the check, presumably preserving their better-than-the-rogue abilities, but maybe there's something about the math that I'm not accounting for due to my unfamiliarity with 4e.

Knock no longer automatically opens the door, chest, gate, etc.! You still have to physically open the thing. It also destroys an arcane lock, which is too bad. No more automatic door traps.

It's important not to miss what the description is omitting! The range isn't specified (I assume it's roughly equivalent to every other version of the spell). Knock doesn't affect secret or stuck doors or chains anymore. The chains thing always seemed weird to me. The secret doors change is a good one. Players should have to figure out how to open those! Unfortunately, the days of more powerful spellcasters being able to open bigger doors are also at an end.

4e's version of arcane lock is similarly interesting:

Now, the Arcana check of the caster sets the DC to open the lock. This is an elegant solution to the "a wizard X or more levels higher than the the one casting the spell" mechanic. Now, the arcane lock caster sets the DC, and the creature trying to beat the lock has to beat the DC. Simple.

Let's not miss perhaps the most important change: You can now open an arcane lock with Thievery! Rogues and the like are not rendered useless by an arcane lock. The way that I read it is that arcane lock can magically reinforce an existing physical locking mechanism on the door, making it more challenging for a Thievery-using character to open. If the door has no physical locking mechanism and arcane lock is instead simply magically holding the door shut, I don't think Thievery would apply.

In addition, the caster can now designate specific individuals or categories of creatures who can open the door at will. The caster can also set a password, which introduces yet another way to bypass the lock.

In addition to now being destroyed by knock, arcane lock is now also destroyed by a successful Thievery check. The caster also knows instantly when the arcane lock is defeated, meaning it doubles as an alarm spell. That makes it a lot more difficult to break into the wizard's vault.

5e rolls back many of 4e's changes and adds a new spin to the spell:

No longer is knock a ritual. Its range is shorter than ever (unless I'm missing something about the 4e version) - 5e took the 60 yard range from pre-Wizards of the Coast D&D but kept feet as the standard unit of measure from 3e, resulting in a range of 60 feet. The spell once again does not require costly components. It affects any "object that contains a mundane or magical means that prevents access". It affects stuck doors again. It only unlocks one lock at a time. It once more only temporarily suppresses an arcane lock.

The author of the spell saved my favorite change for last. A real juicy sucker punch of a sentence: "When you cast the spell, a loud knock, audible from as far away as 300 feet, emanates from the target object." 

Aside from more powerful spellcasters being able to open bigger doors (no I won't let that one go), this is my favorite change yet. This is the best disincentive to using knock. How badly does the party need to open this thing? Are they willing to alert potentially all of the dungeon's inhabitants? What if those inhabitants come looking for them? They'll have to loot the vault quickly and then make a run for it. Can they use the spell as a diversion? Can they use it to set a trap? Should they cast a silence spell first?

It's fantastic.

The 2024 version of the spell is identical, albeit written more tersely and with Game Terms capitalized according to that style. 

5e's arcane lock preserves some of 4e's changes and blends in a bit of 3e:

You can still designate creatures that can bypass the lock, and you can still set a password. Knock only temporarily suppresses the lock. Instead of setting the DC with a spellcasting check like in 4e, you simply add 10 to the existing DC like in 3e. 5e still allows for the possibility of picking a magically augmented lock. 

The 2024 version doesn't seem to change much:

Unfortunately, you can no longer pick a lock affected by arcane lock. It reads as if you can still destroy the door or force it open, but no guidelines are given for modifying the DC. As seems to often be the case with the 2024 rules, this version offers only (often detrimental) simplification without clarity or innovation.

What's your favorite version of knock? How about wizard/arcane lock? The original versions are short and simple, AD&D's are more complex, 3e doesn't change much, 4e totally reworks both spells, and 5e brings them back to their roots with at least one brand new twist. 

Personally, I like some combination of AD&D's weird specificity, 4e's skill contests with a cost, and 5e's clever complications. There's just one problem that no version of knock has ever solved. The caster should have to physically knock on the door! That really grinds my gears. 

Here's my own version of knock, combining some of my favorite elements from several iterations:

Knock

2nd-level transmutation (ritual)

Casting Time: 1 action

Range: Touch

Target: A stuck door or a door, box, chest, set of manacles, or other closed object secured with a lock, bar, arcane lock, or other means preventing access, with a surface area no larger than 10 square feet x your spellcaster level

Components: V, S, M (a key carved from bone and etched with silver-inlaid arcane symbols worth at least 25 gp, which the spell consumes)

Duration: Instantaneous

Classes: Bard, Sorcerer, Wizard 

 

Make a spellcasting ability check with proficiency against the lock, stuck, or barred door's DC. On a success, a target that is held shut by a mundane lock or that is stuck or barred becomes unlocked, unstuck, or unbarred. If the object has multiple locks, you may continue to make spellcasting ability checks until all locks are unlocked or until you fail.

 

If you choose a target that is held shut with arcane lock, make a spellcasting ability check against the DC set by the caster of arcane lock. On a success, that spell is suppressed for 10 minutes, during which time the target can be opened and shut normally.

 

If you know the mage hand spell, you can use your mage hand to deliver the knock. 

When you cast the spell, a loud knock, audible from as far away as 300 feet, emanates from the target object.

(Written for 5e, but you could easily use it with almost any edition by just changing or omitting certain details.)

The spell is now a ritual, which increases its utility, for better or worse. It has a component cost however, so it isn't necessarily spam-able (not that the 25 gp will be much to player characters for long). The caster has to physically knock on the object (unless they have mage hand, which I thought was a fun twist). We're also bringing back the limitation on the size of the target object, because of course we are. 

Because expert characters like rogues are likely to have expertise with thieves' tools, and the knock spell only grants proficiency on the spellcasting check, rogues are still likely to be better lockpickers than their spellcasting counterparts, but a spellcaster remains a viable substitution if a rogue isn't around. If the target is affected by an arcane lock, it triggers a skill contest, which I feel better represents the different capabilities of more or less powerful spellcasters accounted for in arcane lock's early iterations.

And here is my version of arcane lock:

Arcane Lock

2nd-level abjuration (ritual)

Casting Time: 1 action

Range: Touch

Target: A closed door, window, gate, chest, or other entryway, with a surface area no larger than 10 square feet x your spellcaster level

Components: V, S, M (gold dust worth at least 25 gp, which you use to etch arcane glyphs on the object's surface, and which the spell consumes)

Duration: Until dispelled

Classes: Artificer, Wizard 

The target becomes locked and held fast for the duration. You and the creatures you designate when you cast this spell can open the object normally. You can also set a password that, when spoken within 5 feet of the object, suppresses this spell for 1 minute. Otherwise, it is impassable until it is broken or the spell is dispelled or suppressed. 

When you cast this spell, make a spellcasting ability check with proficiency. The result of your check determines the DC required for a knock spell to suppress your arcane lock.

While affected by this spell, the object is more difficult to break or force open, and any physical locks on it are more difficult to pick. The object's AC increased by 2, and it is treated as one size larger for the purposes of determining its hit points. The DC to force the object open or pick any locks on it increases by 10. If the object's locks are successfully picked, the spell continues to hold the object shut.

As with knock, I made arcane lock a ritual and added the limit on the area of effect (couldn't help myself). I made the arcane lock tougher to physically destroy and clarified that picking the locks still requires the door to be forced. 

In my mind, the spell is both making the lock tougher and physically reinforcing the door. Picking the lock doesn't remove the spell, but it makes forcing the door easier than if you were to try forcing the door with multiple locks in place. Hopefully that makes sense.

And that's knock and wizard/arcane lock! They're two classic spells that have gone through a variety of iterations. I am surprised by both the similarities and differences from one version to another, and delighted by the little intricacies and the ways in which each edition's quirks manifest in the spells' descriptions. They are interesting and iconic spells which are often written off for causing problems, but which I personally feel are just additional magical tools - one most likely to be used by the players to solve problems, and another to be used by the DM to cause them (in a good way).

On Shapechanging

If one goes looking for what are commonly considered to be the "problem" spells in 5th edition Dungeons & Dragons, there are some usual suspects which always come up - guidance, find familiar (and summoning in general), healing word, augury (and other divinations), healing spirit, counterspell, tiny hut, simulacrum (and all illusions), wish, and so, so many more. Hyper-online D&D players love to complain about magic in D&D.

I see problems with these spells and others, and have had my share of trouble with magic in D&D, but none of them has ever bothered me quite so much as polymorph, its related spells, and shapechanging magic in general, like the druid's Wild Shape ability.

On Challenge Rating

Polymorph (Player's Handbook, page 266) allows the caster to transform a creature for up to an hour into any beast whose challenge rating is equal to or less than the target. This changes the target's game statistics (including mental ability scores, but not alignment and personality) to that of the beast. 

True polymorph (page 283) uses the same CR = level approach as polymorph, but the new creature doesn't need to be a beast, and the change lasts until dispelled if the caster concentrates on the spell for the full hour. 

Shapechange (page 274) again uses the CR = level approach, but it only affects the caster, the new form needn't be a beast, the caster retains their mental ability scores and skill and saving throw proficiencies, and for the duration the caster can use an action to change shape again.

My problem with this is that the Monster Manual states "An appropriately equipped and well-rested party of four adventurers should be able to defeat a monster that has a challenge rating equal to its level without suffering any deaths" (page 9). In other words, a CR 7 creature is meant to be a challenge for a party of four appropriately equipped and well-rested 7th-level characters.

So...if a CR 7 creature is roughly equivalent to four 7th-level characters, why can a 7th-level character transform into a CR 7 creature, which in and of itself should be a fitting challenge for the entire party? Why does the Monster Manual tell the DM, when building encounters, to equate a CR 7 creature to an entire 7th-level party, but for the sake of this spell and those like it, equate a CR 7 creature with a single 7th-level character?

Mass polymorph (Xanathar's Guide to Everything, page 160) is a bit different - it affects up to 10 creatures, and just like polymorph, the new forms can be beasts only. However, the new form's CR must be equal to or less than half the target's level. I imagine that this caveat is there because it affects more than one creature, but I'm going to interpret it as recognition that an Xth-level character is not equivalent to a CR X monster.

Wild Shape (PHB, page 64) recognizes this too. A 2nd- or 3rd-level druid can Wild Shape into a CR 1/4 beast (i.e., not a CR 2 or 3 beast). A 4th-, 5th-, 6th-, or 7th-level druid can Wild Shape into a CR 1/2 beast. At 8th-level and beyond, they can Wild Shape into a CR 1 beast. The Circle of the Moon druid (page 69) can do a bit better, but even at the height of their powers, they can Wild Shape into a CR 6 beast (at 18th-level) or a CR 10 elemental (at 10th-level, expending two uses of Wild Shape in doing so).

Why is it that a 7th-level bard, sorcerer, or wizard gets access to a spell which immediately allows them to change into a more powerful creature than is allowed by the shapechanging class ability of the shapechangey-est subclass of the game's shapechanging class? (Yes, druids do get access to polymorph too, but that's not the point I'm making.)

Similar to how mass polymorph affects multiple creatures, one could argue that the limitation on Wild Shape is there because the druid can use that ability relatively often (twice per short rest), whereas another spellcaster's spell slots (used to cast polymorph) only return on a long rest. Yet, considering a spellcaster can get up to three 4th-level spell slots per day with which to polymorph - plus more, if they choose to use higher-level slots to do so - this doesn't seem right either.

As an aside, this makes me wish that the Monster Manual or Dungeon Master's Guide included a table which roughly matched characters of a given level to monsters of a given CR, i.e., a CR 1 monster is roughly equivalent to a 3rd-level character - something like that. It would make so many things easier - not just this specific issue, but also encounter balancing (when I feel the need to do that), and designing player characters as monsters and vice versa. 

Want to stat up a 20th-level cleric as a PC and convert them to a monster/nonplayer character? Here's what CR they should be. Want to give a red dragon wyrmling sorcerer levels? Well, a CR 4 creature is roughly equivalent to a 6th-level character, so they can have 6 levels of sorcerer (granted, this would effectively change the CR, but let's be honest - those monsters need some beefing up).

Xanathar's Guide to Everything has something similar in Chapter 2: Dungeon Master's Tools, under Encounter Building (page 88), which gives guidelines for how many creatures of a given CR to use in an encounter against a number of characters of a given level, but these tables also make the laughable claim that a 20th-level character is roughly equivalent to a CR 7, 8, 9, or 10 monster, so I don't really trust that math.

On Hit Points

Perhaps the more problematic element of polymorph is this: "The target assumes the hit points of its new form. When it reverts to its normal form, the creature returns to the number of hit points it had before it transformed." All of the aforementioned shapechanging spells work this way, save for mass polymorph, which does the same thing but with temporary hit points for whatever reason (precluding the affected creatures from benefiting from additional sources of temp HP or from healing), and shapechange, which specifies that when the target changes their form additional times, they don't assume the hit points of their new form if it has more hit points than their last form.

So the sorcerer twins a polymorph spell to target the 8th-level barbarian and fighter, turning them both into tyrannosaurus rexes (D&D is a silly game, which is a good thing). They stomp into combat against the dragon, take 136 points of damage each, then revert to their humanoid forms and...still have their full 100 hit points or whatever from before they were changed.

I thought this was shapechanging magic - I didn't realize the spell encases the target in a t-rex-shaped shell which wards them against harm until the enemy does enough damage to break them out of it. The way I understand it, the target's body is transformed - the t-rex form is still their body. It's body takes damage, the target's body takes damage.

"Well, the target's armor class is usually worse when they transform into a beast, so if they're easier to hit and still take damage as normal then the spell just sucks!" one might say, to which I would say, "No it doesn't!"

The benefit of polymorph and its ilk is that the target can assume myriad forms to suit any situation. With the use of one spell or class feature the affected creature can climb, burrow, fly, swim, become faster, stronger, better able to concentrate on spells, gain new proficiencies, heightened senses, and - with more powerful magic like shapechange, true polymorph, and Elemental Wild Shape - squeeze through cracks as narrow as an inch wide, gain immunities and resistances, cast spells, unlock a new suite of supernatural actions, speak new languages, wield a flame whip and lightning sword (assuming your new form gets the typical equipment included in its statblock, which - don't get me started), and even with all that, they retain their alignment and personality.

So yeah, the tradeoff should be that the target's body is their body and their hit points are their hit points.

Changing Shapechanging

I like that in D&D a creature can change into a dragon or a beholder or a giant or whatever. I really do. It's an awesome ability, but, I also get tired of the t-rex polymorph meta. Shapechanging magic is immensely powerful without having game-changing combat utility. In my most recent 5e game, the warlock spent levels 17 to 18 as a death knight and level 19 as a balor. 

It was really cool when they first did it, but for three whole levels in the last tier of the game, they weren't even playing a warlock anymore, because it was just so much better to transform into an "equivalent" monster. If they really wanted to, they could have permanently transformed the whole party into balors (again, pretty cool, but it feels wrong somehow - they also would have stomped every encounter for the rest of the campaign, easily).

So, I'm done with polymorph as written. My house rule for shapechanging mechanics is presented below:

When a creature changes shape (such as with the polymorph spell), a player character can change shape into a creature whose CR equals (at most) half the character’s level (rounded down), unless the feature specifies that the character can only change shape into a creature with a lower CR (i.e. the Wild Shape class feature). When a creature’s new form is reduced to 0 hit points, it reverts to its normal form. Any hit points lost during the transformation are subtracted from the character’s current hit points.

For example, a 20th level wizard with 142 hit points uses polymorph to transform into a CR 8 tyrannosaurus rex with 136 hit points. The wizard’s tyrannosaurus rex form is reduced to 0 hit points, and the wizard resumes their normal form. The wizard now has 6 hit points (142 - 136 = 6).

Shapechanging abilities grant you only the physical traits of the new form. You always retain your Intelligence, Wisdom, and Charisma scores when you change shape, along with any saving throw or skill proficiencies associated with those scores. You do not gain access to any abilities, actions, proficiencies, or spells which the new form has if those features use Intelligence, Wisdom, or Charisma.

Side Note: I have a second house rule for Circle of the Moon druids which grants them 4 temporary hit points per druid level when they Wild Shape (based on the Circle of Spores's Symbiotic Entity feature - Tasha's Cauldron of Everything, page 36), so that Wild Shape maintains some combat utility for them in light of this change. I would also rule that the limitation on CR doesn't affect Elemental Wild Shape, since that feature specifies that you can transform into a more powerful creature than would otherwise be allowed, and specific rules trump general rules in 5e.

System Shock

When I first decided to change shapechanging in my games, I first looked back at Advanced Dungeons & Dragons 2nd edition, it being my "older edition" reference point. It was definitely still a mess back then, including a complicated and poorly laid out system for polymorphing creatures and objects, for which the duration changes depending on the difference between the kingdom, class, "relationship", size, shape, and intelligence of the creature or object and what it is being turned into (2e PHB, page 242). For example a lion can be polymorphed into a sphinx permanently - because a sphinx is like a lion - but a turnip can only be polymorphed into a purple worm for a few hours - because a turnip is not at all like a purple worm. That is the extent of the guidance given.

2e also has cool ideas like polymorphing a creature into a rock and grinding the rock into dust to kill them, but if one polymorphed the creature directly into dust they'd have to "scatter the dust far and wide" in order to effectively kill the creature. I have no idea how I would apply this to 5e, but it is inspiring.

2e also includes system shock (2e PHB, page 21), which is a percentage chance - based on a character's Constitution - to survive body-altering effects like polymorph, petrification, and magical aging. On a failed system shock roll, the change is so devastating to the creature that it dies instantly. The check is made when the creature changes shape and when it reverts to it original form (page 205). Notably, it doesn't apply to the polymorph self spell, so I think the check is only required if the change is done to an unwilling creature.

If I were to introduce system shock to 5e, I'd probably do it as such:

Whenever a creature is unwillingly subjected to a magical effect which reshapes or ages its body (petrification and reversing petrification, shapechanging, magical aging, etc.), that creature may be instantly killed by the shock to its body and mind. The creature must make a DC 10 Constitution saving throw or instantly die.

I wouldn't necessarily encourage this, as it only enhances polymorph's ability to be a pseudo-death spell, which it already is in 5e, if Rules-As-Written white room theory-crafters are to be given credence. Basically, the caster transforms the enemy into a vulnerable creature like a sheep and then surrounds the target with their allies, readying to attack all at once so that the entire party's focused damage is applied simultaneously, before the creature reverts to its original form. 

This is incredibly cheesy and I would never allow it (the first hit to drop the target to 0 hit points would cause it to revert to its original form, in my opinion - roll initiative to see who strikes first), but I thought I should address it in case Redditors decide to nickel and dime me on this point.

Interestingly, polymorph self in 2nd edition (PHB, page 206) only allows the affected creature to benefit from the new form's movement and breathing abilities. However, they can change shape as often as they please while the spell lasts, similar to 5e's shapechange. This was my inspiration for making it so that shapechanging in 5e only grants you the creature's physical capabilities, not mental ones.

Retaining Personality

I always love to make it a point whenever polymorph comes up in my 5e games that the affected creature retains its alignment and personality. "Don't worry, friends - you can tell that the tyrannosaurus rex is me, your loyal ally, because it has my personality and mannerisms."

Don't worry, this is actually the party's barbarian!

2e has another interesting mechanic by which a polymorphed creature might assume the personality of (and believe itself to truly be) the creature it's polymorphed into (PHB, page 205). A percentage chance is calculated based on the difference in Intelligence and hit dice between forms. The polymorphed creature must roll against this percentage each day it's polymorphed, and on a failure, it effectively becomes the new creature and comes under the control of the DM until recovered by a wish spell or similar magic. This total transformation includes the acquisition of all the new creature's magical and special abilities (like a dragon's innate spellcasting and breath weapon). Furthermore, those who are reverted to their original form after this transformation has occurred maintain the belief that they are actually the polymorphed creature and attempt to return to that form.

Again, this doesn't apply to the polymorph self spell, so it might apply to unwilling creatures only. However, polymorph any object (2e's version of true polymorph - PHB, page 242) does state "The system shock roll must be applied to living creatures, as must the restrictions noted regarding the polymorph other and stone to flesh spells." Which restrictions are we referring to?

Stone to flesh (2e PHB, page 230) has some language about the cubic volume of flesh and stone that can be converted (with a delightful mental image of a wizard creating a tunnel of flesh through a solid stone wall), as well as some language about broken or weathered petrified individuals retaining their damage and deformities when unpetrified, so either or both of these might be the aforementioned "restrictions".

The "restrictions noted regarding the polymorph other...spell" might then be the note about retaining personality when polymorphed. This is how I choose to read it, as although the target of the polymorph any object spell could be willing or unwilling (and previous examples suggest that only unwilling targets are at risk of losing their personality), it makes perfect sense that a creature which spends days, months, or years polymorphed into a demon might be at risk of believing themselves to actually be a demon.

A scientist (AKA modern day wizard) who turns themselves into a fly may be at risk of believing themselves to truly be the Brundlefly.

This becomes a potential solution to the true polymorph problem, wherein a 17th-level character expends one 9th-level spell slot one time and proceeds to spend the rest of their adventuring days in the form of a death knight or balor with no risk of consequence. If I were to use such a mechanic in 5e, I'd do it like so: 

Whenever a creature is subjected to a magical effect which reshapes its body into that of another creature, there is a chance that the creature assumes the alignment, personality, and mental ability scores of its new form. 

At dawn each day after the creature is first transformed, the creature must make a Charisma saving throw. The DC equals 10 plus the difference in Charisma between forms (if the new form's Charisma score is higher) or minus the difference in Charisma between forms (if the new form's Charisma score is lower), plus the difference in levels or CR (if the new form's level or CR is higher) or minus the difference in levels or CR (if the new form's level or CR is lower). 

On a failed saving throw, the creature assumes the alignment, personality, Intelligence, Wisdom, and Charisma scores of the new form, gains all associated abilities and proficiencies, and comes under the control of the DM. The transformation can be reverted as normal, but changes to alignment, personality, mental ability scores, associated abilities and proficiencies, and control of the creature by the DM can be reverted only by way of a greater restoration or wish spell, or similar magic. Until that time, the creature believes itself to be its new true form and attempts to return to that form.

I'm using Charisma instead of Intelligence here, because Charisma in 5e is a creature's "force of personality" (PHB, page 173), and this is a contest of personality/sense of self. I'm tempted to leave it as Intelligence, because Intelligence needs all the help it can get in 5e, but Charisma feels like the "right" ability score to use. 

I'm using CR instead of hit dice because CR is the measure of a monster's power in 5e, whereas the same can be said of hit dice in 2e.

Again, I like that shapechanging magic can accomplish such goofy and cool effects. PCs should be able to transform into powerful monsters at a certain point in the game, but the conversion rates are whacky, transforming into a t-rex every combat is boring after the first time, and one shouldn't be able to live out their days as a dragon without risk of consequence. 

Shapechanging magic is extremely versatile without being the dominant meta-strategy in combat during the later tiers of gameplay. I like its combat potential reined in, its mechanics more sensible, and its drawbacks more nuanced - or rather, in 5e, I'd like it to have any drawbacks at all!

On Specialists

I love wizards. It used to be that when I made a character for a Dungeons & Dragons game, I waited to see what everyone else was playing first, and picked a class which met some unfulfilled role in the party. Because, until recently, there was only one Intelligence-based class in 5th edition, this usually meant I'd be playing a wizard.

One thing I've never liked about 5e wizards (and a lot of 5e's spellcasters in general) is how same-y they all feel. Gone are the days when the Dungeon Master picked which spells the wizard had in their spellbook. Players now have total freedom over their spell selection. The scope of the game has narrowed at many tables to be primarily about combat - some might argue it's always been that way, but early dungeon-centric campaigns like Castle Greyhawk seem to have been at least equally about exploration. There is a proliferation of content on YouTube which concerns itself primarily with ranking the "best" spells, which means experienced players who want to optimize their characters tend to pick the same "must-have" spells every time - mage armor, magic missile, shield, misty step, counterspell, fireball, etc. 

(As an aside, this is why I like playing with new players, and try not to steer them towards the "good" spells unless they ask for my advice - they haven't yet closed themselves off to the possibilities of less "optimal" spells, and it's often a delight to watch them find ways to use those spells.)

Certain subclass abilities incentivize picking spells from an Arcane Tradition's school (Arcane Ward is recharged by casting abjurations, diviners can regain spell slots from casting certain divinations, Sculpt Spells only works with evocations, etc.), but even the Savant feature that each specialist gets only incentivizes learning spells from one's specialized school of magic outside of the leveling system (the feature reduces the gold cost of scribing spells, so a specialist wizard is better off learning other spells on level up and then getting the ones they want from their specialized school by other means).

All wizards tend to have the same spells, aside from those oddballs picked up from spell scrolls in treasure hoards and from enemy wizards' spellbooks. An abjurer's Arcane Ward might make them play slightly differently than a diviner with their Portent or an evoker with Sculpt Spells, but by and large, they're all probably going to have very similar repertoires.

Specializing Specialists

Advanced Dungeons & Dragons 2nd edition did specialist wizards differently. Specialists gained additional spells per level from their chosen school, received a bonus to saving throws against spells of their own school, imposed a saving throw penalty on their enemies when they cast spells of that school, had an improved chance to learn spells of their school, and could more easily create new spells within their school. The trade-off was that each school had a school of opposition, from which the specialist was forbidden to learn spells.

Later editions did away with this aspect of specialists along with most other restrictions on player character choice, and wizards became meta-chasing casters like everyone else, casting whatever spells D&D YouTubers told them were most optimal, albeit with a few bells and whistles of varying usefulness.

In my 5e games, I offer a house rule for wizards who choose the abjuration, conjuration, divination, enchantment, evocation, illusion, necromancy, or transmutation Arcane Traditions:

Starting at 2nd level, whenever you gain a wizard level, you can add one wizard spell to your spellbook, in addition to the two you already learn (as described in Spellcasting). This additional spell must be of the school associated with your Arcane Tradition. Additionally, when you make a saving throw against a spell of the school associated with your Arcane Tradition, you have proficiency on that saving throw attempt, even if you normally lack proficiency. When you cast a spell of the school associated with your Arcane Tradition which forces a saving throw to resist its effects, you can give one target of the spell disadvantage on its first saving throw made against the spell.

Additionally, you have advantage on Intelligence (Arcana) checks to learn spells from your Arcane Tradition’s school from spell scrolls.

It's a small bonus, but I think that it, along with the subclass features already included in each Arcane Tradition, gently nudges wizard players towards leaning into their specialization, if they have one.

Side Note: I also imported a spell research downtime activity based on 2e's system, which benefits specialists who choose to create new spells within their school of magic, but that is probably beyond the scope of this post.

Schools of Opposition

Another way to..."encourage" a player to lean into their wizard archetype is to reintroduce schools of opposition - that is, schools of magic from which a specialist cannot learn any spells. This is not something that fits nicely with new school playstyles, and more experienced players who are familiar with "must-pick" spells and the "meta" of wizards may chafe at the restriction. It may be worthwhile to offer the aforementioned house rule as an optional bonus if the player opts-in to locking themselves out of opposition schools. If that isn't to their liking, the DM can just leave specialist wizards as they are.

The schools of opposition are as follows:

• Abjuration: Illusion and Transmutation
• Conjuration: Divination and Evocation
• Divination: Conjuration
• Enchantment: Evocation and Necromancy
• Evocation: Conjuration and Enchantment
• Illusion: Abjuration, Evocation, and Necromancy
• Necromancy: Enchantment and Illusion
• Transmutation: Abjuration and Necromancy

I'll admit there's not much rhyme or reason to this aside from being how it was done in 2e, but one can derive some meaning from it: Wizards schooled in the more subtle arts of enchantment and illusion don't have the skill or inclination to invest time and gold into learning crude battle magic like abjuration and evocation, or the dark arts of necromancy - why reanimate a corpse when you can trick the living into doing what you want? Inversely, necromancers are much too isolated in their gross crypt laboratories to learn magic that relies on understanding people - and why would a wizard preoccupied with the undead choose to learn spells which largely don't affect them? Summoned creatures get in the way of evoker's blasting magic, and they'd rather blow someone up than charm them. Diviners know too much about the nature of the multiverse to start conjuring creatures from who-knows-where. Conjurers use refined, artisanal magic to create more sophisticated effects than the instantaneous blasts of evocation, and they have no need for divination, as they can simply acquire information from the extraplanar creatures they summon, or traverse the planes and find knowledge for themselves.

When one looks at the schools of magic and which spells belong to which school, it becomes apparent how blurry the lines are - why is necromancy, the magic of life and death, not responsible for curative magic? Why is fear, a spell which triggers an emotional state in its targets, an illusion and not an enchantment? 

I think of this as a feature of D&D's magic system, not a bug. The schools are not indicative of what kind of magic is possible within each school, but rather the approach used to achieve the magic's effects. Necromancy is not responsible for healing magic because it primarily deals in "dark" magic. Fear is an illusion because it works like that scene in Fellowship of the Ring where Gandalf makes himself appear very big in order to scare Bilbo back to his senses ("I am not trying to rob you! I'm trying to help you.").

When one thinks of magic this way, its various schools become almost like monastic traditions - wizards are disciplined in a certain approach to magic, and stand in opposition to incompatible disciplines. An illusionist might see a necromancer as an affront to magic itself. The two might engage in battle whenever they meet, and their approaches would look very different - the illusionist may use their magic to confuse and misdirect the necromancer's simple-minded minions, then shatter the rival wizard's mind with powerful fear effects, or the necromancer might come prepared with undead specifically created to see through such magic, forcing the illusionist to resort to trickery and obfuscation to escape the enemy's minions. They're not both going to stand 120 feet apart slinging fire bolt and blocking with shield.

I don't care much for D&D YouTube - content there is often preoccupied with player character builds, ranking character options into tier lists based on their "power" in a featureless white room combat, or trying to be "funny" ("This is what it would be like if Skyrim players played D&D!"). One of the only channels I really enjoy is WebDM, and they have a great series on the schools of magic in D&D, which is a good place to start thinking about these things.

Other Specialists

It isn't just wizards who need to work within the confines of their discipline in 2e. Priests' spells were broken into spheres, with each deity granting and barring access to specific spheres, while druids were simply priests of natural spirits and old gods, with access to spheres that suited their style of worship.

There's a world in which every Divine Domain, Druid Circle, Sorcerous Origin, Otherworldly Patron, and maybe even Sacred Oath and Ranger Conclave has its own, totally unique spell list. I like Domain, Circle, and Oath spells because they offer a happy middle ground, wherein spellcasters automatically have convenient access to the spells that fit their theme, but can still prepare the "optimal" spells they might prefer (although this usually means that the subclass-specific spells don't see much use unless they're particularly "good").

Even then, this method isn't applied consistently. Otherworldly Patron spells are added to the warlock spell list, but the warlock still has to choose them as known spells (foregoing potentially "better" vanilla warlock spells), and some rangers and sorcerers get bonus spells, but not all.

One solution to this is to give all Ranger Conclaves and Sorcerous Origins bonus spell lists (which I've done in my games). An option for warlocks would be to give them their Otherworldly Patron spells for free (which I've done as well). But how do you encourage players to actually use these spells over the "optimal" ones?

Like with wizards, some subclasses have moved towards solving this problem by granting certain benefits when casting subclass-specific spells (and others that fit with the theme) - Life domain clerics heal better, Circle of the Shepherd druids have more durable summons, Storm sorcerers can cause damage to enemies near them whenever they cast a lightning or thunder spell, etc.

I encourage players to lean in to the theme of their subclass's specialization. When I play an Aberrant Mind sorcerer, I'm only taking spells which can pass for psionic effects. When I play a Divine Soul sorcerer, I'm only taking cleric spells. When I play a Draconic sorcerer, I'm playing an elementalist. My Shadow sorcerers are necromancers. My Storm sorcerers only use weather magic, etc.

I find that the world feels much more rich when spellcasters of a given type use different approaches to magic and a different selection of spells. Each wizard, cleric, druid, or the like feels more unique and memorable. Ultimately, this is something players most likely have to decide they want for themselves. The DM can't force this kind of playstyle, but they can offer incentives for players who embrace it.