Tuesday, August 6, 2013

How old school D&D gets magic (mostly) right

Wow, it's been a long time.  I haven't run a game, nor written anything game-related since that last post about hit points way back in February.  Gaming has been, if not the farthest thing from my mind, at least a few time zones away.  Suffice it to say that anxiety disorders are a royal bitch.  But now things are settling down, the gaming bug has started to creep back in, and the mental tranquility and focus that have eluded me for so long have mostly returned.  With that, let the fire be lit anew in the hearth and the doors of the Flagon cast wide once again!

When last I wrote of things gamey, I had just taken the first few steps down a slightly different road for this blog, seeing the old B/X rules with new (and perhaps old) eyes and appreciating their simple elegance rather than trying to completely overhaul them to conform to some pie-in-the-sky concept of how fantasy settings, stories, and natural laws "should" manifest in a game.  Bottom line, it's a game, not a novel, and I'm all about appreciating it as such.  Perhaps surprisingly, that hasn't changed over my hiatus.

In fact, I find that I even have a new appreciation for the much-maligned quasi-Vancian magic system of early D&D.  Not every aspect of it, mind you, but we'll get to that a bit later...

I think it's fairly self-evident that in a game, magic must be limited and restrained somehow.  It's all well and good in a novel or a movie to have a magic-wielder manifest powers as the plot demands, but a game requires clearly comprehensible limitations.  Unlimited use of automatically successful abilities makes for overpowered characters and boring games.  The tension and excitement of the game comes from uncertainty, which is accomplished in several different ways.  The most obvious one is randomness, a recourse to the dice to decide the success or failure of an action.  Another source of uncertainty is the asymmetry of knowledge between DM and players.  The players don't know what's behind the next door or inside that locked chest until they interact with it in-game, and the DM can never be completely sure what the players will do. Asymmetry of knowledge also generates uncertainty in the management of finite resources.  The utility of using a resource now must always be weighed against possible future need.  A flask of oil might be just the thing against that owlbear, but will you live to regret it when your lantern gutters out three dungeon levels below the surface?  Different methods of introducing uncertainty to an adventure encourage different strategies of coping with the uncertainty, which ultimately is one of the most important factors distinguishing one character class from another.

I find that I really don't care for a "roll for success" mechanic for spells.  It does limit their utility, but it makes magic feel too much like any other skill.  A fighter rolls to hit with his sword; the mage rolls to cast magic missile.  The thief rolls to pick the lock; the mage rolls to cast knock.  The classes, to me, feel very homogenous this way.  The different approaches to adventuring represented by a class system become pure fluff.

Making magic a limited resource allows it to function very differently from mundane skills.  Spells often allow automatic success, but you only have a limited number of them per day, so you must be judicious in deciding when to use them and when to rely on less certain but unlimited means.  Using magic to solve problems becomes more than a question of flavor; it's instead a choice between the uncertainty of the dice and the uncertainty of finite resources against an unknown future. 

Limited Resource systems of magic usually take the forms of spell point systems and spell slot systems, including traditional "Vancian" D&D magic.  I've actually tried a spell point system, and while it wasn't terrible, I found it less than satisfactory.  Various formulas exist to determine a character's pool of spell points, with one of the most common being simply to take the total spell levels usable by a spell caster under the rules-as-written and convert it to spell points; e.g. a fifth-level magic-user who normally gets 2 first, 2 second, and 1 third-level spell would instead have a pool of 9 spell points.  The advantage I perceived in this type of system is that spell points are completely fungible.  Any given point may be used to power a first-level spell or a ninth-level spell, as long as you've got enough of them.  Magic-users aren't bound by a spell table; they can cast whatever spell of whatever level they like, as long as they've got the points for it.  The problem I have with this approach in hindsight is that it's just too fiddly for my taste.  There's that much more math to track, and worse, it takes the focus off the spells and puts it squarely on the math.  It's not that it's difficult math, or even that the numbers are difficult to remember, but in my experience the more adding and subtracting of abstract numbers you have to do at the game table, the more it diverts attention from the adventure itself.  (Yes, sometimes the goals of game simplicity and immersion in the story do align.) 

Another difficulty with a spell point system is low-level spells that scale with caster level.  A 9th-level magic user who gets five magic missiles per casting could dish out 5d6+6 points of damage per round 31 times in a day in a 1-spell-point-per-spell level system.  There's almost no reason at all for him to even bother with fireball or lightning bolt, the wizard's traditional big guns.  Two rounds of magic missiles, for 2 spell points, deals 10d6+10 damage, eclipsing a fireball's 9d6 for 3 spell points.  For that matter, a very high-level wizard could also "cash in" his allocation of eight and ninth-level spells to add multiple 20-die fireballs to his daily limit.

A spell slot system is sort of a spell point system in broad strokes, but with a separate pool of points for each spell level.  At a glance this seems more complex and fiddly, but since we're counting spells instead of spell points, it's actually a degree less abstract than a pure spell point system.  The focus is on the spells themselves, not some nebulous mathematical pool of power. I suppose, if you're one to ponder the underpinnings of your magic system, that this implies some sort of quantum property of magic - magical energy comes in discrete quanta rather than a smooth continuum - but you can also easily hand-wave such esoteric notions and accept it as simply a gaming convenience.  The point is that you and your players are dealing first and foremost with spells, not subtraction problems.

Where I part company with traditional D&D magic is the "fire-and-forget" rule that seems to be the one uniquely Vancian quality of the system, but this really is not an integral part of a spell slot system per se; it's merely an additional rule superimposed upon it, and a spell slot system functions just fine without it.  In my opinion it's overkill; it turns a necessary limitation on magic into a straitjacket.  It forces the player either to anticipate the uncertain future in great detail, or to select only those spells likely to be useful in almost any adventure.  The result in my experience has been almost always the latter, with potentially useful but narrowly focused utility spells eschewed in favor of stuff that will make the biggest impact in combat.  Why bother to memorize read languages or hold portal when the likely outcome is tying up a spell slot that would have been better used for another sleep spell?

Some DMs do away with memorization entirely, and allow magic-using characters to choose freely from among the spells they know when casting, while limiting them to the total number of spells per level given in the rule book - a pure spell slot system with Vancian elements removed.  I've played that way, and it works well.  The major drawback comes into play when a character has a large repertoire of spells, and things bog down as the player is constantly poring over the list for the perfect one for each occasion as it happens.

Maybe you do want a certain element of preparation, but more akin to cramming for a test or drilling a particular set of plays for a sporting event rather than loading your mind like a revolver.  I'm given to understand that the ACKS game allows a character to memorize a number of spells per level equal to the usual allotment, but once they're memorized, the caster can choose freely from among them.  A level 2 magic-user could memorize two first-level spells, and if she needs one of them twice and the other not at all, she can cast accordingly.  She doesn't forget either, until she wishes to memorize a different selection. 

If you like, you could add the character's Intelligence or Wisdom modifier to the number of spells of each level that can be memorized (minimum of one per level to which the character has access, in the case of low scores with penalties), without inflating the number that can be cast per day.  This would make the character's prime requisite relevant, allowing brilliant mages and wise clerics to be a bit more versatile than their less-talented colleagues without making them absolutely more powerful. 

You could also allow a lower-level spell to be cast with a higher-level slot, if all of the proper slots have been spent.  A magic-user who's spent all his first-level spell slots and wants to cast ventriloquism can do so by expending a second-level or greater slot - but it still counts as ONE spell; it can't be broken up into two first-level spells.  Optionally, perhaps using a higher-level slot than necessary might grant extra power to the spell.  This might be a good way to handle level-scaling spells, especially if you find that they unbalance your game under the rules-as-written.  A caster wishing to boost damage from, say, a fireball, could use a fourth or fifth-level slot to cast it, with the damage being 2d6 per level of the spell slot rather than 1d6 per caster level.  Magic missile would render one missile per level of the slot used.  You want the biggest bang, you better be prepared to pay for it.

With another tweak you can model the effects of an exhausted mage attempting a last-ditch blast or a novice exceeding his understanding, by allowing higher-level spells to be crammed into lower-level slots.  Need to disintegrate something but don't have a sixth-level spell slot?  Cross your fingers and use a level four!  It should be fairly simple to devise a mechanic that represents the dangers of such an action, with the hazard increasing the greater the disparity is between spell and spell slot, though I haven't done so here.  (In the case of the novice who can't memorize the higher-level spell he wishes to cast, he could read it directly from a spell book.)

Strip away the Vancian trappings, and the spell slot system is really pretty versatile.


  1. Great post! ACKS actually has a system like you suggest, where mages with a high intelligence gain extra spells known, without any impact on spells they can cast per day. So a second-level mage with a 9 Int can choose from two first level spells when casting, while a 2nd level mage with an 18 int can choose from 5. However, they both can only cast two spells per day. It seems to work well.

  2. Great post. I do like the idea of hit points for spells because it makes a contrast with the fighter. The fighter risks getting hurt, his armor might protect him, but the wizard guarantees he'll get hurt. And it keeps them both tracking the same resource, hit points. As you point out though, you have to make the spell do damage based on the spell level/hp cost it was cast at (which, that can also keep low-level spells useful by casting them at higher level/damage).
    You also got me thinking about mechanics vs role-playing with regards to magic (and why I don't like the Vancian system w-D&D) which I had to write about on my own blog - so thanks for the inspiration too ! :)