@OP
One thing you can do is make offensive magic deal comparable damage to non-magical options and make utility magic useful for new options, but not overpowering and setting defining.
An example of this is the Iron Kingdoms Full Metal Fantasy RPG (not the d20 Iron Kingdoms):
-A moderately optimized spellcaster will do roughly comparable damage to a moderately optimized ranged character, but less than a moderately optimized melee character - and further optimization rarely gives spellcasters additional damage (and when it does, its from their non-spellcasting actions).
-Most spells have little to no out of combat application. The few spells that do give out of combat benefits either enhance existing capabilities (the Occultation spell gives benefits to the Sneak skill) or are only accessible to very niche careers (the two spells that allow characters to walk through a wall can only be accessed by 3 of the 15 spellcasting careers).
-Spells have a duration of 1 round or "as long as I concentrate on it and stay within 200 feet of the target". This limitation is made up for by the fact that characters are not limited in how many spells they can cast a day, but by what they can cast/concentrate on in a given round.


Quote Originally Posted by PhoenixPhyre View Post
I'm late to this conversation, but I strongly prefer solving this by narrowing the day-to-day versatility of casters (as well as letting mundanes have nice things by drop-kicking the "guy at the gym" mentality out the proverbial window). I don't mind that there's versatility, but you should have to build for that, and a blaster caster should have to make trade-offs with other spells. Not "oh, I'm not quite as good at them" or ("oh, give me a day, I'll prepare something else") but "I don't know that type of spell."

I made an attempt along the lines of narrowing the versatility for 5e D&D. IMO, 5e is close enough to balance that my primary concern wasn't balance per se--it was thematicity. Basically, I wanted to drastically increase the number of spell lists and reduce the number of spells on each list. In part, I wanted a resource for NPC spell lists, but also as a thought experiment for PCs.

Here's the google doc I created with the changes. I must stress that it's a WIP, proof of concept, least-change (so not rewriting anything I didn't have to) and not anywhere near finished or balanced. But I think the idea's sound, especially for NPCs. (Feedback would be welcome!)

The essence was that I took all ~400 spells I had (PHB + the free sources) and redistributed them between 30 or so "themes". Classes determine which themes a character has access to (including some that can go "off list"). Each character has a primary theme where they pick most of their spells from. They may have one or more secondary theme (where they occasionally pick spells from) or may have another way of accessing other spells. For example, sorcerers don't get a secondary theme, but can cast one spell (of the appropriate level) from almost any list each day--each day it can be something different. Bard have a narrow primary theme selection, but get to poach spells from any other list periodically. Clerics get different secondary lists by domain, but have a relatively small choice of primary lists.

This helps clerics (for example) feel unified but also differentiated. It's not just 2 spells of each level that are set by the domain--it's about half their list that'll be different between different domains.

A particular attempt was made to spread the spells out--there are no spells unique to a single list. Yes, even iconic spells.

In essence, I made each caster a limited-list caster (like a fixed list, but you might have a choice at character-creation/level-up to choose between two narrow lists). You could go even further and kill off/make up a lot of spells, but this serves my purpose.
One issue I see is that for some classes, it is almost impossible to get spells from their secondary theme on certain spell levels.
For instance, if a bard chooses Illusion as their primary theme (which has fewer spells at every level than the Mesmerism school), they have to learn 3 lvl 1 Illusion spells beyond their starting spells before they can learn a second spell from their secondary theme.
The only way that they can learn a 2nd level spell from their secondary theme is by replacing their singular secondary-themed spell they learned at level 1 with a lvl 2 spell from the same theme.
The earliest that a bard can learn a spell from their secondary theme without wasting magical secrets on something that should be available to them is level 6 (and then, it has to be their 3rd lvl 3 spell known). For a Mesmerist Bard, that would be level 14 (and must be their 4th level 6 spell known).
Bards will never be able to learn a secondary-theme cantrip. (I think; the document doesn't list how bards learn cantrips)
Bards do have magical secrets, but what is the point of giving them secondary themes if they have to use class abilities that literally ignore the themes in order to get those spells?

Maybe a better way to do it would be to say that at each level, at least X of your spells known must be from your primary theme, and you can only have primary theme spells known for your highest level spell slots?