Willie the Duck
Hero
Okay, I have reread a few Dying Earth stories, looked up Gygax's references to spellcasting in the early days, and checked out some of the alternate systems people have referenced. Here are some thoughts --
Regarding Vance's writing -- one thing thing that struck me solidly is that Vance's spells did seem like an arsenal (or as mentioned upthread, Q's Bond gadgets/a Chekov gun, since they always ended up being the perfect solution for a problem which came up). Mostly in that a wizard almost never got to re-charge or change out their allotment in a plot-timely manner. This emulates the supposed original Gygax intent of the spell allotment being per any single adventure (each one usually being a night's worth of gaming), and the framework fuindordm mentions about him thinking highly-powerful/limited use spells would be the most fun. I think that's probably one of the biggest challenges D&D has had to deal with ever since -- stopping players from making the tactically reasonable decision to rest and recharge at every opportunity. Sure there are doom clocks (or environments where resting is as perilous as pressing on with some spells burned), but then you need your adventures to look like that. Excluding 4e/13A, the standard way D&D and D&D-alikes have handled this has been to simply state a standard of expected encounters (6-8 in 5e, 4/day in 3e) and note that if you deviate from this, it will effect the inter-class balance. Certainly a way to deal with it, but also unsatisfying to a significant subset of gamers.
Regarding other systems -- mana points vs. con-damage vs. weaker-but-at-wills and so on seem like... not minutia, but specifics where the other nuances of the game rules may well overcome them in terms of the impact of magic. What strikes me as a near-universal theme is that other systems rarely have the 'usually takes a single combat-frame and always works' quality that D&D has. Sometimes it is a skill test, other times charge up time, or major constraints on when you can readily do so (Tekumel and the 'metal makes you blow up' quality being a pure example there). AD&D specifically had a taste of this in spell disruption (in theory 3e did as well, but getting around it was entirely too easy), but there is so much variability in how much of a hindrance that is (notably whether your caster was behind a retinue of hirelings in 10' corridors or not, but also the incredible variation in whether people actually used the AD&D initiative rules -and correctly. Also then it has no meaning for spells cast away from a fight). Fundamentally I think if utility spells were just another test you made like a skill check there might be less frustration with magic being able to do what a skill-expert does.
Another thing I noticed while reading spell lists (and yeah, massively many other games also use discrete lists of spells with very solidly defined effect parameters): very few games have spells (especially ones without lasting costs or consequences that you can cast day after day) that scale as high as D&D spells do (relative the system as a whole). Nor in many are you usually expected to cast spells as often as even TSR-era D&D casters. There are lots of systems where a TSR-era Charm Person would be considered hugely powerful, or where a Magic Jar-like effect would be something you spent days recovering your power pool from casting (or like Symbaroum --where all magic is risky--has a baleful polymorph-like effect, but it's considered 'the broken spell'). Very few games have magic that is so fire-and-forget, so convenient, and so disruptive. OTOH, this is also true of many D&D magic items, and this leads me back to another thing I realize about AD&D -- getting new spells was hard. They were often found as treasure and became the magic users' part of the loot, very much like magic items. The loss of the uncertainty of having the spells you want is a major change in how the game has progressed, similar to the removal of spell disruption. Still, at the same time, I think each individual change to how inconvenient spellcasting used to be is very very reasonable, since 'this is powerful, but a pain in the rear' isn't necessarily a great balancing mechanism (but then the effects should have been scaled down or the like).
Regarding Gygax:--
Regarding Vance's writing -- one thing thing that struck me solidly is that Vance's spells did seem like an arsenal (or as mentioned upthread, Q's Bond gadgets/a Chekov gun, since they always ended up being the perfect solution for a problem which came up). Mostly in that a wizard almost never got to re-charge or change out their allotment in a plot-timely manner. This emulates the supposed original Gygax intent of the spell allotment being per any single adventure (each one usually being a night's worth of gaming), and the framework fuindordm mentions about him thinking highly-powerful/limited use spells would be the most fun. I think that's probably one of the biggest challenges D&D has had to deal with ever since -- stopping players from making the tactically reasonable decision to rest and recharge at every opportunity. Sure there are doom clocks (or environments where resting is as perilous as pressing on with some spells burned), but then you need your adventures to look like that. Excluding 4e/13A, the standard way D&D and D&D-alikes have handled this has been to simply state a standard of expected encounters (6-8 in 5e, 4/day in 3e) and note that if you deviate from this, it will effect the inter-class balance. Certainly a way to deal with it, but also unsatisfying to a significant subset of gamers.
Regarding other systems -- mana points vs. con-damage vs. weaker-but-at-wills and so on seem like... not minutia, but specifics where the other nuances of the game rules may well overcome them in terms of the impact of magic. What strikes me as a near-universal theme is that other systems rarely have the 'usually takes a single combat-frame and always works' quality that D&D has. Sometimes it is a skill test, other times charge up time, or major constraints on when you can readily do so (Tekumel and the 'metal makes you blow up' quality being a pure example there). AD&D specifically had a taste of this in spell disruption (in theory 3e did as well, but getting around it was entirely too easy), but there is so much variability in how much of a hindrance that is (notably whether your caster was behind a retinue of hirelings in 10' corridors or not, but also the incredible variation in whether people actually used the AD&D initiative rules -and correctly. Also then it has no meaning for spells cast away from a fight). Fundamentally I think if utility spells were just another test you made like a skill check there might be less frustration with magic being able to do what a skill-expert does.
Another thing I noticed while reading spell lists (and yeah, massively many other games also use discrete lists of spells with very solidly defined effect parameters): very few games have spells (especially ones without lasting costs or consequences that you can cast day after day) that scale as high as D&D spells do (relative the system as a whole). Nor in many are you usually expected to cast spells as often as even TSR-era D&D casters. There are lots of systems where a TSR-era Charm Person would be considered hugely powerful, or where a Magic Jar-like effect would be something you spent days recovering your power pool from casting (or like Symbaroum --where all magic is risky--has a baleful polymorph-like effect, but it's considered 'the broken spell'). Very few games have magic that is so fire-and-forget, so convenient, and so disruptive. OTOH, this is also true of many D&D magic items, and this leads me back to another thing I realize about AD&D -- getting new spells was hard. They were often found as treasure and became the magic users' part of the loot, very much like magic items. The loss of the uncertainty of having the spells you want is a major change in how the game has progressed, similar to the removal of spell disruption. Still, at the same time, I think each individual change to how inconvenient spellcasting used to be is very very reasonable, since 'this is powerful, but a pain in the rear' isn't necessarily a great balancing mechanism (but then the effects should have been scaled down or the like).
Regarding Gygax:--
I think this is spot on. While people argue which of Appendix N is most represented in D&D, I think the EGG always just wanted more Conan/Grey Mouser/Cugel than Tolkien set dressing, but his real main focus was a preparation/logistics/weighing-of-options problem-solving game .This all makes clear that the rationale for "Vancian magic" has nothing to do with evoking a certain fantasy feel, or with "playability" in any general sense. (Those may or may not be happy by-products, depending on the game participant.) It's purpose is to create a particular sort of space for game play. Like equipment, it's part of the "load out" aspect of play. (As is well known, it also breaks down at upper levels where load outs become so extensive that the resource management aspect is diluted or near-absent.)
...
The fact that a system intended to emulate classic D&D dungeoneering play nevertheless has to depart from Gygax's design specs for magic-users shows the demandingness of those specs. It's hard to create a game piece that will be fun to play, and whose play is almost entirely the deployment of limited-use problem-solving abilities, but whose access to those abilities won't break the game. Much of the history of D&D design and play approaches - LFQW, class "spotlight" balancing, 4e's distinctive approach to PC building, 5e's "Neo-Vancian" casting, etc - are responses to this design challenge.