In First Fantasy Campaign, Dave Arneson describes in a brief thumbnail the original magic system he used in what would become Dungeons & Dragons.
In Blackmoor, magic followed the "Formula" pattern for most magic. The reason behind limiting the number of spells that a Magic User could take down into the Dungeon was simply that many of the ingredients had to be prepared ahead of time, and of course, once used were then powerless. Special adventures could then be organized by the parties to gain some special ingredients that could only be found in some dangerous place.It strikes me that this is not entirely gone from OD&D - magic-users preparing scrolls (only at wizard level) for 100 GP/level at a rate of 1 week/level seems to take a page from Arneson's non-Vancian style of preparation. Holmes took it even closer by allowing scroll creation at first level, following the same rules as OD&D. By the book, you could actually get pretty close to Arneson's spell system right there.
Progression reflected the increasing ability of the Magic User to mix spells of greater and greater complexity. Study and practice were the most important factors involved. A Magic User did not progress unless he used Spells, either in the Dungeon or in practice (there was no difference) sessions. Since there was always the chance of failure in spells (unless they were practiced) and materials for some spells were limited (determined simply by a die roll) the Magic User did not just go around practicing all the time. The Magic User could practice low level spells all the time, cheaply and safely, but his Constitution determined how often he could practice without rest. Thus, the adventurers might want a Magic User to come with them only to find him lying exhausted.
So to progress to a new level, one first learned the spells, and then got to use that spell. There was no automatic progression, rather it was a slow step by step, spell by spell progression.
Taking a step back, the Vancian system is one that has survived not so much out of sentiment but because it is dead simple to use in a game. Like hit points and armor class, the specific rationale is second to the fact that the system is very effective in game. It limits the magic-user in a readily defined fashion and keeps the bookkeeping manageable.
For many players, though, Vancian casting is a weird limitation. It requires tactical choice, which is good, but very little flexibility. Once out of spells, the MU is useless. A lot of alternative systems such as spell points try to alleviate this by making the MU super-flexible. But with this advantage the game tilts entirely in favor of the spellcasters, who already get powerful at high levels with Vancian casting.
I like Arneson's concept because it invokes a bit more resource management than traditional Vanciancasting. A spell is not just an investment of a spell slot, but is a permanent resource bought with money and/or effort, and available to the magic-user as long as he or she is alive. The decision to use or not use a prepared spell is one that has lasting consequences and cannot be idly cast just to use up a spell for the day.
To use this I think the OD&D pricing is a good start. The prices are about right where an MU will not suddenly become super-powerful, and the referee might give a free spell or two at the game's start to speed things along. Also, an alternative method of preparing spells from ingredients that the MU has to search for at low levels might be a natural source of adventure fodder. The spell components for AD&D are one possibility, as are randomly determined components a la Arneson.
In practice, then: spells would cost 100 GP/level, but take only 1 day/level to prepare. The trade-off is that more spells can be prepared than could be cast. Magic-users would have alternative but difficult methods of finding ingredients. A spell, once cast, is used up. Beginning characters would have 2 spells already prepared. Characters can only cast spells of levels they would be able to cast per the OD&D Magic-User chart (i.e. level 2 at 3rd, etc). I'm also thinking that after some time a spell could start to go unstable.
Has anybody used this kind of system? Any thoughts on potential side effects? Maybe some differentiation in the pricing?