No need to strictly limit ourselves to what has come before in terms of exact mechanics. There are a lot more options that jut those three.
I understand why Vancian spell-casting is so popular for D&D. To a certain a level, it is much easier to track than point-based systems, since you just need to remember which spells you have cast already that day. However, there are a few key problems with it that need to be addressed. While it is pretty easy to track at low levels, when you have only a few spells, if becomes significantly harder to track at higher levels when a Wizard can memorize upwards of four dozen spells from a list of hundreds of spells. There is also the problem that magic spells don't scale well: a level 3 spell starts out as a rare and powerful spell, but eventually becomes a relatively weak and common one.
I think a heavily modified Vancian system that addresses some of these issues would work great for D&D. For one, it would be nice to be able to scale magic spells better. What if you could prepare a level 1 spell like magic missile in a level 6 slot, and it be as strong as any other level 6 spell? 3E metamagic tried to achieve that, but was limited by its "one-size-fits-all" feat approach. What if spells had scalable parameters built into the spell description itself, based on spell level? Alternatively, we could change the spell slot system from a set of ever raising tiers of spells to a simpler Minor, Lesser, Greater (all three of which are prepared daily) scheme, where spells are inherently designed to scale with level better. In this system, a spell like Fireball would always be a Lesser spell, but would do more damage and have a bigger area of effect at higher level. New spells would still be unlocked as the Wizard goes up in level. That way, spells meant to be used less often won't eventually become readily available at higher levels, and the Wizard class complexity would remain reasonably manageable as character level grows.
I understand why Vancian spell-casting is so popular for D&D. To a certain a level, it is much easier to track than point-based systems, since you just need to remember which spells you have cast already that day. However, there are a few key problems with it that need to be addressed. While it is pretty easy to track at low levels, when you have only a few spells, if becomes significantly harder to track at higher levels when a Wizard can memorize upwards of four dozen spells from a list of hundreds of spells. There is also the problem that magic spells don't scale well: a level 3 spell starts out as a rare and powerful spell, but eventually becomes a relatively weak and common one.
I think a heavily modified Vancian system that addresses some of these issues would work great for D&D. For one, it would be nice to be able to scale magic spells better. What if you could prepare a level 1 spell like magic missile in a level 6 slot, and it be as strong as any other level 6 spell? 3E metamagic tried to achieve that, but was limited by its "one-size-fits-all" feat approach. What if spells had scalable parameters built into the spell description itself, based on spell level? Alternatively, we could change the spell slot system from a set of ever raising tiers of spells to a simpler Minor, Lesser, Greater (all three of which are prepared daily) scheme, where spells are inherently designed to scale with level better. In this system, a spell like Fireball would always be a Lesser spell, but would do more damage and have a bigger area of effect at higher level. New spells would still be unlocked as the Wizard goes up in level. That way, spells meant to be used less often won't eventually become readily available at higher levels, and the Wizard class complexity would remain reasonably manageable as character level grows.