When I was a kid, I fell in love with my dad's collection of Dungeons and Dragon's handbooks, and most especially the lists of spells and things. Sure, they were presented kind of prosaically and weren't laid out in a way that was especially useful, but as little scraps of interest, they were incredibly cool to me, and hinted at the neatness of being a wizard... keeping a prop spellbook, and questing for arcane knowledge to grow in awesomeness and power.
When I was first able to roll up a character, then, I was extremely disappointed to learn that what I actually got was a tiny list of what was more-or-less available, with some obvious Best Choices, and of those, I could only cast a very few times a day, and had to decide how many times I thought I might cast a given thing. Which made the Best Choices at the beginning even more important.
What I thought I wanted, at the time, was a removal of the limits on how many spells I could know/keep in spellbook, and a different way of approaching how many spells I could cast in a given day. And to some extent, I am still in favor of some slotted spells, some at-will cantrips, and prepping the spells themselves, rather than individual castings of the spells. That said, having automatic access to everything in the spell-book at level still wasn't, and isn't, to my mind, especially interesting. The problem, at bottom, was that the way one typically gains the spells is hugely uninteresting: to wit, you level up, and then you (most of the time, some DMs vary) just get access to the spells for your level, within the bounds of your int.
Yawn.
I have two basic issues with these things: one which has to do with the way spells are handed out, and the other which has to do with the way one advances at all. We'll address the first to begin with, since it's actually applicable in D&D, and, some might argue, is the way the game ought to be run in the first place.
It boils down to, "treat spells as treasure." Fully, entrench the spells in the story, as items and artifacts that are a part of the world beyond their utility/combat effectiveness. Sure, it is important that a starting mage have spells they can use and have fun with, but couching that in story about how they learned said spells, and creating the expectation that yes, the players will and can uncover new spells over the course of play that can be added to said spell book (whether or not the user can cast them right at that moment) is kind of nifty. I'd like to remove the idea of gaining spells from the idea of "gaining a level," which also brings me to my second point.
The more I play games, the more I think I'm over the idea of levels as the primary form of advancement.
Stands-In-Fire, at one point, suggested a system where you could learn one new skill/ability each session, and the more I think about it, the more I like that idea. I get, and I can pretty much be convinced that, for mages, it's important to limit capacity (ability to cast more powerful spells, and/or number of spells one can cast) in the early game and have that expand, just as I don't necessarily disagree with the argument that a fighter should get better at accuracy and damage as the game progresses. Doing something like that, however, would be antithetical enough to D&D that you might as well start building a new game system around that idea. Which is certainly fine by me. The main idea here, for mages in specific (though I really love the too-underutilized idea of training/specialties as treasure from 4e), is that what advances is now much and at what power you can cast, so the level that matters is the level of the spell. You get new spells through play/research, and pretty specifically through play/research, so it's not grubbing through the boring Player's Handbook pick lists. Importantly to me, though maybe not so interesting to other people, this also removes the desire to plan ahead with spells, because you don't actually know what you're going to find in the course of play, but must think of cool ways to use the spells you earn/learn/find as you go along. Personally, I am not a fan of playing the build game, and I would very much like advancement to be tied directly to what happens in the course of play, rather than planned out in advance based on what is in a rulebook.
But like I said, that could very well just be me.
Since I touched on training treasure above, I want to take a moment to talk about Feat Systems, and why I don't care for the ones I've seen from 3.x to the current iteration in Next (yes, this includes 4e). Largely, I think they encourage optimal combo combo building, which encourages deep system mastery at the expense of newcomers, less devoted to build-play players, and in some cases, the balance of the game itself. The habit of stacking customization in feats, and then handing those feats out at regular level intervals has the effect (once again, to me), of making non-feat levels basically boring stops on one's way to grind to get to the next set of feats.
I don't know what you like to play, but any tabletop game where it seems like a good and worthwhile use of time to grind high-xp monsters over and over to get closer to leveling has missed out on the thing I enjoy about playing a role-playing game: the figuring out of what one needs to go to achieve each individual goal. The goal in these situations is meta (more XP to get to the next level) and it's obvious what one needs to do (grind difficult monsters to get there). And yes, I'm not just talking about 4e-- this happened in 3.x games too. I prefer if the goal is less meta (one of the only times you'll see me say 'less meta' is desirable), more concrete, and more story oriented, for example, "we are seeking the monk who lives atop Forsaken Mountain that she might teach us the Awesome Ways of Awesome." Still, we're talking about advancement, but we're talking about something that's in the story.
And now, I admit that it isn't the feats themselves I mind, it's the way they're handed out, and how they're presented to the players as, essentially, a shopping list to handle at level. I would have no objection to treating them, once again, as treasure, earned through acts and adventuring on the part of the players... you could have tomes of Dwarven Lumberjacking (as a tropetastic example) which teach superior skill in axes, or a cool ability that allows one to do a nifty maneuver while wielding an axe. Or enchantments/blessings that one can temporarily add to weapons, or meditations that allow one to focus one's energy on repelling blows, and improve AC for a time. And these items could pretty easily map to feats as they're written in any of the aforementioned editions, tying their distribution to and through play, rather than to something so arbitrary as level.
To be completely fair, Arcana Evolved _did_ try to do this some with their spellcasting feats, by requiring ceremonies involving certain kinds of beings or creatures in order to obtain them, though the way they were presented was pretty much entirely though the rulebook, rather than being knowledge I think it would ever occur to anyone to organically acquire in play.
Anyway, I also get that, for the most part, this can be handled through the GM's style of game running, rather than needing to be ensconced in rules, unless one is modifying how advancement works, of course. And I imagine that there are already systems in place that do the things I'm talking about more naturally.
But huh, I'd also be open to running a gladiator/arena kind of game where the point was to build an optimal character from the beginning, and swap out pieces/choices between each session to tweak for effectiveness, as something of a crunchy 180 from my play-style, here. Hrm.