From January of 2009 to January of 2011, I ran a D&D 4th edition game set in a reinterpreted version of the Dust to Dust setting (as Dust to Dust does not natively have dwarves, eladrin, half-elves, or tieflings, this was necessary). We had a lot of good times with this campaign, and over the course of 45 sessions, the party advanced from 4th level to 13th level. There was a modest degree of player turnover (a total of eight players came through the campaign, though the party had six players for most of its run), and no character death.
I will eventually run 4e again. For now, though, I'll be running Mage: the Awakening, as it was the most common #1 choice when I polled the prospective players (out of a list of five or six options).
I've mentioned before in this blog the ways in which I find 4e dissatisfying. On the whole, I find 4e to be a very good game that is not intrinsically designed for deep characterization; if there is to be deep characterization, it has to come from the players and the GM. D&D 4e is instead a remarkable example of a good pick-up-and-play game at low to mid levels; I say this because I have watched players who are very inexperienced with 4e's mechanics (and are not just natural tactical geniuses) play with a high degree of proficiency even when playing a pre-generated character in the 5-7 level range. This is a meaningful virtue: heroic-tier one-shots are a better game experience in 4e than any one-shot experience I've had in previous editions.
I get the strong impression that the designers had no interest in creating a deep-characterization game. They wanted Wahoo! action and they wanted it fast. Games do need to be approachable for new players, and they need to hook players quickly, so I'm on board that far. A DM and group of players just need to go several steps beyond the stages of character creation listed in the book in order to set up conflicts that will drive the game, but the text explicitly pooh-poohs such an idea as the gameplay of amateur thespians rather than Real Gamers. God forbid you should want characters to have motivations beyond wondering where they'll get their next pile of cash.
The game has elements that steer players away from deep characterization. The first of these is that this game's default state exerts a nigh-overwhelming pressure to Be A Team Player. Not only is the game more fun for everyone else if you're a team player, it's also more fun for you. The downside of this is that the game makes it pretty hard to feel like you, personally, are a rockstar. Killing a non-minion enemy all by yourself is not common. (It isn't amazingly common in previous editions, but it's much more feasible.) This has a lot to do with the scaling of hit points and the importance of granting bonuses to other players, as well as the way combat is balanced so that healers usually end fights without any of their per-battle heals remaining. I can't criticize this too much, as it's where a fight's tension comes from, but even this good design decision has hidden drawbacks for characterization. This situation is not one I would look to change directly, but a wider use of creatures that are lower-level than the party, and a broader use of easy-to-moderate encounters for the sake of attrition, could help to address the Not A Rockstar drawback.
(Actually, let me skip back to Problem Zero: DDI. DDI makes houseruling obnoxiously difficult. My solution to this is to go back to maintaining the party's character sheets by hand. The follow-up problem is that the books-as-printed now have very little in common with the rules-as-revised. The rules patches are pretty insane at this point, and far more than I can keep in my head, which damages gameplay because I don't feel confident in rules calls based on what's printed. DDI is really good for core-rules-only play, such as Organized Play, and terrible the moment you want to step beyond that.)
The second element that draws players away from deep characterization is the economy. I get why the economy of 4e at release was designed the way it was, I see some of the problems of the 3e-and-prior ways, but I still don't like 4e's answers. Maybe next time I run 4e, WotC will have published enough of its material on magic item rarity that I can use that system, because it sounds pretty cool. At present, they've released the basic rules of the rarity system, but haven't classified items by rarity, and it's unclear what - if anything - of the currently published material should be a Rare item. Getting rid of the limit on how many Daily item effects a single character can activate is a welcome change, though.
The third element that draws players away from deep characterization is a strong emphasis on point-buy stats. In 4e, having an 18 in your attack stat at 1st level doesn't make you feel awesome - it's the assumed power level, and falling noticeably below that means you're not going to have a lot of fun. Samhaine and I have both written about this issue at some length. Some of my preferred solutions I have stated in a previous post.
The fourth element that draws players away from deep characterization is that magic lacks any sense of internal logic or structure. There are no Laws of Magic (not in terms of how it works, and not in terms of what is socially permissible), so there's no feeling of explorationor secrecy. Arcane magic... isn't. The different arcane classes don't feel like distinct ways to approach the same universal force of magic. Welcome to Harbinger's soapbox, guys. If wizards don't need to research their spells or capture them from enemy wizards or find them in a crypt somewhere, a fundamental part of what I like about wizards is missing. In theory, 4e's ritual system helps to address this by giving players a different kind of treasure to seek; in practice, it makes the problem even worse because there's no concept of an in-play system, and there's such a strong temptation to mark off components (particularly residuum, the universal component) and call it done. I would be satisfied if I felt like the rules supported having an order of wizards who could in some way benefit from one another, but because the rules don't make players dependent on individual instructors or organizations for any kind of training, there doesn't feel like much reason to include them. The training-as-treasure ideas posited in the DMG2 are a solid start to addressing this last part.
Addressing the general lack of structure to the ways of magic is tough, though. I understand Essentials is doing something about this, but I haven't picked up Essentials stuff. Also, this is a design blog, in case you missed it, and coming up with my own solutions is kind of a thing for me.
My idea for how to fix this is a much more invasive solution that absolutely requires my solution to Problem Zero. My thinking is: Assume that PH 1-3 represent the core functions of classes; for example, the spells listed for wizards are the Common spells that any wizard can learn. The exploits listed for fighters are the Cool Moves that anyone can learn. Things from other sources - Arcane Power, Dragon Magazine, whatever - are hidden lore. To learn things from those sources, you need to do research, loot an enemy wizard's spellbook, or raid a forgotten crypt.
First problem: this pushes some really interesting builds into the realm of Everything Must Be Researched. I would give those builds, such as the brawler build of fighter that I love so much, a special dispensation, and would come up with some other way to sort them out.
Second problem: So you're telling me that you'll let me put forth a ton of effort so that I can retrain one of my existing powers into something else that is mechanically equal*?
Answer: No. So the wizard class (and the swordmage class, for the cost of a feat) has the ability to learn two different daily powers at each level that grants a daily power. At the start of the day, they choose one to prepare. This is really neat, but shouldn't have been as restricted as it is. Give fighters the ability to learn alternate powers that still spend out of the same pool - you still don't get more than one first-level daily power in a day, but it might be Crushing Blow or it might be Seize and Stab, and it all depends on your situation. Since this undermines a wizard's Cool Thing, I would give the wizard more of same to compensate: they get two spell options as a baseline, and then more with research.
The fifth element that draws players away from deep characterization is that just as there is no internal consistency to magic, there exists no internal consistency to what monsters can do. It's designed to make a fun fight, not communicate world information. In 4e, unlike in previous editions, when an NPC wizard casts some godawful spell at you, there's no point to thinking, "That's super cool, and I look forward to learning that spell myself." (Now, some of the spellcaster stat blocks in the Encounter Builder are written this way, but it's rare enough as to not be worth mentioning.) One of the things the designers saw as a 3.x flaw was that PCs and NPCs were built according to the same rules, so NPCs were crushingly complex to create. What they lost by doing away with that was any sense that paying attention to the horrible things the NPC does to you reveals interesting information. Players don't really care that their math won't be the same for that attack as the NPC's was, since they couldn't see the math in the first place.
My means of addressing this is another logical conclusion from my solution to Problem Four. If the NPC is human (well, mortal; "on a PC power scale" is what I'm trying to get across here), he should be statted with something generally like the same powers that a comparable PC uses, specifically so that when an NPC or group of NPCs starts doing something unfamiliar, that unfamiliarity has meaning. "Where did you learn to do that?" can be a valid question. If it sounds cool enough, maybe it's something a comparable PC will want to do, and suddenly you have a new motivation, minor quest goal, and built-in reward.
That's all I've got on this topic for now, but another massive and rambling post on this subject is a constant danger. The Blog of Damocles, if you will.