Interesting. I find that I build more characters for fun in 4E than I did in 3E. Just because I make the same number of decisions when building a character at any given level, does not mean that those decisions are necessarily the same. I find that with 4E, I always want to pick more things than I have slots for. It forces me to make a decision on what to choose. If I choose one option, I have to choose to give up a second or third good option. I never feel that the character is done and look forward to choosing something more the next level. In the end, even if I do not plan to play the character, I build them with the intention that they would play differently.
I'm not surprised some find 4e better for building characters. There are at least a few reasons I'm not that guy, with examples after I list them. First, in 3e feats were more likely to be very meaningful or transformative to the character. That is, you could base an entire concept around a set of feats, or even a single feat, that would completely change the way character was built, played, and in some cases
felt. Second, I really love the idea (even just the illusion) of subverting designer intent. That led to both problems and awesome in 3e, but it was certainly much easier to do there. Third, the different mechanical subsystems made building certain characters from the same blocks could lead to wildly different choices (and criterion for choices) while building as well as during play. Fourth, the multiclassing system let me put all those elements together in unexpected fashions. In short, building 3e characters was a minigame with a lot of freedom. Freedom not necessarily defined as number of choices, but by what impact the choices made can have.
Examples:
1. Transformative feats: Arcane Manipulation, Arcane Strike, Knowledge Devotion, Spring Attack, to name but a few. Each of these feats could form the basis of doing something unique with a character. In retrospect, I like that 3e has fewer feats because each one could have greater relative significance than in 4e. Yes, there are still more good feats I want in 4e than a character can afford to have, but they are less able to redefine a character than simply refine it. This was intentional, to make sure class was central to character identity, but the designers went a little too far for my taste.
2. Subverting intent: Any trip to a charop board would demonstrate these. I'm not advocating broken builds, but those which achieve something unexpected. A non-optimal but resonant example for me is Wizard/Combat Medic.
3. Different mechanical subsystems: Sorcerer vs. Wizard, both of which choose from (almost) exactly the same spell lists but achieve different results. I prefer sorcerer because I have to make careful choices once, and the character must stick with them to the end of his days. When I add in the constraint of a theme (say storms) the challenge of truly exemplifying the theme while remaining mechanically up-to-snuff is intoxicating.
4. Mix and match: Third edition multiclassing is the glue that makes much of the above possible. I mourn its loss in 4e, even as I celebrate some of the benefits that same loss brings to the table.
More concretely, let me describe one of the last 3.5 characters I built (for a friend) to see if that helps. The initial theme was knowledge, and was basically non-negotiable. This friend enjoys characters which smash things physically, and especially liked the idea of using knowledge specifically to smash things better. This idea turned into a Con/Int based dual-wielding ranger with the Knowledge Devotion feat. It traded spellcasting for a few bonus feats and otherwise depended on knowing everything about monsters for attacks and damage. In concept and execution, I am fond of that build. It was by no means uber (not lich-proof as it turned out), but it tanked at least as well as a decent fighter, had surprisingly good defenses, and met the knowledgeable character niche of the party in an interesting way. Aside from not being multi-classed at all, I feel it exemplifies the ideas above.
I like a lot about 4e character building as well, just not enough on the balance. I'm mostly happy that skill points are gone, the execution of racial feats is pretty solid, multiclassing does allow easier and effective dabbling, and the hybrid classes show promise (1st iteration was rough, second iteration not quite good enough but darn close IMHO). Some of the newer feat options are also excellent (in play and from my perspective about building). In particular, I really enjoy the elegant 4e implementation of familiars -- the active/passive mechanic is nearly perfect.
I personally got over the fact that non-casters use one mechanic for their abilities and casters use another. I personally see no value in one character that has a lack of options and another character having an overwhelming amount of options.
Spellcasters got way too much face-time in 3e. That said, my problem isn't usually with the number of options available (numerically staggering even this early on in 4e...1300 feats!) but with what any given option can achieve. A character has a limited and more-or-less fixed number of choices that can be made, and so the limit of character customization is not defined primarily by the 1300 feats, but by the 18 feats you can actually choose (unless there aren't at least 18 feats you want). In my case, there are plenty of feats that I want for a given character, but very few resonate with me as being somehow key to the character. I currently play a 4e monk (multiclass wizard) that is both an alchemist and ritual user. That resonates with me thematically and mechanically, and it helps define the future course of the character. I've just found building characters in this fashion in 4e to be much more difficult than in 3e.
I'm not sure what your complaint is here. If you want to keep your Sleep spell from level 1 to 30, you can.
As for the reason for the power swapping, I think that the intent is to not overwhelm the player with so many encounter and daily powers. I find that a level 1 character already has a great many options in combat. Adding 3 more encounters, 3 more dailies, plus a bunch of utility powers and feats would progressively increase the number of options and things to keep track of at higher levels. Adding 4 more encounters and dailies may not really add much value when playing epic characters.
I think I didn't really state my problem clearly. You are certainly correct that power swapping is designed to keep the player from becoming overwhelmed. That is a good design intent, and the solution used is pretty clean. But it does have this troublesome feature: It means the character (or player) must potentially choose between the thematically optimal choice or mechanically optimal choice
after having already decided a particular power was appropriate for one reason or another. Theme vs. mechanics is always a present tension, but please keep that out of character continuity unless it actually makes sense for the character. Thus, I like that the option exists for a player to keep the sleep spell from level 1-30, even though it is suboptimal. But, I dislike that the mechanical incentive to swap spells is so strong. In 3e this was less a problem: You learned a spell, and even though it became less useful over time you still knew it and could probably find new spells along the same lines at higher levels (in part because so many spells existed).
What I would prefer is that 4e adopt a freer power system (as in my first post). Some classes might not ever swap out powers (leading to greater resource management) and some players really love that style. Others have very limited resource management (maybe only a few slowly improving at-will powers) because some players really love that style. Or another class where the at-will/encounter/daily scheme is kept intact, but when you would normally swap a power you instead apply an "upgrade" to an existing power of the same type (my preferred solution for sleep, since I don't want a bazillion power cards). For some classes the power swap idea gives me no pause but for others it does. The overall number of options printed between classes could remain about the same, but the ways in which those options can be utilized is more diverse than standard 4e.
yarael said:
So to sum up. Detail guy looks at 4E and sees significance in his choices because it limits his options. He sees variety do the sheer number of choices he gets to make, as well as the plethora of options available for each choice.
Big Picture guy looks a 4E and sees no significance in his choices, because they do not affect the quantity and types of choices he makes. He sees homogeneity in all his choices because none of them have the magnitude of significance upon later choices in character creation or game play that 3.X did.
That was an excellent post, and I think sums up my thoughts better than I've yet thought them.

One small difference -- I'm definitely "big picture" guy in your analogy, but I do like choices that limit my options. Key point: I like making those choices myself, and not having them made for me. It's like poetry, sometimes the most constrained forms yield the most fascinating results, even though "free poetry" could strictly speaking contain any possible poem ever. But I get to choose the form, and then make something wonderful with it.