I really need to find a nice 4e group, while I don't really think your game experience with it is anomalous, it is still far beyond what I have experienced.
Thanks for the nice reply!
I think post-Gygaxian D&D - as the rules and guidelines are written up by TSR and WotC - sometimes suffers from a lack of transparency about how the game works. (And this is somewhat linked to the thread topic of "balance".)
What I have in mind is this: on the one hand it posits that the GM has principal authority over putting challenges in front of the PCs (and at least in 3E and 4e, gives the GM lots of advice and statistical info to help set the level of these challenges); and on the other hand, the way the PC build rules are presented to players tends to create an implicaion that there is the possibility of building a PC that will "beat the GM" (or, at least, the GM's challenges).
In Gygaxian D&D, the idea of better or worse builds would make sense (although the classic D&D rules don't really support the idea of "builds" for non-spellcasters): after all, it is the
players who choose their challenges (by scoping out the dungeon, making choices about what they will hit, and thereby doing their best to amass treasure - see eg [MENTION=3586]MerricB[/MENTION]'s recent thread about his current AD&D game).
But in a game in which the GM sets the challenges, then assuming (i) that the challenge-building advice for the GM is actually reliable, and (ii) that the GM is applying it sensibly, all "builds" really do is re-arrange the balance of expertise among the PCs. But if the GM is setting challenges that speak to the concerns of the players, this will not be a problem - it will be what is desired. So, for instance, because I have a player whose PC is (among other things) about making the world safe from chaos by eliminating magical anomalies, I am more likely to set encounters in which such anomalies are part of the encounter (and the skill challenge/trap/whatever is factored into the XP budget). And then the players as a whole are no worse off for the party containing the builds that it does: if they had another strong combatant in lieu of the mage, I would just set fewer challenges involving magical anomalies and more involving straight-up combat.
In this context, the principal balance issue is one of overshadowing/niche-protection: ie it shouldn't be the case that two players, both building (say) ranged strikers, can end up with vastly different capabilities in ranged combat but with no offsetting domain of alternative expertise for the weaker striker. I don't think 4e is perfect for this, but luckily I've found it to be reasonably robust. (Oddly enough in my game it is the sorcerer who is in danger of overshadowing the ranger rather than vice versa.)
I don't know if the above makes sense - the ideas are reasonably clear to me, but I'm not sure I'm expressing them well.
But ploughing on, two further comments:
The most transparent RPG I know that reflects the thinking above is Robin Laws's HeroQuest revised. PC builds are numbers next to freely chosen descriptors - so all PCs start identically in numerical terms, and differences are entirely in the fiction. So there is not even an illusion of "winning" the game via clever build - all you do at build is choose what sort of character you want to play in descriptive/concept terms, and then actual play is all about doing your best to engage with and transform the ingame fictional situation so that you can bring your best descriptors to bear. (A bit like a 4e skill challenge - I think the latter was at least somewhat influenced by HQ in its design.)
A corollary of this transparency of PC building is the simplicity of the encounter-building guidelines for HQ revised: it is simply a DC chart (again, quite a bit like the one for 4e), where the GM sets the DC based on (i) the average bonus the PCs have next to their descriptors, and (ii) the desired difficulty of the encounter. So there is not even the illusion that the playes could somehow "beat" the GM by clever PC building - all the intellectual and emotional effort is shifted into engaging with the fiction.
The second comment: everything I have said above about post-Gygaxian D&D will break down if the GM is not actually setting encounters based on the players and PCs s/he is GMing, but is simply following someone else's script (eg a module or adventure path). Once this is happening, if the the players want their PCs to survive and prosper then they
will have to build so as to match themselves to the pre-scripted challenges. Furthermore, if they know that the GM won't deviate from the script even if the encounters it contains are not mechanically adequate to the players' builds for their PCs, then it does become possible to "beat the GM". I don't know how much of 3E and 4e play takes place in this sort of context, but my impression is that it is a non-neglible amount.
My own view is that this sort of context generate a significant and distorting pressure on the game design, and that it is a very challenging context in which to generate both satisfactory rules and really satisfying play experiences. But I think my view on this is probably a minority one.