There's a real argument to be made, however, that "game balance" is in some cases a forebear to "narrative balance" and "story balance." Which is, if I'm reading 4e players correctly, the strongest argument to be made about 4e as a whole --- "The 'game' is balanced between classes, particularly in combat, and therefore is able to better support narrative balance and story balance, because the mechanics drive the players into more equitable situations at the table where their involvement matters."
I really liked your post.
It seems to me that a lot of these discussions about balance make assumptions about the
point of RPG play, the limits on mechanical design, the roles of the participants, etc, that are simply unwarranted, and perhaps suggest a lack of familiarity with the range of games out there.
For instance, a striking feature of D&D in all its iterations - including 4e with some marginal departures from this in relation to skill challenges - is that players always have an incentive to bring everything to bear so that their PCs win. For instance, the only reason you wouldn't try and enter a combat with full hit points is because you might not have the resources to achieve such a goal (or might think that those resources need to be saved for something else). Contrast this with Burning Wheel (as just one example), in which mechanical advancement of your PC requires facing a certain number of (near-)unbeatable challenges - and hence players have a reason not to always try to bring their maximum number of dice to bear, have a reason to sometimes have their PCs engage in conflict even when injured or otherwise depleted of mojo, etc.
This difference between D&D and BW means that balance has a quite different meaning across the two games. For instance, in BW a PC who starts with lower stats will face more (near-)unbeatable challlenges, and hence advance more quickly, than a PC who starts with higher stats. So the game's balance is (somewhat) self-correcting.
Now you might say that this is just like old-fashioned D&D, in which a 1st level PC adventuring with an otherwise 5th level group will gain levels more quickly, and hence also generate a self-balancing outcome. But that 1st level PC is likely to be killed by any opponents or challenge that troubles the 5th level PCs; whereas another feature of BW is its action resolution mechancis, which mean that
failure generally does not entail PC death. And it is not just the action resolution mechanics that support this in BW; it is also the world-building mechanics, which give the players more control over the gameworld than is typical in D&D, and therefore helps them play a bigger role in defining failure conditions for their PCs, and hence makes failure less unappetising to the player (even if just as unappetising to the
PC) than it would be in D&D.
Now the above paragraphs are just a modest comparison of some features of two fantasy adventure RPGs. More contrasts could be drawn between D&D and BW, and it's not as if these two games cover anything like the whole design spectrum for even more-or-less mainstream fantasy RPGs.
But with these comparisons in mind we can already see some ways in which a certain sort of balance in respect of mechanical effectiveness might be more significant in D&D than in BW, and might also be related to narrative balance and to story balance in a different way from how it is in BW.
My view of good RPG design is not that it slavishly does one thing or another, but that it understands the features of the RPG in question - which includes understanding how they differ from other possible designs, and push in favour of one sort of experience and away from other possible experiences - and designs with this in mind.
I think 4e did pretty well by this standard. Which is not to say that anyone has a reason to play it - the fact that a game is well-designed in this sense is only one consideration in favour of playing it, very easily outweighted by a multitude of contrary considerations including that one may not care for the sort of experience that the designers deliberately set out to achieve (eg in the case of 4e, one may not care for the metagame elements that it deploys to simultaneously mechanical, story and narrative balance).
even if you're looking in a very reductionistic manner at game balance, mechanical elements are still too diverse to be characterized independent of context. Is it better, for example, to spend your feat on Skill Focus or Weapon Focus? It depends. It depends on whether you are likely to be able to acquire a good version of the specific weapon. It depends on how many attack rolls you are likely to roll, how likely they are to hit, and on a variety of other factors related to the difficulty of combat. It depends on what skill the Skill focus is for, how often it is likely to be rolled, what the DCs are, how useful the skill is likely to be...and a lot more. Most of which is decided by the DM.
There seem to be a number of assumptions in here about the role of the GM - in deciding the nature of conflicts faced, the mechanical difficulty of them, etc.
But of course all that can be flipped on its head. It is possible to design a game system which gives players control over what sorts of conflicts are faced, and/or over their mechanical difficulty.
In a classic sandbox, for instance, the players choose the sorts of challenges their PCs confront. So the GM does not "mostly decide" which skills are useful, how many combats will be faced, etc.
And in a modern "indie"-style game with non-objective DCs (eg Maelstrom Storytelling, HeroWars/Quest, The Dying Earth, Marvel Heroic RP, and default 4e) the game system itself determines the difficulties confronted in action resolution. As a result, bonuses to skills or to combat in those systems can be meaningly compared for their impact - hence the appeal of uniform resolution systems (The Dying Earth, Maelstrom, MHRP, HW/Q) and the multiple errata to 4e level-approriate DCs and damage, as the designers got a better handle on the mathematical complexities and implications of their comparitively non-uniform PC build and action resolution rules.
And in Burning Wheel, which very much emphasises
player authority over the key sites and elements of conflict, whether it is "better" to develop swordplay or rhetoric for your PC depends primarily (not exclusively) on what sort of campaign you want to play.
to my knowledge, all of 4e's "balance" is based purely on COMBAT effectiveness, which is predicated on the ENCOUNTER as being the primary locus of action resolution, which assumes a DEFAULT LEVEL of combat activity per encounter (assuming we're not running a skill challenge).
4e's balance is intended to extend beyond combat to non-combat resolution also - via the skill challenge (as you note, at least indirectly).
But the PC build and action resolution mechanics are quite different for the non-combat/combat divide.
When it comes to combat, key measures of balance are defences (in my 21st level party the gap in AC is 3, and the biggest gap in defences is 8, between the invoker's and the paladin's Fortitude), hit points (107 for the invoker, 175 for the fighter - and 7 surges for the former vs 15 for the latter), and attack bonus (the best in my group - which doesn't use Expertise feats - is +25 vs NADs for the sorcerer, while the martial PCs are all at +26 vs AC using low-bonus weapons). Damage ouput varies wildly, from as low as 20-ish for the invoker to as high as 100-ish for the sorcerer, but this is a design feature and not as such a mark of imbalance (the invoker has noticeably better control than the sorcerer; and also has many more build resources devoted to non-combat than to combat, whereas the sorcerer is a combat magic machine).
What I'm trying to convey here is that the key defence and attack numbers are very close, and the differences of role manifest most noticeably in surges available and in the effects of hitting in combat, plus in utility features available (eg healing from leaders). This reflects the fairly nitty-gritty nature of D&D combat, which is all about making atacks, taking hits, and moving effectively on a battlefield. 4e's approach to combat balance reflects this.
Turning to non-combat, the numerical disparities are much bigger. Compare the invoker's +37 History check to the +11s and +12s of the other PCs. Or even moving away from the least combat-oriented PC in the group, compare +28 Intimidate (paladin) to +12 (fighter), or +23 Athletics (fighter) to +13 (ranger, and second-best in the party). 4e's non-combat is designed to be balanced within a more abstract approach to action resolution: the GM is expected to frame and adjudicate skill challenges in such a way that all the PCs can contribute, but
not by all doing the same thing. Different skills are meant to be able to be brought to bear, so that each PCs distinctive schtick will be displayed in resolving the challenge. So the very noticeable disparities in skill bonuses shouldn't be compared to the comparitively tight balance of defences and attack numbers, but rather to the big differences in damage output and non-damage effects that PCs generaate in combat.
(This way of balancing non-combat does cause problems in certain sorts of situations, like sneaking through a castle where Stealth is the only skill roll that makes sense within the fiction: the designers invented a special Group Check mechanic, found in DMG2 and Essentials, to cope with this, though it's far from perfect.)
The shift in 4e play from hyper-detailed combat resolution to abstract and much more free-flowing non-combat resolution is one of the distincitve features of the system. (For many I can imagine it being not just a feature but a flaw.)