If skill challenges are simulations, how can they be shared narrations, while combat simulations are not?
For what it's worth, 4e combat resolution also has narrativist dimensions (eg the Come and Get It fighter power, which allows a 1x/enc pull of foes within a certain radius towards the PC, and which therefore in effect empowers the player to 1x/enc specify how, in the gameworld, it comes about that those foes move closer to his/her PC).
The DM has to base the existence of said tower or chandelier on his knowledge of what the world may hold or use whatever system is in place to determine such.
This is equally true in narrativist play, except that the logical relation will typically be one of consistency rather than entailment (which in any event is, in practice, unlikely to be made out given the paucity of detail about the gameworld). And the system in place to determine which of the various possibilities obtains (each consistent with the prior state of the gameworld, but all mutually inconsistent as extensions of that state) is one of game-mechanically-distributed stipulation.
If you want to include said determination into a Skill Challenge dice roll, you've changed role-players into role-players plus world creators.
Yes, for a certain value of "role-playing". Of course, in my view "roleplaying", as used to describe the activity of playing an RPG, includes the act of stipulating the state of the gameworld when this takes place during the course of play.
This doesn't work for most folks as the challenge stops being "beating your opponents as your character" into "beating your opponents by halfway wishing the world into existence to win."
<snip>
This is the difference between people jumping up from the table shouting, "WE KICKED YOUR BUTT!!" to "That was a good story we made."
<snip>
It is my assertion success is as important to role-players as it is to wargamers and cardsharks.
It's interesting that you seem to agree with Ron Edwards about the inconsistency of winning and storytelling. I think the tension between the two is not as great as you (and Edwards) are suggesting. For example, provided that the mechanics place certain constraints on narrative distribution, then there can still be a challenge in taking the story to where you want it to go (eg victory for one's PC). And overcoming that challenge might still be fun.
I also think that you are wrong in suggesting that roleplaying (in your sense) and narration are inconsistent speech acts. In many cases I think they are performed simultaneously. I'm reminded of Davidson's essay on Quotation (1979), in which he says (pp 80-81 in Inquiries into Truth and Interpretation):
"I once resolved to adopt a consistent way of using quotation marks in my professional writing. My plan was to use single quotation marks when I wanted to refer to the expression a token of which was within, but double quotation marks when I wanted to use the expression in its usual meaning while at the same time indicating that the word was odd or special ('scare quotes'). I blush to admit that I struggled with this absurd and unworkable formula for a couple of years before it dawned on me that the second category contained the seeds of its own destruction. Consider . . . [my earlier remark that] Quine says that quotation '. . . has a certain anomolous feature'. Are the quoted words used or mentioned? Obviously mentioned, since the words are Quine's own, and I want to mark the fact. But eqaully obvious is the fact that the words are used".
As Davidson notes, it is possible to both use a word and to mention it at the same time. Likewise, I think it is possible to both tell a story (ie occupy the authorial "god" role) and to play a role (ie occupy the protagonist role) at the same time - an example would be a player who says "I take a drink from my water-bottle, which of course I refilled before we left town." Here the players is both playing the role of his/her PC and occupying the authorial role.
This is why I think that Edwards is correct that no particular role (protagonist or authorial) is indicative of whether play is simulationist or narrativist. To work that out you need to look at what sorts of expectations and constraints determine what is done by any given player occupying any given role.