I'm trying to follow along. Are you saying that none of the natural laws are known, until a PC checks them?
Known to whom? And checked by whom?
Here is an example of the sort of thing [MENTION=27160]Balesir[/MENTION] has in mind:
Player, in the voice of his PC: I wonder if I can internalise all the chaos energy emanating from this dead firedrake so as to turn this jewelled but otherwise mundane horn into a magical Fire Horn?
GM: How are you proposing to do that?
Player: I'll leap atop the body of the dragon, and then cast Cyclonic Vortex to suck in all the chaos energy!
GM: OK, make an Arcana check against DC [whatever is level-appropriate]. Add +2 for expending an encounter power.
Player: I succeed!
GM: OK, as you stand atop the dead dragon, the spiralling winds of your Cyclonic Vortex draw in the ambient energy, and concentrate it about your person. As you focus intently on the horn in your hand, you feel the energy permeating it, transforming it into a Fire Horn. Also, you see some flying creatures approaching you from the distant mountains. Their wings look like bats - but they are so far from you (though closing fast) that if they are bats, they must be giant bats!
Table, in chorus: I guess we just learned two things about the natural and magical laws of this fantasy world: (1) it is possible to use ambient chaos energy to turn mundane items into magical ones; (2) when you do so, you run the risk of drawing the attention of other chaos beings, who can sense your magical doings.
I think Balesir was also suggesting the possibility of being a bit more systematic about it (eg the next time this is attempted, the DC might be lower, or it might work even if the mundane item is not bejewelled) so that we gradually establish, via play, a tighter and clearer set of patterns as to what is and is not possible.
In either case, not only are you handing narrative power over to the players (!), but you're tying that power directly into the skill check of the characters! If the player wants it to be possible to channel firedrake essence into a ring, then that theory works because the character knows a lot about magic. My character knows a lot about magic, therefore he knows that my theory is true.
Two things.
First, you are deploying a particular theory of character building which is not the only one. For instance, when a player builds a PC with a high Arcana score, it may not be that s/he is simply authoring a certain sort of person in the shared fiction. It might also be that s/he is establishing a "play position" from which s/he is going to be able to exert more influence over the direction of the game when magic is involved then s/he would if the Arcana skill were weaker.
Second,
someone has to decide what it is that is known by a character who knows a lot about magic. Why should it not be the player? ("Say yes".) Or the dice? Or some interaction between the two (which is what [MENTION=27160]Balesir[/MENTION] is suggesting)?
I mean, even if you did want players to have some narrative control over the setting, it just seems like a bad idea to tie that into skill checks. Since narrative control is a player resource, you probably want to share that equally, rather than saying that stronger character in-game also has more power across the meta-game.
You may have noticed that one recurring thing that people who like 4e like about it is the relative balance across the various PC build options. So there typically is no clearly stronger character in-game.
There are two design reasons I know of for connecting player authority over the fiction to aspects of PC build. The first I already mentioned in this post: by building a PC with strong Arcana skill I am guaranteeing that when magical affairs are afoot, my PC will be the one who is on top of them - whether that is me declaring novel actions (like the chaos energies example) or me responding to a situation that the GM has put in front of me (eg opening or closing a magical portal).
The second is that it makes the integration between framing, action declaration and resolution, and the fiction, smoother. I'll give an example that involves the Burning Wheel Circles mechanic (this is a "skill" that PCs have in BW, which allows them to make contact with helpful NPCs - the player makes the action declaration, and characterises the sort of NPC s/he wants to meet, based on elements of background that automatically become established as part of the PC generation process in that system).
In my BW game so far there have been two Circles checks in five sessions.
The second was (almost) pure metagame: the PCs were shipwrecked, floating in the ocean on rafts made of bits of the wreckage of their ship, and the player of the elven princess made a Circles check to see if a friendly elven sea captain came sailing by to rescue them. Given the relative improbability of such an event the DC was high, but the character has a very strong Circles ability, and the check was successful.
One part of the check was not metagame, though: the wizard PC turned himself into a falcon, and flew high and far looking for ships. This obviously makes it more likely that any elven sea captain in the area might find the PCs, and so I gave the player of the princess a bonus die on her check. This is the standard mechanic,in BW, for resolving checks made with an advantage, and the fact that Circles is mechanically handled in the same way meant that factoring in the advantage was very easy.
The first check was not pure metagame, though. The PCs were in the town of Hardby, down on their luck and looking for work, and so the wizard (who is a member of a sorcerous cabal) decided to reach out to Jabal, another member of that cabal, to see if Jabal had any work to offer. The check was a failure, and so an offer of work from Jabal was not forthcoming - rather, he sent a thug to run the PCs out of town.
In this case, the Circles mechanic is working basically the same as a Streetwise check, with the one exception being that the player is also free to establish the existence, in the fiction, of the NPC with whom the PC wants to connect. If you separated the metagame from the ingame aspect, then you have to (i) have the player spend a "plot point" to establish that Jabal exists, then (ii) have the player make a Streetwise (or similar) check to connect with Jabal. Which is mechanically clunkier, makes it harder to narrate interesting failures (eg perhaps Jabal doesn't really exist at all, but is just a rumour propagated by someone else to take advantage of unsuspecting junior cabal members), and reduces the sense of continuous engagement with the fiction which arises when the player just declares: "I reach out to Jabal of the cabal - what's the difficulty for the Circles check?"
As opposed to my (much more traditional) method, where that action either is or is not possible
This is a misdescription, which again is the result of a category confusion.
In the gameworld it either is or is not possible -but that is equally so in any of the approaches that I, Balesir and others have described. But that fact of possibility has to be authored by someone. And if it is not to be authored by the player, or by the dice, then it will be authored by the GM. So what you are advocating for, in the real world of game play, is that the player's desired feat of magic either is or is not
permitted by the GM, based on the GM's preferences as to how the game should unfold.
Illusionist GMing is all about obscuring the occurrence, the circumstances and the motivations of these GM decisions. The sort of transparent GMing that several in this thread have connected to 4e is about making these decisions overt. So if the GM says "no", this is made clear - and the GM, presumably, offers some reason for vetoing the attempted move by the player - rather than hidden behind obscure mutterings and concealed dice rolls related to uncertain rules (like eg [MENTION=6696971]Manbearcat[/MENTION]'s wilderness shelter table that he mentioned upthread).
