Fiat
Player: "I cast Color Spray. These guys in the area need to make a saving throw, DC 15."
vs.
No Fiat
Player: "I want to make a big cone of dazzly light to stun or dazzle these guys. What sort of roll do I make?"
It's an interesting example you've posited. My sense of things is that the further back you go through the editions, the more spells look like the second option. I don't think they were codified as they have been with the intent of granting wizard players additional control over the game, however. I think spell effects were rendered increasingly specific and objective and arbitrary in order to balance character abilities, speed up the game, and remove some of the DM's influence. I look at your "player fiat" as an unintended (if natural) consequence of this path of development.
Thing is, I think the magic system personified by the first option is bad game design. If anything, I think the second should be the goal.
I mean, if you really want to blind someone with magic, there should be any number of ways to do it, right? If you shoot a Melf's Acid Arrrow right in someone's eye, or a Sunburst that goes right by their face, that's going to diminish their visual acuity. But more and more, we've accepted that blinding can only happen when the spell text specifically says it can. Glitterdust blinds you, because it says it does. Lightning bolt doesn't. Lightning bolts don't bounce off of walls any more either.
This is a bad thing. (In my opinion). We buy into this unnatural specificity because "it's magic". In many cases, really "it's magic" is a blanket excuse for bad game design.
The thing is, a fighter doesn't even have that excuse. If you tell me that I can only shoot flames twice a day before I run out of magic, or I can cast dominate person, but it only works on humanoids, or that I can blind someone with Glitterdust but not with another spell that creates bright light or could physically damage someone's eyes, I guess I'll buy into that because it's magic. If, however, you tell me I can only swing my sword a certain way twice a day before I run out of sword swings, or that I can only get an extra attack from Cleave only when a enemy is felled (which really has nothing to do with how quickly I attack), or most pointedly, that I can only blind someone if I've told the DM I'm using Blinding Strike (which I may even have to have selected in advance as a character option), that's not going to fly. Why, I ask as a player, is it impossible for someone to poke someone in the eye without the player specifically intending that effect? Why is it so easy to achieve when I select a certain ability? That's not how combat works at all; it's far more improvisational, and its outcomes are far more open-ended.
Frankly, I think our standards for quality in how nonmagical abilities are written are higher than for magical ones in a variety of ways. The nonmagical abilities have to be written more clearly and comprehensively, balanced better, and hew more closely to some sense of reality (even if it's the reality of a fantasy world). 3e's skill system, for example, is much better than its magic system, in every meaningful way. It's simpler in execution, but deeper and broader in scope. It's more open-ended, but more balanced and fair. It's more realistic, but also more dramatic.
That's why, as I've said earlier, I think on a broad conceptual level, it's much more natural to try and make the magical abilities more like everything else than to try to add quasi-magical abilities to the fighter. Or, to your point, it makes more sense to take away "player fiat" from a class that has an excess of it, rather than add fiat to a class that shouldn't have any more of it than it already does.
(And really - it is a crystal clear example of player fiat, particularly the pre-errata version.) I just want to be clear - while I, personally, think CaGI is pretty great, I understand our focus here should be to talk about Next, and not take another go around the edition wars. I know that CaGI (and its lesser-known cousins like the Rogue's Compel the Craven) are on the far end of 4e's more cinematic powers, and if they must go for a compromise, then go they must.
