This may be true as far as it goes, but doesn't really address Starfox's point. Because the question then becomes, How is that situation decided upon? If the GM decides on the situation regardless of the players deciding to play the GDS trio - and would have used exactly the same situation if the players had build a Sorcerer Supreme, a Goliath Brawler and a Valkyrie - then is the GM still running a game? Or just authoring his/her own story? The answer to this seems to me to depend crucially on the preferences and expectations of the players. But there is no default presumption that what this GM is doing is the proper way to run an RPG.
In my case the starting situation would remain exactly the same. The unfolding narrative is likely to be highly different depending on the areas the PCs engage, their choices, and the relationships explored. I do not author the game. I present the starting context and let the players choices guide the narrative.
I have experience that the player choices strongly affect the way the narrative and situation unfolds. Several times, I've been asked to run a 2nd or 3rd simultaneous campaign with a different group. In order to cut down on my workload, I start the new group in the original situation with the same starting parameters. Typically, by the third session, the circumstances have changed enough that I no longer have a reduced workload. The situations evolve away from each other quite quickly from the differences in PC choice and success.
Another great post! Key to Starfox's hypothetical scenario is that the GM has persisted with his/her own conception of the gameworld regardless of the fact that for whatever reason, the players are unmoved by the bits the GM is very excited by. If the players end up being upset by this, when the GM brings to an end the game that the players have been enjoying, the GM should hardly be surprised!
Whereas I would never run a game like that - I constantly rework the backstory to my campaign world in light of the unfolding interests of the players revealed in the course of play, in order to better support situations that build on those interests and allow the players to keep driving the game forward in the way they want to.
I constantly rework the forestory and the current situation by adding previously undefined elements. Once an element is in play, its nature remains constant. As an example, from a much earlier discussion; under my scheme if a PCs long-lost mother is known to be in an orcish camp for sacrifice and the PCs leave without interrupting the ritual then the mother is sacrificed. The backstory is not modified because
I think the player may want to further engage the NPC in a thematic way. I let the players choose what interests them from the milieu. Perhaps the players wants the angst of knowing he failed to save her? Perhaps the player originally had her long-lost because he doesn't want to interact with the mother at all? Allowing the players to author the story by respecting their choices -- both good and bad -- is a form of empowerment.
As a player my interests are ecletic and best known by me not the DM. I'll engage those things that catch my fancy and drop them again just as quickly if they don't keep my interest. Having someone rebuild to feed my apparent interest would be disconcerting and likely frustrating for the other person.
As Starfox says, a GM must be prepared to take responsibility for his/her own world design decisions. If you know that your players want to explore a world in which you already have a default timeline worked out, and in which your choices for what will be significant in the setting (like the end-of-the-world plot) are settled prior to play, then go for it - take responsibility! But a lot of players don't want to play that way, don't want to explore the GM's world but rather want to play their PCs in the setting as they conceive of it (and as their conception of it evolves over the course of play) - and there is nothing defective about their preferences in this respect.
Of course, I take responsibility for my designs and game play. I am upfront with the prospective players about my style, the starting situation, and base expectation for the game. I despise bait-and-switch gaems where the character is designed under one premise and rapidly thrust into a different one without player choice because it is deceptive on the part of the DM regarding design and game play.
I insist the players take responsibility for their designs and game play too. The unfolding narrative is constructed out their actions and consequences. The PCs do not instigate all actions; they may not appreciate all developments; but the PCs can affect any outcome. If a player doesn't want this form of play, as a DM I hope he takes the responsible route and finds a game more to his taste.
Dragonlance is a Railroad because as you point out it does not support player freedom in any significant dimension, the players are essentially just along for the ride. Where there is significant player freedom in any of the dimensions you have identified, it is not a railroad.
