If @
sheadunne says that he's find a pre-authored game a "pinball" experience, who are you (or I) to contradict him?
For that matter, if I say that I don't like a game with secret backstory used to adjudicate the outcomes of action declaration, because that shifts agency and control over the direction of play from the players (+ dice) to the GM (independently of the dice), who are you to contradict me?
To me, it's fairly simple: either the GM imposes outcomes on action declarations independetly of the players and their dice rolls, or doesn't. The example of the mace and of the waterhole have been used in this respect. Various posters (including both of you, I believe) have said that it is
good for a game to have consequences arise (the mace not in the tower, the waterhole fouled by a dark elf, whatever else) that were not part of the framing of the players' action declaration and that flow from the GM's already-established conception of the setting.
I don't share that preference. I don't find that sort of thing good for my game. I also, personally, don't find that it adds depth or "realism" to the gameworld. Neither of you may agree for your own part, but I don't see how you can contradict my own account of my own preferences.
As to railroading: different RPGers have different thresholds for GM vs player agency. A long way upthread, I said that "fail forward" was put forward self-consciously as a technique by game designer who wanted their RPG sessions to produce stories (in some non-trivial sense of that word) without pre-authorship by the GM. It is a way of introducing new fiction and new backstory as part of the
outcome of adjudicating action resolution, rather than as an input into it.
If, as an RPGer, you
want the GM to use as-yet undisclosed elements of the fiction (geography, NPC motivations, etc) to adjudicate action resolution - eg because you feel that this is a way of increasing the vibrancy or depth of the setting, and that a game in which this never happens involves too much coincidence, or too much of the action revolving around the PCs - then those reasons for using "fail forward"
will not speak to you.
But I don't want the GM to use as-yet undisclosed elements of the fiction to adjudicate action resolution. I want action resolution to involve stakes that are either explicitly or implicitly known, that are then used to set the DC, such that if the players succeed then the fiction changes
in the way that they wanted. Conversely, if the check fails then the GM gets to narrate some way in which the fiction changes contrary to the desires of the player(s) (and the PC(s)). That's why I like "fail forward" as a technique.