Celebrim
Legend
Which is why narrative control lies at both ends of the spectrum. You can give players all the rules you want to do whatever they want, but if the guide is the only narrator, the players have agency only to the extent that it's given by the narrator.
I believe you misunderstood my point. I'm suggesting that in games where players can engage in authorship, they can have less agency than in games where they can't.
Consider my case of a railroaded traditional RPG where the players have only limited tactical choices and can't actually shape the overall story. We both agree this represents low player agency. The game is on rails.
Consider a hypothetical game with typical Nar mechanics. It will have rules for allowing players to engage in authorship, but only in a finite way. However, the storyteller - or 'guide' - still has the unlimited authorship of a world builder and secret keeper. Additionally, Nar games often flat out encourage the guide to engage in traditional railroading techniques such as Schrodinger's Map or Schrodinger's Stat Block.
Now suppose we are engaged in some sort of story arc and have reached the climatic encounter (or a climatic encounter) with a villain or foil. Because the guide has unlimited authorship and is flat out encouraged to "do what is best for the story" or "do what is fun", the guide can decide that since this is a climatic encounter then it ought to be a tense and exciting combat. And as such, he can using his authoring authority adjust the encounter on the fly so that - for the good of the game - the villain does not go down like a chump resulting in an "unfun" anti-climatic end of the story arc. Likewise, if the villain seems to be getting the upper hand, then he can - for the good of the game - adjust the combat on the fly so that just as it seems all is lost, one good die throw turns the tide at "the last moment'.
I put to you that this game is identical in terms of agency with the traditionally railroaded traditional RPG, despite the ability of the players to engage in authoring during the game. The reason is that ultimately, all that authoring is going to amount to what are essentially tactical choices of small import, while the actual results are beyond their ability to actually effect.
I came to this conclusion after watching game on youtube with mechanics that allowed player authorship. The game was more on rails than a game I run with traditional preparation and mechanics. Indeed, because the game actually encouraged much stronger authorship by the guide than is normally validated by a traditional RPG, the actual player agency relative to the GM was less than a traditional RPG despite sharing the authoring role with the players.
In Numenera, players roll all the dice, they use XP to change the GM's mind, and they use abilities to determine/adjust how difficult a task is. The book encourages the GM to let players describe their actions, and gives many examples of negotiation between player and GM of certain outcomes. I'd put its Player Agency rating at about 40%, as in, the GM still has most of the control, but the players have a lot of influence in what happens. I don't know how the rules go, but if you've seen Titansgrave (the web series), you'd know that a game can be played with more player agency than Numenera. The players in the series do a lot of GM-level narrating, so I'd put that at a good 50%, if not more.
I'll have to watch that but don't be too surprised if I don't have the same analysis that you do.