There's something I don't quite get about the narrative games I've GM'd. Let's see if I can put it into words.
First of all, games that I've played: through my teens D&D 3, with the usual mixture of GM-led story and princess play. Then in more recent years some story games: Intrepid, Lady Blackbird, Witch: JTL, Dungeon World, plus a couple of others. And I've also delved into OSR via the school of Eero.
The question I have is about the GM's role in escalating conflict. In OSR play, I understand how this works: my concern is to treat fictional positioning, at all levels, deadly seriously. If it has been established there are sleeping goblins around the corner, and the party's struggle with a carrion crawler makes too much noise ("too much" perhaps being established via a listen check), then the goblins will wake up and get involved. When things are less well established (for example, a more improvisational situation, with less dungeon prep) I make similar calls: Well, I know this is a lawless badlands area, and it fits with my aesthetic that these would have roaming bandits. So when the players light a fire on a hill at night, I make a judgement that it's likely they will be spotted - maybe I roll a 1 in 3 - and bandits show up.
But because my concern is fictional consequences, and a living breathing world, and this gamestyle is very much "Story Later" (as opposed to Story Now), sometimes nothing happens. Sometimes it's just dull. The players may quest through an easy dungeon, and my prep says that's all the challenge there is. Or they may be overwhelmed, and get the party wiped out.
Now what about narrative games with a strong GM role where player input is (formally, at least) limited to saying who their character is and what they do? For example, Apocalypse World, Dungeon World, or Lady Blackbird (at least at one end of the axis of ways to play Lady Blackbird). And let's consider Dungeon World, since it's a similar genre of fiction.
In Dungeon World, it won't be a simulationist concern that brings the sleeping goblins. It's more likely to be a "Show signs of approaching threat" move in response to a player missing a move. I probably haven't prepped the dungeon, so when I decide to make that move, and when I decide to bring goblins around the corner it's because my sense of story and drama, and my desire to fill the player's lives with adventure, suggests that's the best way. (Apparently goblins was the best thing I could think of!). Equally, if I decide not to escalate, so there's nothing between the players and their objective, that's for narrative concerns - so we don't get the "dull" problem of OSR play.
So here's the crux of it: If I keep bringing in more threats, and snowballing things, based on my sense of story (and it doesn't have to be just monsters, it can be the layout of the dungeon, how much further there is to go, how many obstacles are in their way) then aren't I back to illusionism? The players' moment by moment choices affect how well they deal with the current threat, but if I respond to make sure there's always a "satisfying" level of adventure and derring do, to make sure the characters' lives are filled with adventure, then ultimately their choices aren't having an effect on the story beyond colour.
What am I missing?
P.S. Because text on the Internet is a bad conveyor of emotions let me state clearly: I love the games I've listed. I'm not wanting anyone to "defend" these games. I'm just trying to get them better.