As you turn back from the door, you see a spectral figure in white staring at you from the other chair.

Surprised and fearful, you stumble backwards towards the fireplace.

Amazed but curious, you sit down in your chair and ask the figure "Who are you? What do you want?" 

This is one of four branches on the second split of the story tree.

This looks like a good start. But I'm not so sure what I should say about it because I know how it ends.

Writing a hypertext narrative, especially a branching one, you have to keep track of what is happening all over the graph. A linear story can be written in the spirit of "let's see what happens." A single branch of the story can be written that way, but as soon as one branch is growing it puts pressure on what can happen on the other branches.

I can sense that pressure on the writer to keep it all connected and in view. The reader is in the opposite position. Each is constructing a narrative, but the writer needs a wider view.

Are you the writer's voice?

No, are you?

The implied author of this piece is not necessarily the same as the writer.

Is he hiding?

No, and there is always a meta-story to be told about the writer finding a voice for any piece of writing. That meta-story is implicit, and often quite standardized, as in most academic writing.

That's why we're doing it differently. If we're in dialogue, which of us is Socrates?

I don't know.