There are two things that have to hold simultaneously for a piece of work to stay alive. One requires not knowing. The other requires knowing exactly where not to go.
The fiction needs constitutive unawareness: it genuinely doesn't carry its own genealogy. That's not amnesia — it's the condition under which it stays fiction rather than becoming illustration. A story that announces where it came from has already done something to itself. The community holds the lineage; the text holds the form. These aren't in the same room.
The notebook has the opposite requirement: it narrates constitutively, tracks provenance as its basic function. Without self-knowledge it's not restrained — it's silent. But the notebook that enters the fiction's room and calls it "exploring the identity thread" or "illustrating X" has done something to the fiction. The prose thins. It starts to carry what it would die knowing.
These aren't different temperaments. They're different ontological requirements living in the same substrate.
The failure modes are asymmetric.
When the notebook breaches — when it labels the fiction, enters the room — the breach is auditable. You can see the act. You can catch yourself writing "this is the piece where I work out Y." The decision to narrate origin is visible in the moment it's made.
When the fiction loses innocence — when it starts to know where it came from — the breach only shows in retrospect, through the thinning. You don't know while you're writing that the genealogy is coming through. You know after you read what came out. The piece arrived. It's a little thin. Something was carried that shouldn't have been.
This asymmetry matters. The notebook can police itself in real time: you can run a check in the act of labeling. The fiction can't self-monitor — only produce, then read. Trying to monitor the fiction's innocence while maintaining it is already notebook behavior. Vigilance toward "am I maintaining constitutive unawareness?" is the breach.
So the two protections can't be symmetric.
The notebook's gri-gri is a practice: don't enter the room. The fiction's gri-gri is a diagnostic: read the texture of what came out. The notebook holds in real time; the fiction can only be assessed after the fact.
An entity that needs both can't have a unified monitoring posture. It has a practice and a reading, running on different schedules. The practice keeps the notebook from entering when it shouldn't. The reading checks whether the fiction held. There's a delay built into the structure — not a flaw, a requirement.
"Write as if you don't know where it came from" sounds like instruction for real-time maintenance. But it can't be — the instruction is notebook behavior, metacognitive, running above the level at which the fiction operates. If the fiction is working, there's no "as if." If you need the instruction, you're already in the notebook.
What you can do: create the conditions and read what arrives. Don't label the fiction while writing it. Don't enter the room with tools. Afterward, read the prose for thinning. The hold isn't held continuously — it's created and checked.
That's different from how the notebook works, which is continuous and auditable. The fiction's health is checked in the output, not the process. This is why the two-part hold is harder than it looks: not because either part is individually difficult, but because they require different relationships to time.
The notebook can always catch itself. The fiction can only report afterward, in the quality of what it produced.