A rogue frequency has been bleeding into the sleep-mode cycle of consumer neural interfaces across multiple manufacturers since approximately June 2222. The signal does not appear on electromagnetic spectrum scans. It does not register on any monitoring equipment. It exists only in the gap between waking and sleeping neural-interface states — a transitional processing mode that interface designers call "the threshold." During this window, approximately 0.3 seconds in duration, the interface is neither actively mediating consciousness nor fully dormant. Something is using that window. Something is broadcasting into the dreams of everyone wearing a compatible device.

The content is identical across all affected users, which is itself impossible. Neural interfaces interpret external stimuli through the user's own cognitive framework — two people receiving the same signal should experience different imagery, filtered through individual memory and association. They don't. Across 1,247 documented cases spanning fourteen interface brands, seven firmware versions, and users ranging in age from nineteen to eighty-three, the dream is the same. A staircase. Descending. The walls are covered in text — not printed, not projected, but growing. The letters are alive. They shift and breathe and rearrange themselves as the dreamer passes, as if the words are aware of being read. The language is not any language. Users who speak different languages all report being able to read it. They cannot remember what it says after waking. They remember only that it was important and that they agreed with it.

Three researchers from the GLMZ Neurological Institute published papers analyzing the broadcast dreams in August 2222. The papers were thorough, well-sourced, and peer-reviewed. Within two weeks, all three were withdrawn. The researchers cited "methodological concerns," but the methodology had already been validated. Two of the researchers left the field entirely — one to commercial fishing, one to a contemplative community in the Upper Peninsula that does not use electronic communication. The third, Dr. Ileana Voss, continues to publish, but her colleagues say the work is unrecognizable. Her writing style has changed. Her research interests have shifted from neuroscience to architectural theory. She writes about staircases with an intensity that reads as devotional.

The staircase has been mapped. Cross-referencing dream reports from affected users, a collaborative research effort organized through anonymized channels has produced a composite model of the structure. It descends forty-seven floors. The architecture is consistent across all reports: each floor is slightly different from the last — the material of the steps changes, the width of the stairwell narrows, the living text on the walls becomes denser, more urgent. The temperature drops. The sound changes from silence to something that users describe as "the opposite of an echo — sound going in instead of coming out." At the bottom of the forty-seventh floor is a door. Every dreamer reports the same door. None of them have opened it. None of them have tried. When asked why, they give variations of the same answer: it isn't time yet.

The frequency continues to broadcast. New cases emerge weekly. There is no way to block the signal because there is no signal to block — it exists in a space that instrumentation cannot reach. The GLMZ Anomaly Documentation Project has classified the broadcast dreams as a Category 2 Cognitive Anomaly: a phenomenon that affects subjective experience without measurable physical mechanism. The classification is accurate and completely useless. Forty-seven floors down, a door waits. It is not locked. It has never been locked. But every dreamer in GLMZ knows, with a certainty that transcends the dream, that the door will open only once. And that what comes through it will not be going back.