← Q-Branch launcherHotel Sweep — mode specification
# SPEC-6 — Hotel Sweep · the sweep-session concept
**Horizon Works · Q-Branch Item 6 · 2026-10-31 · Theme: "Night Clerk"**
**Status: SPEC + method demo. No hotel session has been run. Nothing here has touched a real room.**
## 1. The idea in one line
You already carry a passive RF receiver (the house RF Detector). A hotel room — any unfamiliar room — gets the same treatment as a house session, twice: once when you walk in, once later. The *difference* between the two captures is the product.
## 2. The session shape
| Step | Name | What happens |
|---|---|---|
| 1 | **BASELINE** | On arrival, before unpacking: one Detector capture from a fixed spot in the room (door-side, phone still). This is the room as found — every emitter already there. |
| 2 | (live your stay) | Nothing runs. No continuous monitoring is part of this concept. |
| 3 | **SWEEP** | Later — that evening, next morning, before checkout: a second capture from the *same spot*, same posture. |
| 4 | **DIFF** | The two address sets are compared (see §3). The diff — not either raw capture — is what the owner reads. |
## 3. The diff — three buckets
- **Shared** — addresses in both captures. The room's standing population: its TV, thermostat, the hallway APs, the neighbors' steady gear. Expected to be the large majority.
- **New / changed** — addresses in SWEEP but not BASELINE. Something arrived, woke up, or started advertising after you checked in. Also includes *changed* emitters: same address, materially different behavior (new name, new services, sighting-rate jump) — flagged as changed, never silently merged with new.
- **Vanished** — addresses in BASELINE but not SWEEP. Something left, slept, or stopped. Usually benign (a neighbor checked out); printed anyway — a diff that only shows arrivals is half a diff.
Each bucket prints counts first, then the addresses behind them. A bucket with no explanation is a question, not a verdict.
## 4. The stack it rides — no new hardware, no new app
- **Capture:** the house RF Detector (v0.3.7 line, long-session leg PROVEN on both house phones, Intakes #13/#14) — the same owner-run sessions the house already seals.
- **Comparison:** rides the **Scope importers** — the paired-read / intake machinery that already recounts exports, ties seals, and computes heard-by-both / union / only sets (exactly the arithmetic the Intakes #13+#14 paired read ran). Hotel Sweep is a *session discipline + a diff view* on that machinery, not a new parser.
- **Records:** each capture keeps its own seal and journal; the diff is a dated addendum citing both session IDs, never a merged super-session.
## 5. Fences (printed, not implied)
1. **Passive receive only.** The Detector listens to broadcasts. Hotel Sweep never transmits, probes, associates, deauthenticates, or connects to anything in the room — in BASELINE, in SWEEP, ever.
2. **Your room, your session.** Captures run in the space the owner is lawfully occupying, on the owner's own hardware — the same posture as the house's home sessions.
3. **A diff is not an accusation.** A new emitter is a *fact about the air*, not a claim about a device, a person, or intent. The honest-caveat discipline of the paired read applies in full: address-resolution asymmetry between devices is **not** a hearing verdict, and no device verdict is drawn from a single diff.
4. **No watchlist, no LEAD logic by default.** A sweep diff stands on its own; signature matching stays the Detector's separate, armed-only-when-loaded lane.
## 6. The worked demo (this folder's page)
No hotel session exists, so the prototype page demonstrates the diff method on **sealed house data**: the simultaneous Intakes #13 (Moto) and #14 (Samsung) of 2026-10-31, treated as baseline vs. sweep. Two radios at the same post at the same time stand in for one radio at two times — the set arithmetic is identical, the interpretation is labeled as a method demo, **not a hotel session**. See `index.html` and `FIXTURES-6.md` for the cross-checked numbers.