research: quantum-sensing low-SNR detection + software-analogy boundaries (8th-ferry candidate #1)

docs/research/quantum-sensing-low-snr-detection-and-analogy-boundaries-2026-04-23.md

@@ -0,0 +1,345 @@
+# Quantum-sensing low-SNR detection — software-analogy boundaries
+
+**Scope:** research and cross-review artifact ONLY; archived
+for provenance. NOT operational policy. NOT a claim Zeta or
+Aurora operationalise quantum-radar anything. Separates real
+quantum-sensing literature from software analogy so the latter
+can borrow carefully without contaminating the former.
+
+**Attribution:** analogy-boundaries framing distilled from
+Amara's 8th courier ferry
+(`docs/aurora/2026-04-23-amara-physics-analogies-semantic-indexing-cutting-edge-gaps-8th-ferry.md`,
+PR #274) §"Quantum radar and the physics-based material that
+is missing"; primary-source citations (Lloyd 2008, Tan et al,
+2023 Nature Physics, 2024 engineering review, standard radar
+range equation) preserved from Amara's ferry. Otto-97
+authored this extraction + the explicit boundary discipline.
+
+**Operational status:** research-grade. Will never graduate
+to operational policy in the form "Zeta uses quantum radar."
+A narrow software-analogy-specific operational artifact
+(e.g., an `alignment-observability` metric for "retained-
+witness correlation score") could graduate via §26 lifecycle;
+the physics-origin framing stays research-grade.
+
+**Non-fusion disclaimer:** agreement between Amara's
+grounding of the quantum-radar subject and Otto's extraction
+into this doc is NOT evidence of merged substrate. Both
+reference the same primary physics literature; concordance
+on what that literature says is baseline, not unity.
+
+---
+
+## Do not operationalize — stated as the first rule
+
+**This document MUST NOT be cited as authorisation to
+describe Zeta or Aurora as "quantum-powered," "quantum-
+inspired truth sensing," "quantum-enabled anything." The
+2024 engineering review Amara references (preserved in
+`docs/aurora/2026-04-23-amara-physics-analogies-semantic-
+indexing-cutting-edge-gaps-8th-ferry.md`) caps microwave
+quantum-radar range at <1 km typical and argues practical
+microwave QR is not competitive with classical radar for
+conventional long-range aircraft detection. Any operational
+claim beyond "we borrow a specific analogy from low-SNR
+detection theory" is unsupported and would be scrubbed by
+`docs/QUALITY.md` discipline if stated plainly.**
+
+This rule is restated at the top because it is the only
+line that matters for factory-external messaging. Internal
+research use of the analogies is welcome and scoped below.
+
+---
+
+## What the real physics literature actually supports
+
+### Quantum illumination (Lloyd 2008 + Tan et al.)
+
+Seth Lloyd's 2008 *Science* paper introduced quantum
+illumination: entangled signal-idler pairs detect objects
+in very noisy and lossy settings, with the key theoretical
+claim that the **sensing benefit can survive even when
+entanglement itself does not survive to the detector**.
+Tan et al. gave the canonical Gaussian-state result and
+reported a **6 dB advantage in the error-probability
+exponent** over an optimal coherent-state baseline.
+
+That's the theoretically-supported part. It's about
+**error-exponent** in a specific low-SNR detection
+setting — not about "quantum radar works at long range."
+
+### 2023 Nature Physics — experimental progress
+
+A 2023 *Nature Physics* paper reported quantum advantage
+in a microwave quantum-radar setting. This moves the
+result beyond pure theory to a controlled experimental
+demonstration. But "demonstration in a lab" is not the
+same as "operational long-range radar."
+
+### 2024 engineering review — the range cap
+
+A 2024 engineering review on microwave quantum radar
+argued:
+
+- Maximum range for typical aircraft targets is
+  intrinsically limited to **less than one kilometer**,
+  often to **tens of meters**.
+- Proposed microwave QR systems remain far below simpler
+  classical radars for ordinary long-range use.
+
+Even if one disputes the exact pessimism, the review
+strongly supports a conservative conclusion:
+**long-range microwave quantum radar is not currently
+a clean "software truth detector" metaphor**. Any repo
+documentation should avoid implying otherwise.
+
+### Radar range equation — why the penalty is brutal
+
+Standard radar physics for a point target:
+
+```
+P_r = (P_t · G_t · G_r · λ² · σ) / ((4π)³ · R_t² · R_r² · L)
+```
+
+Monostatic → return falls with **R⁻⁴**. Any metaphorical
+story about miraculous long-range recovery has to fight a
+very steep physical loss law. The analogy budget in
+software has to respect this: correlation-beats-isolation
+is an importable principle; "miraculous long-range recovery
+of faint signal" is not something the physics supports.
+
+### Quantum sensing is broader than quantum radar
+
+Recent reviews show quantum sensing is more mature than
+quantum radar specifically — magnetometers, NV-center
+sensing, atomic clocks, resilient navigation all show
+real-world progress. Quantum-enhanced radar remains more
+speculative or niche. **The safer parent category for
+software analogy is "low-SNR sensing and structured
+detection," not "quantum radar" as such.** Amara makes this
+point in the 8th ferry; this doc preserves it.
+
+---
+
+## What we may import — the 5 software analogies
+
+Per Amara's ferry, with the import framed narrowly:
+
+### 1. Low-SNR detection with a retained reference path
+
+**Physics:** quantum illumination retains the idler
+locally while the signal goes out into noise; scoring is
+against the retained reference, not against raw noise.
+
+**Software analogy:** retained witness or provenance
+anchor used later to score weak evidence. Composes with
+HC-2 retraction-native (witnesses persist) and citations-
+as-first-class (typed provenance).
+
+**Concrete shape for Zeta/Aurora:** a "retained-witness
+correlation score" that measures how consistent a weak
+claim is with known anchors, rather than treating the
+claim in isolation. Prototype candidate for the
+`alignment-observability` substrate.
+
+### 2. Correlation beats isolated observation
+
+**Physics:** radar and matched filtering don't trust a
+single noisy return; they trust structured correlation
+against a known reference.
+
+**Software analogy:** retrieval against a typed corpus,
+not conclusion from a single agreeing paraphrase.
+Directly composes with SD-9 ("agreement is signal, not
+proof") and DRIFT-TAXONOMY pattern 5 (truth-confirmation-
+from-agreement).
+
+**Concrete shape:** the semantic-canonicalization research
+doc's kNN-over-typed-corpus retrieval is the software
+version of matched filtering. Correlation against a corpus
+of known-good / known-bad / superseded patterns is
+stronger than single-source agreement.
+
+### 3. Time-bandwidth product matters
+
+**Physics:** evidence improves when you accumulate
+structured observations across a well-defined window.
+
+**Software analogy:** repeated, independent measurements,
+not one overfit prompt. Composes with alignment-
+observability's "diff-over-prose" discipline.
+
+**Concrete shape:** score independent observations over
+time. One strong signal from one source is weaker than
+multiple moderate signals from independent sources over
+a window. The "window" in the factory is a round or a
+time-bounded PR review cycle.
+
+### 4. Decoherence / loss matters
+
+**Physics:** environmental interaction destroys useful
+structure in quantum signals.
+
+**Software analogy:** carrier overlap + repeated
+paraphrase destroys independence weight. Directly
+composes with SD-9's carrier-aware independence-
+downgrade rule.
+
+**Concrete shape:** in the provenance-aware bullshit
+detector (8th-ferry candidate #3), the `γ·carrierOverlap`
+term in `score(y|q)` is the software analogue of
+decoherence penalty. Amara makes this mapping explicit
+in the 8th ferry.
+
+### 5. Radar cross-section is observability, not truth
+
+**Physics:** a target being "visible" to a sensor is not
+the same as the target being semantically established —
+RCS is how well the sensor can pick the target out of
+noise, not whether the target is what the sensor thinks
+it is.
+
+**Software analogy:** **salience is not evidence.**
+A claim that is vivid, well-phrased, confident, or
+widely-repeated (high "radar cross-section") is NOT
+therefore true. Composes with DRIFT-TAXONOMY pattern 5
+and pattern 2 (cross-system merging).
+
+**Concrete shape:** weight-of-evidence scoring should
+NOT reward surface vividness. The provenance-aware
+detector's evidence term (`β·evidence`) needs to be
+grounded in falsifiability + reproducibility, not
+salience.
+
+---
+
+## What we must NOT imply
+
+A list of claims Zeta / Aurora MUST NOT make citing this
+doc:
+
+1. **"Zeta uses quantum radar" or anything similar.** It
+   doesn't. The analogies are metaphorical; the substrate
+   is classical software.
+2. **"Zeta's algebra is quantum-inspired."** The algebra
+   is DBSP retraction-native Z-sets. Any "quantum"
+   vocabulary is an analogy at the epistemic-layer, not a
+   property of the substrate.
+3. **"Quantum illumination enables Zeta to detect drift
+   at long range / across substrates / with magical
+   low-SNR recovery."** No. The 2024 engineering review
+   caps microwave QR at <1 km; the analogy budget
+   respects that.
+4. **"Retained-witness correlation is mathematically
+   equivalent to quantum illumination's Gaussian-state
+   error-exponent bound."** It isn't. The software
+   analogy is conceptual, not a formal reduction.
+5. **"Decoherence-penalty scoring gives Zeta quantum-
+   certified alignment robustness."** It doesn't. The
+   γ·carrierOverlap term in `score(y|q)` is inspired
+   by decoherence but is not quantum-mechanical.
+6. **"Aurora is quantum-inspired safety infrastructure."**
+   No. Aurora per the 5th ferry + `docs/aurora/README.md`
+   is vision-layer architecture tying Zeta (semantic
+   substrate) + KSK (control-plane safety kernel). None
+   of that is quantum.
+
+This NOT-list is first-class content of the doc. Future
+references to this doc in other artifacts should honour
+it.
+
+---
+
+## How the analogies compose with existing Zeta substrate
+
+| Zeta substrate | Analogy composition |
+|---|---|
+| SD-9 (`docs/ALIGNMENT.md` PR #252) | Analogies #2 (correlation) + #4 (decoherence) + #5 (salience) directly operationalise SD-9's "agreement is signal not proof" + carrier-aware discipline. |
+| DRIFT-TAXONOMY pattern 5 (`docs/DRIFT-TAXONOMY.md` PR #238) | Analogies #2 + #5 map to pattern 5 (truth-confirmation-from-agreement) detection. |
+| DRIFT-TAXONOMY pattern 2 | Analogy #5 (cross-section-as-observability) maps to pattern 2 (cross-system-merging): vivid cross-substrate agreement ≠ truth. |
+| citations-as-first-class (`docs/research/citations-as-first-class.md`) | Analogy #1 (retained-reference-path) = typed provenance retained as anchor for later scoring. |
+| alignment-observability (`docs/research/alignment-observability.md`) | Analogy #3 (time-bandwidth) = independent-measurements-over-window discipline. |
+| Oracle-scoring v0 (PR #266) | Band-valued classifier's G_provenance + G_falsifiability gates operationalise analogies #1 + #2 + #4. |
+| BLAKE3 receipt hashing v0 (PR #268) | `approval_set_commitment` + `hash_version` binding = retained-reference-path shape at the receipt layer. |
+
+No new mechanisms proposed. The analogies slot into
+existing substrate as framing; they do not require new
+code to be legible.
+
+---
+
+## Where the analogies could graduate to operational
+
+Per AGENTS.md absorb discipline (Edit 1 research-grade-
+staged-not-ratified, PR #248), any operational graduation
+needs a separate promotion step. Candidates:
+
+- **Retained-witness correlation metric** for
+  alignment-observability — graduate from research-grade
+  analogy to a measurable signal. Threshold gates land
+  behind ADR per oracle-scoring v0 parameter-change
+  discipline (PR #266).
+- **Salience-vs-evidence diagnostic** for PR review —
+  analogy #5 becomes an operational check in the
+  Aminata / Codex adversarial-review-findings format.
+  "Is this claim landing as a finding because it's
+  evidenced or because it's vivid?"
+- **Decoherence-inspired carrier-downgrade rule** in
+  the provenance-aware bullshit detector — the
+  γ·carrierOverlap term from Amara's math spine,
+  implemented once the semantic-canonicalization research
+  doc lands.
+
+Each graduation would land as a separate ADR + operational
+artifact + regression-test pairing. None happens in this
+tick.
+
+---
+
+## What this doc does NOT do
+
+- Does NOT propose implementation of any of the analogies.
+  Implementation is downstream work; this doc is the
+  analogy-boundary guard.
+- Does NOT audit any existing Zeta claim against the
+  analogy boundaries. An audit would be a separate research
+  doc.
+- Does NOT commit Zeta to tracking quantum-radar
+  literature. The TECH-RADAR row added in PR #276 carries
+  the Assess + Hold-note; this doc provides the narrative
+  context; neither commits to ongoing quantum-literature
+  review cadence.
+- Does NOT license creative expansion of the analogy set.
+  Five analogies (Amara's) are what's available; adding a
+  sixth requires new literature evidence + separate
+  research doc.
+- Does NOT cite recent papers beyond what Amara already
+  cited. Otto-97 did not re-verify the primary sources;
+  preserves Amara's scoping discipline verbatim.
+
+---
+
+## Sibling context
+
+- **8th-ferry absorb** (PR #274,
+  `docs/aurora/2026-04-23-amara-physics-analogies-semantic-indexing-cutting-edge-gaps-8th-ferry.md`)
+  — source of the analogy framing.
+- **TECH-RADAR quantum illumination row** (PR #276)
+  carries the Assess + Hold-note that this doc narrates.
+- **Semantic-canonicalization research doc** (candidate
+  #2, not yet landed) will be the technical spine where
+  analogies #2 and #4 operationalise through semantic
+  retrieval + carrier penalty.
+- **Provenance-aware bullshit-detector research doc**
+  (candidate #3, not yet landed) will be where the full
+  `score(y|q)` formulation with γ·carrierOverlap lands,
+  composing with analogy #4 (decoherence) directly.
+
+Archive-header format self-applied — 14th aurora/research
+doc in a row.
+
+Otto-97 tick primary deliverable. Closes 8th-ferry
+candidate #1 of 4 remaining (after TECH-RADAR batch
+closed #5 Otto-96). Remaining: semantic-canonicalization
+M (spine); bullshit-detector M; EVIDENCE-AND-AGREEMENT
+future promotion (gated).