# SuperStack Final-Polish Report

**Candidate:** JPL_AUTONOMY_EDL_06
**Document:** `dissertation.md` (Landing-Ellipse Contraction as a Technology Learning Curve) + `deck.md`
**Mode:** ACADEMIC. **Branding:** NEUTRAL (no MITRE markings, no vendor-AI names).
**Date:** 2026-06-16
**Method:** Targeted edits in place, section by section. No from-scratch rewrite. Full length and all 130 references preserved.


## Passes run (canonical stack order)

1. **SPACE-SCIENTIFIC-WRITING** (domain rigor). Verified domain fluency and source traceability across all eight chapters. Every load-bearing numerical claim is either drawn from a cited mission record or explicitly labeled illustrative/design-stage. No citation points to a fabricated source; all 130 references resolve to real DOIs or NTRS citation URLs. No changes required; the body already met the bar.
2. **HAAF LOGIC BAND** (mandatory tier; HAAF wraps Toulmin at Layer 4). Verified the nine-layer flow holds: answer-first executive thesis stated as H0/H1; problem frame (current state, desired state, gap, consequence) present per chapter; major claims carry full Toulmin (Claim x8, Grounds x7, Warrant x8, Backing x8, Qualifier x7, Rebuttal x8, including symmetric H0/H1 treatment in Ch7); evidence interpreted as claim-evidence-reasoning; every causal claim names a mechanism (driver to mechanism to effect to consequence to implication); confidence stated with what-would-move-it, calibrated to the DESIGN-STAGE grade; assurance-case summary (top goal, five subclaims, assumptions, context, residual risk) present and consistent. Layer 9 (architecture traceability) correctly OUT OF SCOPE, consistent with `research/expansion_plan.md` section (e). No HAAF layer was missing or broken; no fabrication needed to fill any element.
3. **SUPER-HUMANIZER** (anti-slop, authentic voice). Scanned for AI tells and scaffolding. Zero occurrences of: "it is important to note", "in conclusion", "delve", "tapestry", "landscape", "realm", "moreover". Abstractions are named; sentence-length burstiness is present. No changes required.
4. **MITRE-HUMANIZER** (discipline only; neutral work). Em dash count enforced to ZERO (see table). Banned-phrase purge: "robust" (vague-adjective sense), "cutting-edge", "leveraging", "seamless", "navigate the complexities", "testament to", "at the end of the day" all absent. Tightened three soft intensifiers (see "Edits made"). The zero-negative-words rule was deliberately NOT applied: null hypotheses, limitations, and threats to validity remain stated plainly, as the academic register requires.
5. **NOVEL-POLISH** (academic). Sentence-length distribution, lexical variety, and paragraph transitions hold to the Kepler (precision) and Vox (argument-clarity) bar. Opening (Abstract, Introduction 1.1) and closing (Conclusion 8.7) are strong. No changes required.
6. **EXECUTIVE-STORYTELLING** (light, academic). Abstract and Introduction are stakes-first (the two-order-of-magnitude ellipse contraction and its decision relevance lead). No process-label scaffolding ("this section will") found. Payoff arc clear. No changes required.
7. **KING CHARLES REGISTER** (framing only). Ran `query_style_exemplars.py --mode ceremonial --limit 6`; inferred the dignified, measured stewardship-and-service register. The executive-facing framing already carried it: Abstract opening (the precision improvement across the robotic era), Introduction opening framing (the ellipse as the most consequential number), and Conclusion closing 8.7 ("the quiet history of the discipline ... The contribution is the instrument and the discipline; the verdict awaits the data"). Technical body left in its analytic register. Guardrails honored: inspiration not impersonation, no royal signature forms, no copied transcript phrasing, no claim of Royal Household approval. No changes required.
8. **LITERARY-METHODS** (Deictic Shift + Rhetoric of Science). Temporal references anchored to calendar dates (mission years given throughout: Viking 1976, MSL 2012, InSight 2018, Mars 2020 2021). Entities introduced before pronominal reference. Epistemic modality calibrated to the design-stage grade: estimates stated in conditional/future terms ("would estimate", "expected sign", "if H1 holds"), never as executed findings. No changes required.
9. **HAAF RECHECK** (closing logic pass). After the prose and citation passes, confirmed no HAAF layer broke: every claim retains warrant and qualifier; causal mechanisms were not flattened to correlations by the humanizers (the time-technology collinearity caveat survives intact); confidence statements survived; the architecture trace remains correctly out of scope and resolves to plain-prose decision relevance. Reference: `haaf/references/haaf_quality_gate.md`.

## Citation hygiene (final step)

Unified every in-text citation to a single clickable numbered scheme. The corpus keys already mapped bijectively to sequential reference numbers (S01..S24 -> 1..24, R025..R130 -> 25..130), verified zero mismatches. Converted **399 in-text citation groups** (760 individual key references) from the `[S05]` / `[S03, S04, R075]` corpus-key form to clickable `[\[N\]](#ref-N)` links, splitting multi-key brackets and expanding the one appendix range (`R071-R078`) to its members. Added `<span id="ref-N"></span>` anchors to all **130** reference-list entries (kept the human-readable `**[KEY]**` label alongside). All 115 DOIs are clickable `https://doi.org/...` links; the 13 NTRS reports without journal DOIs carry resolvable NTRS citation URLs. No citation was repointed to a different source; only style normalized and links added.

## Metrics: before and after

| Metric | Before | After |
|---|---|---|
| Em dashes (dissertation) | 0 | 0 |
| Em dashes (deck) | 0 | 0 |
| Clickable numbered body citations | 0 | 760 |
| Reference-entry `<span id="ref-N">` anchors | 0 | 130 |
| Unresolved corpus-key citation markers | 528 (raw `[S/R..]`) | 0 |
| Body link targets missing a reference anchor | n/a | 0 |
| Clickable DOI links in reference list | 115 | 115 |
| Resolvable NTRS URLs (no journal DOI) | 13 | 13 |
| Word count (dissertation) | 64,350 | 64,484 |

The em-dash count was already zero on intake (the document used commas, semicolons, colons, and parentheses throughout); the grep at the end confirms zero remain. Three legitimate en dashes (U+2013) remain in numeric ranges and a formula middle-dot in the appendix table; these are correct and are not em dashes.

## Banned phrases removed / soft intensifiers tightened

- "the **crucial** counter-current" -> "the **decisive** counter-current" (Ch 1.3.1, InSight; "decisive" is load-bearing: InSight is the identifying case).
- "and, **crucially,** the dissemination" -> "and the dissemination" (Ch 2.3; intensifier removed).
- "made flexible and **robust enough**" -> "made **general enough**" (Ch 3.2; the cited finding is about generality across vehicle/mission range).

## Preserved deliberately (NOT removed)

- "robustness battery" and all "robust to X" / "outcome-robust" uses: these are econometric terms of art (a result invariant under perturbation X), load-bearing in this design.
- Two "robust nonlinear compound controller" instances: the verbatim title of reference R051 (now [51]).
- Null hypothesis, limitations, threats-to-validity, and "fragile/weakly identified" language: required by the academic register; the zero-negative-words rule was intentionally not applied.

## Flagged for the author

1. **No substantive items.** The prose passes (layers 1-9) were already complete and high quality on intake; the only outstanding work was the citation-hygiene normalization, which is now done.
2. **Reference coverage:** 111 of 130 references are cited in-body; 19 (e.g., S10, S11) appear in the reference list but are not cited in the running text. This is normal for a reference list but worth a glance if the author wants every listed source cited.
3. **Heredoc caveat (process note, not a document issue):** intermediate verification using shell heredocs mangled backslash regexes and produced false-negative counts; all final counts above were produced with file-based scripts and are authoritative.