Su-CC: Quality engineering (Step 4b)

Project: Ching-Chuan Su (C.-C. Su), National Tsing Hua University, Hsinchu: the "local-ether" model and "Quantum Electromagnetics". QE date: 2026-10-06. Inputs checked: _text/manifest.json, _textprep.log, all 18 digest files (read in full), deep/ (13 files, searched and sampled), facts.json, claims.md, sites.md, research/ (notes and file list), physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py (run). Tags: [read] from the extracted text or a review file; [web] with URL; [memory] with confidence; [run] output of a command I ran.

Verdict: MINOR

The work is complete, careful and honest. Every manifest document was digested and every substantive paper got a deep read with page-image checks. The load-bearing numbers I checked match the source text. The primer follows STEP 4 and the reader profile, and example.py runs. None of the defects below is likely to change a central verdict. They are: three predictions Su lists that nobody reviewed, one related experimental record the reviews did not know about, a few retention gaps in the arbiter and primer, and some process slips (an Investigator run that stopped early, and one verdict resting on unchecked [memory]). All can be fixed by updating facts.json and claims.md, adding a "QE revision" to arbiter.md and updating primer.md.

1. Coverage

No text was skipped that should have been read. 17 manifest entries, 17 facts.json entries [read: facts.json]. _textprep.log reports 17 documents, 18 chunks, all extracted [read].

2. Shortcuts

• No central claim, experiment or derivation rests on a digest alone. Every one of the six claims is tied to deep-read documents.
• The outline is digest-only. That is fine for a summary, except that it is the only document in the folder that lists four of Su's predictions (section 3, items 1-3). Those predictions now rest on a digest line and were never examined.
• Investigator downloads that bear on the claims were read only at abstract level by the reviewers: physics/0208083 (modified Schrödinger equation, atom-radiation interaction), f2c (Riemann force law, 2006) and f2b/f3b (AP-S digests) [read: mathematician.md section 4, "not read in depth; I rely on abstracts"]. The Contrarian read part of f2d (Barnett). For f2c, f2b and f3b this is acceptable because they repeat claim 4. It is not acceptable for physics/0208083, which probably holds the "stronger second-order effect in atom-radiation interaction" prediction (item 3 below).
• Spot reading of the experimental papers Su cites (Brillet-Hall, Staudenmann, Krisher and others) was not possible. The source-gap note says so and is repeated faithfully in every later file [read: claims.md, arbiter.md, primer.md]. Acceptable.

3. Missed material (from my own reading of the digests)

1. Transition-dependent gravitational redshift (a testable prediction, not reviewed). The outline says: "in the gravitational redshift the quantum origin and the transition-dependence are predicted" [read: outline p.8, _text lines 431-432]. Su's chapter 5 abstract (Investigator download) says: "the derived gravitational redshift originates from a quantum nature of the involved atoms and can be different in some other interaction. This discrepancy then provides a means to test the local-ether wave equation" [read: research/files/wayback-qem/ch5.htm]. In f6 Su also says different mass or speed dependences "caused by some other interactions are not precluded" [read: deep/f6 p.5]. This makes the gravitational redshift depend on which atomic transition is used, which violates the universality of gravitational redshift (local position invariance). Mainstream tests compare clocks of different species (hydrogen maser, caesium, rubidium, optical clocks) as the solar potential changes over the year, and find no species dependence at roughly the 1e-6 level of the redshift [memory, medium; to be web-checked]. Neither claims.md, the three reviews, the arbiter nor the primer mentions this prediction. It may be Su's one gravitational prediction with an existing test. Action: deep reads 2 and 3, then a verdict.
2. Stellar downshift depending on the star's mass. Outline p.8: "in the stellar spectrum a gravitation-induced downshift depending on the mass of the star is predicted" [read]. Not reviewed. It probably coincides with the ordinary GR surface redshift (for example white dwarfs) at weak field and differs only in Su's non-GR strong-field form (f5c). Someone should say which. Action: arbiter note after deep read 1.
3. Two more listed predictions. Outline p.8 also lists "in the magnetic deflection an effect of the relative motion between the deflection yokes and the accelerating electrodes" and "in the interaction of atoms with electromagnetic radiation, a stronger second-order effect" [read]. f2a's CRT section (pp.8-9) argues for a null result in the matrix frame [read: deep/f2a section 7]. The non-null prediction (electrodes moving relative to the yoke) is not sized anywhere in the review. Action: deep read 4 for the CRT effect; read physics/0208083 for the atom-radiation item.
4. Experimental record bearing on Su's F2 term (related work missing). f2a predicts an extra conservative electric force F2, proportional to drift speed squared over c^2, on a charge at rest next to a neutral loop carrying a steady current [read: deep/f2a p.10, eq. 28]. The arbiter and primer call it "Never properly tested". There is a published experimental line on exactly this class of effect (a second-order electric field outside a closed steady-current circuit, as predicted by Weber-type laws, the tradition Su's 2006 Riemann paper joins): - Edwards, Kenyon and Lemon (Utah State University), Phys. Rev. D 14, 922 (1976): they reported a potential from a superconducting Nb-Ti coil that scaled as I^n with n = 2.02 +/- 0.05 [web: https://arxiv.org/pdf/1009.0474 (abstract summarising the 1976 result); https://isidore.co/misc/Physics%20papers%20and%20books/Zotero/storage/C6B7H3S5/PhysRevD.14.html]. - Later null tests, as I recall them: Bartlett and Maglic, Rev. Sci. Instrum. 61 (1990) [memory, low-medium; not found by my one search]. This is pertinent supporting and contrary evidence for claim 4 whatever its sign, and it is a candidate "anomalous result" in the Contrarian's sense. Action: the Investigator or arbiter should web-check both papers, and someone should compare Su's F2 coefficient with the measured or bounded field. The claim 4 verdict for F2 may then move from "Never properly tested" to "Tested, not supported" or "contested".
5. Index-dependence in Sagnac loops (f3a, Harzer 1914, Dufour and Prunier 1942). Su offers the "various index-dependences" (n^2 on the loop rotation rate, 1 on Earth's rotation) as a test [read: deep/f3a p.10; outline p.8]. The deep reader found both coefficients are probably what SR gives (Post 1967), so the test does not discriminate [read: deep/f3a lines 195, 215]. No reviewer, nor the arbiter, carried this into a verdict; the item sits only in facts.json. Action: arbiter one-liner, "Established, not discriminating", if the reviewers agree.
6. Hughes-Drever isotropy. Su says his ECI-speed argument explains it "whatever the dependence of nuclear magnetic interaction on mass and speed" [read: deep/f6 p.8]. The deep reader shows the agreement comes from symmetry, not calculation. Only the Contrarian mentions Hughes-Drever. The arbiter's claim 3 "Established" list and the primer leave it out. Minor; add it to the list with that caveat.
7. CMBR second-order Doppler test. Su proposes that a second-order measurement of the CMBR dipole could discriminate propagation models, and treats the CMBR sources as a further enclosing local ether [read: digest EPJC p.11-12]. Not in claims.md or the reviews. Peripheral; one line in facts.json and claims.md item 5 is enough.
8. Smaller numbers not carried forward. The Japan-Korea link Sagnac error of about 40 ns [read: digest EPJC p.3-4]; the Hafele-Keating measured values (-59 +/- 10 ns and +273 +/- 7 ns [memory, high]) next to the predictions -40 and +275 ns that Su quotes (the Physicist notes these are Hafele and Keating's own predictions; nobody gives the measured values with a source); the Vessot (Gravity Probe A) link Su discusses [read: digest f6 p.7]. Minor.

4. Accuracy spot checks (against _text/)

I found no transcription errors in the load-bearing figures.

5. Retention

Pertinent material that the reviews or the Investigator recorded but that the arbiter or primer dropped. I see no sign it was dropped because it was judged negatively; it looks like condensation. Still, the rule applies.

• Su's observational evidence for quasar ejection and for a single governing potential. Su cites: the X-ray quasar 8 arcsec from the NGC 7319 nucleus with a V-shaped filament (Galianni et al. 2005); X-ray filaments from NGC 3628 to quasars at z = 0.995 and 2.15 (Arp et al. 2002); the NEQ3 triple (z = 0.1935, 0.1939, 0.2229); the hydrogen bridge from Arp 220; and the abrupt, non-overlapping H-alpha redshift boundary in MCG 7-25-46 (z = 0.003 and 0.098) and NGC 7320, which he argues supports his single-potential rule while noting the original authors' projection reading [read: digest f5c p.5-7; deep/f5c section 2]. The Contrarian keeps part of this (NEQ3, MCG 7-25-46, NGC 7320). The arbiter gives only "not settled ... interest is recorded", and the primer gives a generic "sometimes apparently linked by filaments". The "abrupt redshift change" is a key claim in facts.json (f5c) with no verdict anywhere. Action: the arbiter lists these cases with their sources and stakes (Arp, Burbidge, Lopez-Corredoira and Gutierrez as proponents of association) and gives the abrupt-boundary argument a verdict or an explicit "not assessed". The primer adds two or three of them to section 3.8 or to the evidence table.
• Investigator finds not weighed or not carried forward. Wu and Tsao 2019 (J. Mod. Phys.) was downloaded but appears in no review, nor in the arbiter or primer. Persson (Physics Essays 2021) and the other works citing Su are listed only in sites.md. These are minor; the primer should name them in one line under "Related and supporting work", or say why they are left out.
• Hughes-Drever (section 3, item 6) and the Sagnac index-dependence (item 5): present in the deep reads and facts.json, absent from the arbiter and primer.
• Everything else is retained well: Marsen's reading of the Brillet-Hall residual, Beckmann's priority, Gift, Field, Kipreos, Kennard, the Barnett prediction, the railgun paper, the engineering citations of Su's Sagnac formula, and the full absence-of-evidence analysis of Su's record after 2006.

6. Process

• Source tags. Present and honest throughout. [memory] items carry confidence. Advocates (Marsen, Gift) and Su himself are marked as interested parties. The single-source death report is described ("documented in one published advocate source; no obituary ... found yet"), not labelled. Good.
• Web checks. The Physicist used 4 of about 6 calls, the Contrarian 5, the arbiter 4 of 8. One verdict still rests on medium-confidence [memory] that was not web-checked although budget remained: claim 3, "Compton as wave-only scattering: Tested, not supported", which rests on the Bothe-Geiger and Compton-Simon coincidence experiments and on Kimble, Dagenais and Mandel 1977, all [memory, medium]. This is a defect against the "Web verification" rule. Also unchecked: Hafele-Keating's measured values, Mandelberg-Witten's precision and Le Bellac and Levy-Leblond. These are lower stakes because no verdict turns on them.
• Verdict scale. Used consistently, and split verdicts are clearly separated. "Established, not discriminating" is a qualifier on a scale verdict, which is acceptable. The arbiter's reasoning for choosing "Internally inconsistent" over "Refuted by calculation" (D4) is explicit and sound.
• Investigator time rule. The run used about 6.6 of the 10 minutes [read: sites.md header], yet nine sites are marked "not reached (time limit)". Under prompt.txt they should read "not reached (stopped early)", and the run should have continued to at least 9.5 minutes. Su's Chinese name was never established, which blocks the obituary and NTHU-record search the arbiter names as deciding. The arbiter's guess (蘇清泉) is flagged as unverified. Good.
• Arbiter's added test (Herrmann et al. 2009). This is the review's most consequential new step. It is properly sourced to the arXiv text [web: https://arxiv.org/pdf/1002.1284], and its two caveats (an incidental test; two-cavity equivalence argued, not derived) are carried into the primer. The primer's Step 5 supplies the two-cavity identity using Su's eq. 13, which closes much of the second caveat. Acceptable.
• Primer against STEP 4 and the reader profile. All nine sections are present in order. Prerequisites have refreshers and books, with chapter numbers flagged where uncertain. Symbols and acronyms are defined, the Klein-Gordon equation is introduced from scratch, the worked example uses real numbers, there are five check-yourself questions with answers, and the further reading includes Su's own works. Good. One small point: §3.4 gives "about 200 microseconds" for the Earth-Venus Shapiro delay [memory, high] beside Su's 230 (232.5 recomputed) without saying why they differ (geometry and conjunction). One sentence would fix it.
• example.py. It runs cleanly with the venv, prints "All checks passed", and reproduces 62.5 Hz, 4.46e-13 and the ratio 4.4e4 [run: .venv/bin/python example.py].

7. Remediation actions

Deep reads (most important first; at most 8):

1. outline-c283c7, pp.4-9, Sonnet. List every prediction Su states (p.8 list and the section 4-5 text) with the page. For each, say which paper in the folder derives it, or that none does. Flag the four not reviewed: transition-dependent gravitational redshift, star-mass-dependent stellar downshift, the yoke/electrode relative-motion effect, and the stronger second-order atom-radiation effect.
2. f6-03029f, pp.4-6, Opus. Does Su's model predict a transition- or species-dependent gravitational redshift (and speed shift) for hyperfine, Zeeman or nuclear transitions versus electronic ones? Give the form, any stated size, and whether Su commits to a number that can be compared with clock-comparison tests of redshift universality.
3. f5a-374ccb, pp.4-7, Sonnet. Same question for the gravitational part: how the gravitational potential enters the quantum-state energy, whether the coupling is the same for all states, and where Su says it "can be different in some other interaction".
4. f2a-45067c, pp.8-11, Opus. (a) What does the force law predict for a CRT whose accelerating electrodes move relative to the deflection yoke (the outline's "magnetic deflection" prediction), and how large is it? (b) Evaluate F2 (eq. 28) for a multi-turn coil of known current, and give the potential difference it predicts between the coil and a nearby electrode, so it can be compared with Edwards, Kenyon and Lemon 1976 (field quadratic in I).
5. f5c-c6f6cd, pp.5-7, Sonnet. Tabulate every observational case Su cites for ejection and for the abrupt redshift boundary: object, redshifts, separations, the source he cites, and the alternative reading he records. This is for retention in the arbiter and primer (the deep read already covers part of it).
6. A-local-ether-model-of-propagation-of-electromagnetic-wave-177f1c, pp.11-12, Sonnet. Su's CMBR treatment: the enclosing CMBR "local ether", the proposed second-order Doppler test, and any number given.

Other reads and web checks (not deep reads of manifest documents): - research/files/arxiv/physics-0208083v1.pdf, in full (Sonnet): find and size the "stronger second-order effect in the interaction of atoms with electromagnetic radiation". - research/files/wayback-qem/f2d.pdf (Barnett), in full: the predicted size, so the Barnett item in claim 4 has a number. - Web: Edwards, Kenyon and Lemon, Phys. Rev. D 14, 922 (1976), and the follow-up and null tests (Bartlett and Maglic 1990; Lemon, Edwards and Kenyon 1992). Also the Compton coincidence and antibunching papers behind the claim 3 verdict, and tests of redshift universality between clock species. - Investigator, next run: establish Su's Chinese name, then the NTHU alumni e-news of 9 May 2008, the NTHU repository and NDLTD; use the full time allowance; relabel this run's unreached sites "not reached (stopped early)".

Changes to later files: - facts.json: add to the outline entry the four unreviewed predictions with page refs, and change how_read if deep read 1 is done. Add the transition-dependence statement to f6 and f5a key_claims. Add the CMBR test to the EPJC entry. Add F2's experimental context to mainstream_conflict in f2a. - claims.md: claim 2 gains "redshift may depend on the transition (Su: 'can be different in some other interaction')". Claim 5's prediction list gains the transition-dependent redshift, the star-mass downshift, the yoke/electrode effect, the atom-radiation effect and the CMBR test. - arbiter.md, add a "QE revision" section with: (i) verdicts for the transition-dependent redshift (probably "Tested, not supported" if Su commits to a species dependence, otherwise "Not testable as stated"), the stellar downshift, the yoke/electrode effect and the atom-radiation item; (ii) the claim 4 F2 verdict reconsidered against Edwards et al. 1976 and the later tests, with both results stated; (iii) the Sagnac index-dependence ("Established, not discriminating", per deep/f3a) and Hughes-Drever added to claim 3; (iv) the quasar observational cases and the abrupt-boundary argument stated, with a verdict or "not assessed"; (v) the Compton coincidence verdict web-checked or explicitly marked as resting on [memory]. - primer.md: add the new predictions and verdicts to section 4 and the evidence table; one or two lines in §3.8 naming Su's cited quasar cases; a sentence on the Edwards et al. experiments in §3.6 or the claim 4 row; Wu and Tsao and Persson in "Related and supporting work"; one sentence in §3.4 reconciling 200 and 230 microseconds. Rerun example.py only if section 5 changes (it should not).

Recheck

Date: 2026-10-06. I re-read arbiter.md (QE revision, Q1-Q8), primer.md, claims.md, facts.json and sites.md, and reran example.py ([run]: prints "All checks passed").

Remaining items are three: Hafele-Keating/Japan-Korea/Vessot numbers (minor), the unread physics/0208083 and f2d, and the sites.md relabel with the Chinese-name search. None changes a verdict, and each is disclosed in the files.

Final: OPEN (3 issues)