QE (Step 4b): Wallace-Byran, "The Farce of Physics" (1993)

Date: 2026-10-04. QE agent: Opus. Inputs checked: _text/manifest.json, _textprep.log, digest/farce-d10910.01.md (the only digest), deep/farce-d10910.md, facts.json, claims.md, sites.md, research/ (9 notes; research/files/ is empty), physicist.md, mathematician.md, contrarian.md, arbiter.md, primer.md, example.py. Also checked directly: the extracted text _text/farce-d10910.txt, lines 594-4921 (Chapters 2-6) read by me; the original farce.txt decoded as CP437 for two numbers; and Wallace's 1969 paper (Spectroscopy Letters 2, 361-367), which I downloaded from http://sciliterature.50webs.com/SpecLetters1969-p361-367.pdf to /tmp/qe-wallace/ and read as page images (pp.364-367).

Verdict: MINOR

The strong-form verdicts (light speed is c+v; the mass-dynamics model; the cosmology) are not affected. Two things do need fixing: - One central sub-verdict, claim 1(a) "Internally inconsistent", rests on a misreading that I confirmed against the page image and against the book. - The Step 1a digest covers only the opening of each chapter. As a result, a good deal of pertinent material in Chapters 2-6 never reached facts.json, claims.md, the arbiter or the primer.

The MINOR path covers both: deep reads, then updates to facts.json and claims.md, a "QE revision" section in arbiter.md, and an updated primer.md. None of it requires Steps 2-4 to be redone.

---

1. Coverage

Only one document is in the manifest, and textprep skipped nothing. The problem is the digest itself: - The whole 300 KB book arrived as one "page", so it went to Haiku as one chunk. The digest has 9 entries. Each of Chapters 1-6 is summarised from its opening pages only and then ends "[text continues]". - The Chapter 7 entry is empty ("[Section appears to continue...] [text continues]"). - My estimate is that the digest covers about 15-20% of Chapters 2-6. Examples: it stops Ch.3 at "mathematics is a language" (line ~1560 of 2466), Ch.5 at "Lincoln Laboratory made a complete c analysis" (line ~3445 of 4128), and Ch.6 at the 1973 Found. Phys. abstract.

facts.json's how_read says "Chapters 1-6 rest on the digest only". That overstates the coverage, because most of those chapters were never digested at all.

Mitigation: - The Contrarian and the Physicist read the Ch.4-5 text directly in places (line refs 2480-2560, 3401-3414, ~3535-3600, ~2826-2830). - The Mathematician, the Contrarian and the Primer author read the 1969 paper as page images. - Claim 1 therefore does not rest on the digest alone.

Nobody read Chapter 3 after line ~1560, most of Chapter 2 after the Langmuir section, Ch.5 lines ~3800-4128 or Ch.6 lines ~4200-4921 (see section 3).

Process defect for the tools: textprep emits a plain-text file as a single page, so a 300 KB book becomes one 300 KB chunk. That is beyond what one Haiku digest call will summarise. textprep should split long plain-text files into synthetic pages (for example every 3,000 characters or every 60 lines) so the 60,000-character chunking and page-by-page digesting work. This probably affects other single-file text projects too.

2. Shortcuts

• Chapter 7 (the theory) got an Opus deep read. That was acceptable and well done: the CP437 re-decode recovered "2½", "µA" and "90°", and I confirmed "2½" and "1.2 and 1.3 µA" in the original myself.
• Claim 1 (radar). No deep read of Chapters 4-5 was ordered, which is a defect under STEP 1b. It was partly cured by the reviewers' own reads and by the 1969 paper. The radar correspondence in Ch.5 (lines 3344-4128) holds the most load-bearing non-paper evidence for claims 1, 2 and 6 and still needs a systematic read. Examples: Wallace's account of how Lincoln Lab removed the daily variations, the Duncombe-corrections argument, the Moyer argument and the Kotel'nikov 1975 Soviet radar result.
• Claim 6 (obstruction). It rests on the parts of Ch.2, 4 and 5 that the reviewers happened to read. Chapter 6's accounts were not used: the Pulkovo dismissals, McVittie's complaint, and Shapiro's published 1985 reply.
• Missing source papers. Nobody read the 1971 radar paper (Spectroscopy Letters 4, 79), although sites.md says it is free at GSJ. Every later step says so honestly, so this is a disclosed gap rather than a hidden shortcut.

3. Missed material (from my reading of Chapters 2-6)

Each item below is missing from facts.json, claims.md, the arbiter and the primer. A reviewer touching an item is noted in brackets. All are [read: p.1, line N] of the extracted text.

Radar and light-speed evidence (claims 1-2) 1. Shapiro published a reply. I. Shapiro, Scientific Ethics 1(2), 10 (1985) [ref 72, L5595]. It answers Wallace's "The Great Speed of Light in Space Coverup", Sci. Ethics 1(1), 2 (1985) [ref 71]; Wallace mentions it at L4510. This is Shapiro's side in print, from an interested party. Nobody located it, although it bears directly on claim 2 and on the arbiter's "what would settle it". The Investigator should target it. 2. How the daily variations disappeared, in Wallace's account. Wallace says Lincoln Lab removed the large daily variations by switching to a constant observing time (12:00 UT), "even when the planet was not observed or in some cases was not even visible", citing Evans et al., AJ 70, 486 (1965) [L3606-3611]. He also says the synodic variation in the GR fit of Ash, Shapiro and Smith, Fig. 4, could only have been removed by "empirical corrections similar to Duncombe's corrections" [L3715-3735]. Both are specific, checkable charges against the reduction method. They are the core of Wallace's case for claim 1(b), and neither review weighs them. Evans et al. is in sites.md only as an ADS listing. 3. Kotel'nikov et al., Astron. Zh. 53, 1270 (1976) [ref 79; L4018-4035]. This Soviet 1975 Venus radar paper reports measured-minus-calculated delays reaching "3500 microseconds, which ... comprises 500 km", with "different characters" at different conjunctions. Wallace cites it as support. It is a published residual of the same size as his "3 ms". It is not specific to c+v, since it is measured against planetary theory, but it is pertinent and absent everywhere. 4. The Moyer argument. Wallace argues that JPL's Moyer, Celes. Mech. 23, 33 (1981), evaluates distance at signal reception without the −tv/2 term and calls the main term "the Newtonian light time", which Wallace reads as c+v [L3810-3838, L4740-4750; also L2826-2830]. The Physicist addressed it and judged it a confusion. The arbiter and the primer dropped it entirely. Retention defect: it is Wallace's main claim about modern ranging, and it stays in the record beside the Physicist's rebuttal. 5. McVittie. G. C. McVittie, AJ 75, 287 (1970) [ref 81], and a 1970 letter quoted by Wallace saying that Shapiro's secrecy about methods and results "makes him wonder whether there is something to be concealed" [L4523-4529]. This is a second named astronomer's complaint about data access. It bears on claim 1(b) and on the "Absence of evidence" analysis. It comes from one published source (Wallace's book) and should be quoted and weighed as such. 6. Hollingsworth (AFCRL), letter of 7 Nov 1969: "a number of reasons why it is impossible for him to persuade Dr. Shapiro or other Lincoln Laboratory people to prove that the velocity of light in space can be measured as anything else than a constant value c" [L3753-3757]. Shapiro's reply of 9 Dec 1969 was that he was "too much occupied ... to direct any large-scale investigation" [L3770-3772]. These sharpen the data-access record (arbiter explanation 3). 7. Einstein's statements of falsifiability [L2914-2930]. To Freundlich (1913): "If the speed of light is the least bit affected by the speed of the light source, then my whole theory of relativity and theory of gravity is false." On Dayton Miller's ether-drift repetition (1921): otherwise "the whole relativity theory collapses like a house of cards". The arbiter mentions "Einstein's own statements (1912-13)". Dayton Miller's ether-drift result is a well-known anomaly claim and appears nowhere. 8. Salamin's binary-star objection and Wallace's reply [L4697-4750]. Salamin argued that c+v would make some binaries show simultaneous redshifts from both members. Wallace replied by citing Struve's "gas whirlpools" and Fox's remark that Struve's hypothesis has "consequences similar to those of the Ritz theory". Wallace also states there that solar-system transit times "do not suffer from the same ambiguity" as binaries. This is evidence against claim 1 together with the inventor's answer; both belong in the record.

Relativity tests other than radar (new peripheral claim) 9. The eclipse claim. Wallace cites F. Schmeidler, Sky & Telescope 27(4), 217 (1964) [ref 49; L2446-2464]: a plot of 92 star shifts from the 1922 eclipse "going in all directions, many of them going the wrong way". He uses it to say the 1919/1922 light-bending results depended on the reduction method and the omission of stars, and he calls them "Trimming and Cooking". Essen's letter adds that a former Pulkovo director showed the "1915 [sic] eclipse results did not support Eddington's claim" [L4279-4282]. This is a testable claim about the record and is missing from claims.md and every review. The mainstream answer (later radio-interferometric deflection measurements agree with GR to about 1e-4) is [memory, high] and should be web-checked when it is added. 10. Essen's critique of special relativity: thought experiments, SR as "a new system of units", "no experimental support" because accelerations are always involved [L2062-2160]. Essen built the first caesium clock, so this is the strongest professional ally Wallace quotes. The reviews mention Essen only in connection with Hafele-Keating. 11. Parshin (Aeroflot Academy of Civil Aviation) reportedly ran "a modern variation of the Ive's experiment". He could publish the data in a Soviet journal but not his theoretical analysis [L4394-4398]. This is a related experiment and a suppression account; no source was given.

Cosmology (claim 5) 12. LaViolette, ApJ 301, 544 (1986) [ref 101; L1240-1305]: a tired-light model fits four cosmological tests better than a no-evolution expanding model. It is a peer-reviewed paper Wallace relies on, and it appears nowhere. The supernova time-dilation test the arbiter uses against tired light came later (1990s). The arbiter should say so and keep LaViolette in the record. 13. Arp's loss of telescope time (Burbidge's review, quoted at L1311-1360): "interminable refereeing, blackballing of speakers ... denial of telescope time". Also Segal's chronometric cosmology (MIT) on the background radiation [L1361-1368]. The arbiter leaves Arp "open" but omits the documented treatment, which is a suppression account from a third party.

Obstruction (claim 6) 14. Pulkovo Observatory. Tolchelnikova's 1988 letter says "two scientists were dismissed because they discovered some facts which contradicted Einstein" [L2565-2566, L4204-4205]. Wallace also reports that under Director Abalakin two anti-relativist women were forced into early retirement and later reinstated [L4458-4467]. Both come from one published source (the book, relaying a letter and conversations); no second source has been found yet. 15. Physics Today and APS. Physics Today printed no letter of Wallace's after Davis's resignation; Lubkin rejected his 9 Jul 1984 "speed of light is c+v" letter on 4 Jan 1985 [L2800-2850] (the Physicist notes the Lubkin rejection). A PRL editor rejected a paper without refereeing [L2752-2755]. R. R. Wilson's 1985 APS address joked about ostracizing "the culprit" and holding up publication "by the use of our well-practiced referees" [L2680-2685]. Beckmann (Galilean Electrodynamics) cut Wallace's reply to Salamin [L4751-4756]. That last one is notable because it is an anti-relativity journal declining him. 16. NSF 1978 wording, in full. The rejection called his model "beset with a number of fatal conceptual difficulties" [L943-946]. The arbiter quotes only the first half. Those "difficulties" are the referees' side and should be quoted.

Balance (retention) 17. What Wallace accepts. "I certainly have no problem with E=mc² ... So in general, much of relativity theory is true, but many of the original arguments are not" [L4549-4552]. He also accepts Hafele-Keating's dilation as real, relative to an absolute frame. Primer §1 says he "concluded that the whole edifice of twentieth-century physics was 'an elaborate farce'". That is his title, but the primer should add this sentence so his position is not overstated. 18. Leads dropped between the reviews and the arbiter/primer. Rapier, Spectroscopy Letters 4, 303 (1971), and Wegener & Øhrstrøm, Nuovo Cimento B 30, 291 (1975), are both Investigator leads citing Wallace. All three reviewers mention them; the arbiter and primer drop them. Also: Beklyamishev's 1991 Leningrad paper on Wallace's Venus analysis [L4867-4880]; Vigier's offer to publish in Physics Letters A and his stated "doubts on Prof. Shapiro's observations" [L4262-4268]; and Hayden's view that SR is "inherently irrefutable" [L4681-4690]. These are minor, but they are pertinent supporting or related material.

4. Accuracy spot checks

Checks 4 and 5 change the basis of the claim-1(a) verdict (see Remediation, item A).

5. Retention check

• Moyer/JPL argument (missed item 4): weighed by the Physicist, then dropped by the arbiter and primer. Defect. Restore it in both, with the Physicist's rebuttal beside it.
• Rapier 1971 and Wegener & Øhrstrøm 1975 (Investigator leads, in all three reviews): dropped by the arbiter and primer. Minor defect.
• The arbiter's handling of negatively judged material is otherwise good. It explicitly keeps several items "on the record, not suppressed": Ash-Shapiro-Smith's station incompatibility, Standish, Tolchelnikova-Murri, Hollingsworth, Sekerin, Kelly's FOIA reanalysis and Essen on Hafele-Keating, and the Arp anomalies. The primer carries most of them into its evidence table and its "record goes dark" section.
• NSF referees' "fatal conceptual difficulties" (missed item 16): half-quoted. That is not a suppression, but the full sentence is the referees' side and should be quoted.
• Primer §1 framing (missed item 17): Wallace's acceptance of E = mc² and of much of relativity should be stated.
• Contrarian line 70 calls Shapiro's admission "a single-source account" as a verdict label. SOURCE HONESTY forbids reducing evidence to such a label. The arbiter's wording ("documented in one published source ... no transcript found") is correct, so nothing downstream inherited the problem.

6. Process

• Source tags. Present and honest throughout. The primer introduces [calc], which is acceptable since it is defined. The arbiter lists every remaining [memory] fact a verdict depends on: electron point-likeness, the CMB and supernova (1+z) time dilation, Newcomb's accuracy, ranging accuracy after 1969. These should be web-checked in the remediation round, especially the electron point-likeness limit, which carries the claim-4(a) verdict "Tested, not supported", and supernova (1+z), which carries claim 5(a).
• Web checks. The arbiter used 8 of 8 (Hafele-Keating, Jupiter, the Brecher citation, Shapiro-Ash-Tausner, the Texas Symposium) and corrected the Physicist's Brecher citation. Good.
• Verdict scale. Applied consistently and with split verdicts. Two notes:
• Claim 6 gets "Not testable as stated" as physics plus a prose finding on the events. That is reasonable.
• Claim 1(a) "Internally inconsistent" is the wrong category even before the misreading: the algebra is consistent for the stated rule (check 4).
• Primer and STEP 4. All nine sections are present in order. Prerequisites have textbooks. The worked example uses real 1961 numbers and reproduces Fig. 1 from orbital geometry. The check-yourself answers are at the end. It is written for the reader profile and defines every symbol. The primer author did the right thing on the epoch question: it flagged the correction to QE instead of issuing its own verdict.
• example.py. Runs cleanly in the project venv. Its outputs match the primer: tv/2 table, 30,000%, 128 km versus 257 km, Alvager |k| < 2.1e-4, spin ratio 33.7, radius 1.93e-13 m.
• claims.md. It still says Shapiro's side "is not given" and that the residuals were analysed "1961-1966". The arbiter corrected both (corrections 1 and 2), but claims.md was never updated.
• research/files/ is empty. The 1969 PDF the reviewers used is not saved in the project; each reviewer downloaded it to scratch. sites.md records "Owner: approve download". Recommend the owner approve saving it there.

7. Remediation (one round)

Deep reads (all on farce-d10910, the only manifest slug; most important first)

Also recommended (outside the manifest, for the Investigator's next run rather than as deep reads): - Shapiro, Sci. Ethics 1(2), 10 (1985), and Wallace, Sci. Ethics 1(1), 2 (1985). - McVittie, AJ 75, 287 (1970). - Kotel'nikov et al., Astron. Zh. 53, 1270 (1976) (Sov. Astron. translation). - Evans et al., AJ 70, 486 (1965). - Moyer, Celes. Mech. 23, 33 (1981), to read eq. (A6) and the epoch convention. - Schmeidler, Sky & Telescope 27, 217 (1964). - Wallace's 1971 radar paper, Spectroscopy Letters 4, 79 (GSJ). - Save the 1969 PDF into research/files/ if the owner approves.

What each later file must change

A. arbiter.md: add a "QE revision" section 1. Claim 1(a): withdraw "Internally inconsistent". The 1969 paper (p.364) and the book (L3810-3822) both put Eq. (1) and Eq. (2) at the reception epoch, and each follows exactly from its stated reflection rule. Suggested replacement, split as follows: - (i) The derivation: consistent for the c+2v reflection rule. The prediction depends on which reflection rule an emission theory adopts (c+v gives tc/2 − tv/4). - (ii) The 1961 eight-point comparison as published: Tested, not supported, because by Wallace's own yardstick (Newcomb's N, Fig. 1) the constant-c curve E lies closer than G. The caveats: Newcomb's tables are themselves uncertain by an amount the paper does not state, and G's agreement comes only from Wallace's own Cowell integration with unstated initial conditions. If the arbiter judges Newcomb too imprecise to decide either way, "Never properly tested" is the defensible alternative. State which, and why. - 1(b) and 1(c) are unchanged. 2. Delete open question 1 and the matching row of the settling table; the question is answered. Note that the Mathematician's reading of Fig. 1 was wrong. 3. Add to claim 1, kept on record: - Wallace's Moyer/JPL argument beside the Physicist's rebuttal. - The Kotel'nikov 1975 residuals (up to 3500 µs or 500 km against planetary theory; not c+v-specific). - The 12:00 UT and Duncombe charges, marked as testable against Evans et al. 1965 and Ash et al. 1967. - McVittie's 1970 complaint about Shapiro's secrecy. 4. Add to claim 2: Shapiro's published reply exists (Sci. Ethics 1(2), 10, 1985) and has not been read. It is the first record to obtain. 5. Add to claim 5: LaViolette (ApJ 1986) as the peer-reviewed tired-light support Wallace used, with the date note that supernova (1+z) tests came later. Add Arp's telescope-time denial as documented by Burbidge. Web-check the supernova (1+z) and CMB facts the verdict rests on. 6. Add to claim 6: - The Pulkovo dismissal and early-retirement accounts, as a witness account in one published source. - The Physics Today, PRL no-referee and Beckmann refusals. - The full NSF sentence ("fatal conceptual difficulties"). - Rapier 1971 and Wegener & Øhrstrøm 1975 as citing work. 7. New peripheral claim 7 (eclipse light-bending data were selected; Schmeidler 1922 plot): verdict after the Ch.3 deep read and one web check. Likely "Tested, not supported" if modern radio deflection measurements are confirmed [memory, high, to verify]. 8. Web-check the electron point-likeness limit behind claim 4(a).

B. primer.md - Evidence table row 1a: replace the arbiter verdict with the QE-revised one and remove the "Primer note for QE" (it is now resolved). - Remove the "New in this primer, for QE" box, or turn it into a one-line note that QE confirmed it. - §1: add Wallace's own statement that "much of relativity theory is true" and that he accepts E = mc². - §6 or §7: add the Moyer argument with its rebuttal, the Kotel'nikov residuals, Shapiro's 1985 published reply (unread), LaViolette, and the eclipse claim (one row). - §9 Further reading: add Shapiro, Sci. Ethics 1985; Essen's booklet (ref 55); LaViolette ApJ 301, 544.

C. facts.json - Fix how_read so it states that the digest covers only the opening of each chapter, and list the new deep reads. - Add key_claims for: the eclipse claim; the Moyer claim; the Kotel'nikov residuals; the 12:00 UT and Duncombe charges; the Pulkovo dismissals; the "much of relativity theory is true" position. - Add to summary: Shapiro's 1985 published reply (ref 72) and McVittie.

D. claims.md - Apply arbiter corrections 1-2: Shapiro's side is given (letters of Jan 1969, and a 1985 published reply exists); split claim 1 into the 1961 eight-point fit and the later residuals. - Update the source-gap note: Ch.4-6 are now deep-read and the 1969 paper has been read. - Add claim 7 (eclipse).

E. Tools (outside this project) textprep should page long plain-text files so Haiku digests do not truncate. Re-run Step 1a for this project after the fix if the owner wants a complete digest, although deep reads 1-6 cover the gap.

F. Owner Approve saving the 1969 PDF (and the 1971 paper) into research/files/.

---

Recheck (Step 4b)

Date: 2026-10-04. I re-read arbiter.md (QE revision A-G and the corrected claim list), primer.md, facts.json (also parsed as valid JSON) and claims.md against the remediation list in section 7.

No defect in the project's content remains. The two open items (E and F) are a tool change and an owner permission. They affect no verdict, and the project files disclose each of them.

Final: PASS