Progress and how to resume

Last updated 2026-10-05 (deliverable format change). The method is in prompt.txt; file types are in aaa-index-to-artifacts.md. The reader's site is site/ (open site/index.html; copy the whole folder to publish). The old Markdown index is kept at work/aaa-interesting-index.md.

Where things stand

Current focus: TernionAG and Morray-TH. Government is parked. The remaining 63 non-empty projects under Energy/Researchers are being reviewed in batches (Next steps item 5).

QE status of the 31 (last Final: line in qe.md): PASS: Ainslie, Babcock, Becquerel, Bedini, Blondlot, Brown, Builder, Compton, Correa, Distinti, DLHotson, Farnsworth, Frollov, Gardi, Goodenough, Gray, Heaviside, Maxwell. OPEN: Ampere (3), Aspden (1), Baranski (3), Bearden (1), Brush (1), Dollard (1), Dunning-Davis (1), Evans (2), Fedderson (2), Flynn (1), Gift (1), Government (1).

Records only the owner can obtain

The Investigator stops at registration, ordering and FOIA requests.

Add obtained documents to the project's source folder, then re-run Step 0 (tools/textprep.py) for that project and a review update.

User follow-ups (STEP F)

Morray-TH framing (owner, 2026-10-08): Moray's valve is not a rectifier in the AC-to-DC sense. Sea of Energy 4th ed. p.139: the stone is "a valve which allows energy waves to pass in only one direction at a time ... without necessarily changing from H.F. or A.C. to D.C."; p.116: oscillations in the first tank, reinforced by "harmonic coupling to the cosmic wave frequencies", "spill" over through the valve, which "prevents the return or feed-back of energy", into the next stage at a lower frequency, stage after stage; "syphon" (Beyond the Light Rays p.3; Valley State College 1962 p.1-2, where he says the transformation "obeys the laws of the conservation of energy"). The arbiter/primer claim 4 wording "rectifier-type detector" and the primer's crystal-rectifier background misframe this; correct at the next Morray-TH review update (one-way coupler between tank stages; both readings: a non-reciprocal element, or a threshold element passing peak pulses whose next tank rings at its own frequency).

Morray-TH, for the next review update (orchestrator check, 2026-10-08, of owner-recalled anomalies): spark on breaking the antenna lead is 6 in (4th ed. pp.126 and 180, the 1932 booklet), "about nine inches" (Fletcher demonstration, 4th ed. p.133), and 18 in (John Moray: 1980 USPA talk, _text/Media-Audio-John-E-Moray-1980..., line ~206; IECEC 1991 paper, citing Jensen's 60 W early device); no 17 in found. The 18 in figure is not yet in facts.json/arbiter/primer: add it, attributed to the son. Physics to weigh: Freau's empirical Tesla-coil rule L(in) = 1.7 sqrt(P in W) gives about 110 W for 18 in, 28 W for 9 in, 12 W for 6 in, consistent with the 60 W early device; and a long HF spark at the antenna terminal means the box drives the antenna to high voltage (Moray's "antenna capacitor", 4th ed. p.139), not what a passive receiving antenna gives. The series lamp was in the antenna lead (Jensen 1938; 5th ed. p.110 "in series with the device"), not the ground lead, and unscrewing it put the output lamp out.

Morray-TH and LeBon-G, for the next review updates (owner question, 2026-10-09): Moray's 'special helium ... spontaneously vanishing' (Sea of Energy 4th ed. printed p.80; 5th ed. ~p.232; Radiant Energy 1926 text line ~2953) is copied from Le Bon, Evolution of Matter (Legge 1907) pp.140-141. Moray changed 'must be a very special helium' to 'is a special helium!', dropped Le Bon's conclusion 'It therefore seems very difficult to admit the transformation of radium into helium', and dropped Legge's footnotes (p.115: Soddy, Radio-activity p.163, the helium 'disappearance' is the alpha particles 'burying themselves in the glass'; p.140: Cooke 1906 helium-cadmium compound). Physics: alpha embedding in the wall and discharge-tube clean-up of tiny samples [memory, high: web-check]; settled by Rutherford and Royds 1909 (thin-walled tube, helium spectrum grew in the outer vessel; recorded in Rutherford-Ernest arbiter line ~78). The folder's zz-Claude-Analysis-02 ('factually wrong, 4He is stable') misses that the observation was real. Add to the Disintegration of Matter theme card too.

Morray-TH, also for the next review update (orchestrator check 2026-10-08): the "vapor of water ... act only by their presence" catalysis passage (Radiant Energy "1926" scan pp.57-58; Sea of Energy 4th ed. ~line 3502) is Le Bon, The Evolution of Matter, nearly verbatim (research/pages/lebonmat.htm: acetylene + potassium hydride, carbonic acid -> formiate, dried NH3 + HCl, quinine sulphate); it is about chemical catalysis, not a tube ingredient. Moray's own tube fills on record are argon + mercury vapor (US 2,460,707; NASA contractor note: 29 argon-mercury tubes); no water-vapor fill found. "Jet" (Radiant Energy p.12) is an ion-propulsion aircraft idea; "turbine" (p.62) is a metaphor ("a high-speed oscillating energy turbine"; a generator is a "pump").

Morray-TH gap (orchestrator, 2026-10-08): the owner's aa-new multipactor papers (Kishek flyer, "Multipactor discharge on a dielectric", "Temporal evolution of multipactor discharge") are digested in facts.json but not weighed in arbiter/primer. Next update: weigh multipactor as the mechanism of the cold-cathode valve (secondary-emission conduction answers the Patent Office's "cathodes cannot emit"; radioactive coatings as seed electrons; voltage/frequency window = threshold, "spill over"; transient current overshoot = "surges"), not as an energy source (it absorbs RF power). Cross-link Farnsworth-PT (multipactor tube; mechanism Established, tube Never properly tested) and Dollard-Eric (multipactor as free energy: Not supported).

Tesla gap (orchestrator, 2026-10-08, owner question): Tesla's "stout copper bar" experiments (newer researchers call it the "Tesla hairpin" circuit; the word "hairpin" appears in none of the folder texts, so search the web under both names) are in the folder but not weighed in arbiter/primer: 1891 AIEE lecture (Inventions, Researches and Writings p.194-195, Figs. 128-129: lamps short-circuited by a bent thick bar light; nodes along the bar; "the thicker the copper bar ... the better") and 1893 On Light p.36-37 Fig. 19a (low-voltage, 50 V, 100 V lamps and an exhausted tube all in parallel on two stout bars, each at proper brightness by position). Next Tesla update: weigh them (bar inductive reactance and standing waves; a stiff low-impedance source "adapts to the load"; Lindemann's "hairpin circuit" and underwater-bulb replications, which also work at mains frequency in tap water).

TernionAG framing (owner, 2026-10-07): Ternion applies energy mainly by displacement current in a plate resonator (Kassner's "standing field"), and also by radiated RF through a wand-type applicator (the "radiating field"). A03 and Q05 are reworded; the arbiter and primer still say "RF-assisted electrolysis" for claim 11, to be corrected at the next TernionAG review update.

A dropped question (the full recipe of Moray's detector, first numbered Q05; the number was reused) was deleted unanswered at the owner's request. The follow-up research found that Radiant Energy pp.53-56 is copied, uncredited, from Le Bon; this is sites.md leads 29-33, for the next Morray-TH review update to weigh.

Next steps (owner's call)

0. To do 2026-10-10 (owner, 2026-10-09): add a Selected Theme "Flat vs. round earth". Needs a NEW PROJECT with new source content from both flat-earth authors and mainstream scientists. Include the NASA technical reports that flat-earth authors cite for their "flat, non-rotating earth" models (e.g. NASA RP-1207, 1988, a linear aircraft model; find the others they cite), and weigh what those reports actually assume and why. Then add the theme to work/themes/themes.json (selected: true, with scope and boundary), run the project through the workflow, write the theme card (STEP 5c) and re-render the site. Settle scope and the source list with the owner first. - 2026-10-10 (owner): scope settled. Theme scope: the flat-earth case from zetetic astronomy to the modern movement, each cited evidence item with flat vs globe prediction and measurement (sightlines/Bedford Level, horizon, rotation: Foucault, Michelson-Gale, ring-laser gyros, Sagnac; geodesy/gravity; the NASA reports). Boundary: rotation experiments link to Kelly-AG; hollow/concave earth (Koresh) mention only; medieval flat-earth myth as a history note. Project Researchers/Flat-Earth (source folder by gatherer agent; owner-approved write to that folder only); modern flat-earth sources = free primary only (author PDFs, FES wiki, Behind the Curve gyro/light tests); full altsci-review-batch.js run, normal Investigator cap; then theme (selected) in themes.json, STEP 5c card, render. - 2026-10-10: Flat-Earth source folder built (71 files, 100 MB; 28 primary incl. Rowbotham 1865/1881, Hampden 1869, Carpenter 1885, Blount Review, 18 TFES wiki pages, Dubay 200 proofs, two flat-earth 'NASA documents' lists; NASA-Reports/ 22 (16 NASA, 3 ARL, PTTI, TU Delft course summary) with citing source and quoted assumption in README; Related-Research/ 20 (Wallace 1905 chapter, Oldham 1901, Foucault 1851, Michelson-Gale 1925, Airy 1856, WGS84, IS-GPS-200N, Hirt 2010 refraction, Russell essay, Behind the Curve press reports, fact checks)). Gatherer found ARL-TR-1812/2683/PTTI quotes in the [R] list do not appear in those reports. Owner go 2026-10-10: full workflow + theme card; Rob Skiba added as a researcher to consider (gatherer adding transcripts/writings to Flat-Earth/Skiba-Rob/). Theme flat-vs-round-earth added to themes.json (selected); site render fails until the batch's curator rebuilds the theme map (expected). Skiba-Rob/ added (32 files: 15 auto-caption transcripts, 13 web pages, 4 secondary). Flat-Earth added to selected.json (project card in Selected). Full run launched wf_57820188-5a2. DONE 2026-10-10 (330 agents, 29.6M tokens, 2h20m): QE MINOR, remediation (6 deep reads, arbiter section 9), recheck OPEN 2 (stale physicist/contrarian figures: UA energy ratio 5e7 not 1e11, heat-budget remark, Naylor 1894 held, MGP 269) fixed by orchestrator -> PASS; card QE notes updated. Rankings run 16: Flat-Earth rank 13 (Selected card). Theme map 03; cards flat-vs-round-earth (new) + ether-drift, alternative-gravity, pathological-science, suppression-gatekeeping refreshed, all PASS after fix and independent recheck. Privacy STEP P: 0 H, 0 M, 10 L (Flat-Earth clean). Site 1827 pages, 101 projects, links OK.

1. Idea, not started (2026-10-09): images, D3 graphs, and a wiki-style site (landing page with search, one page per project/theme/person instead of the 2.5 MB single index). Findings, sources, licence rules, pilot and open decisions: work/2026-10-09-images-and-wiki-ideas.md. Owner to revisit.

2. 2026-10-08 owner go-ahead: "resolve Q09 and anything else outstanding" (chosen: pending review updates, Tesla loose ends, B5-B7; NOT the evidence rules on the old 31). Q09 researched (done, run wf_f40305b5-4ca); then the chain tools/workflows/history/altsci-outstanding-2026-10-08.js (run wf_aebf7f57-c8c FAILED: workflow nesting is limited to one level, so the chained scripts could not call altsci-followups/altsci-curate; only the Tesla reads, update and recheck ran. Now run one top-level workflow at a time: investigate reviewOnly wf_0b48f4a7-e47, then curate Tesla, then B5, B6, B7): Morray-TH + TernionAG review update (reviewOnly; TernionAG claim-11 wording note), Tesla loose ends (history/altsci-tesla-loose-ends.js), B5, B6, B7 (Tesla removed from B7).

3. Before the next full workflow run (owner, 2026-10-07): token-efficiency changes, then the Ives document update. - Measured on the Ives run (wf_134bcc69-3fd): about 95% of tokens are cached re-reads of growing contexts; the Investigator alone was about 40% (303 turns, 43 full-page get_page_text calls, context up to 300k). Cached reads cost the same on Opus 5.5 and Sonnet 5.5, so model choice is not the lever. - Change 1: Investigator and STEP F save each page's text to research/ files and grep them, keeping only excerpts in context (est. -25% of a full run). - Change 2: QE, primer, arbiter and reviewers read long review files by section (grep/headings) instead of whole files repeatedly; source text still read in full where the role requires it (est. -8%). - Dropped by owner: splitting prompt.txt by role (about 5%). Not using aisidekick memory for this (assessed: no gain for exact-identifier search, harms QE completeness, pollutes the recall hook). - APPLIED 2026-10-07 to prompt.txt (ROLES AT A GLANCE "Context economy"; STEP I "Pages to files", which STEP F inherits). First run under them: Tesla. - Then measure tokens per project against Ives (3.8M) and Whittaker (7.1M) and spot-check one card. - Ives document update: DONE 2026-10-07 (owner said run it before the efficiency changes; run wf_12bda5fc-932, QE MAJOR -> Steps 2-4 redone, recheck OPEN 2 minor, applied by orchestrator -> PASS; 4.3M tokens, 1h16m; rankings run 08; owner question recorded as Q13). Was: 31 public-domain Ives papers are in Researchers/Ives/Ives-papers/ and Step 0 has been re-run (32 new chunks); run altsci-qe.js with their slugs (from review/Ives/_text/manifest.json, docs under "Ives-papers/") after changes 1-2. The owner will obtain library copies (Internet Archive loan) of JOSA 42:540 (1952) and the two JOSA 43 (1953) notes and add them to Ives-papers/.

4. Cancelled 2026-10-07: cards for every ranked project. The owner rolled this back: only the top 20 and the projects in work/rankings/selected.json ("Selected scientists/inventors of interest": Heaviside, Maxwell, Eskridge-Amy, Whittaker-ET, Ives) get cards; everything else is one line under "Other ranked projects". Entry files for projects outside these remain in work/rankings/entries/ but are not shown.

5. Done 2026-10-04: Investigator run 3 + review update on TernionAG and Morray-TH (run wf_8782d1fe-acd; first run under the new STEP I priorities; runs 1-3 findings now weighed). QE update check found 7 defects (TernionAG 4, Morray-TH 3); the remediation round (run wf_34f12826-3b7, fixOnly) fixed all of them: both projects Final: PASS. HathiTrust needs a Chrome site permission for the Investigator.

6. Investigator run 4 when wanted: the rechecks skipped in run 3 (TernionAG: USPTO, SHAB, CAP, CourtListener, BAR, swisscollections, Bodleian, NARA, HathiTrust).
7. Apply the evidence rules (SOURCE HONESTY: no suppression, interested parties, absence of evidence, keep pertinent information; altsci-evidence-rules.js) to the 31 projects reviewed before the rules existed.
8. Government: DONE (found 2026-10-08: the zipped document's 4 pieces are digested and weighed in arbiter/facts/QE; this item was stale). Was: tools/textprep.py now unpacks .zip archives, but Government's Step 0 has not been re-run, so the zipped "EM effects on human body" document is still unread. Re-run Step 0 for Government, then a review update.
9. Remaining 63 projects (owner go-ahead 2026-10-04): running in sequential batches, smallest first, with altsci-review-batch.js. Kohler and Pais-Salvatore are empty and skipped. - B1 (run wf_e91186fe-df9, done: 9 PASS, Perreault-Bruce OPEN 1; 16.3M tokens, 1h56m): zzz-Unknown, SilverTooth, Pappas-PT, Wallace-Byran, Gunderson, Radiant_One_Youtube, Perreault-Bruce, NRGFromTheVac, Popa-Simil, SchusterAndHemsalech - B2 (run wf_3c3feef3-c3e, done: 8 PASS, Kelly-AG and Hieronymus-TG OPEN 1 each; 19.7M tokens, 2h02m): Znidarsic-Frank, Trichel, Malove-Eugene, Kelly-AG, van Vlaenederen Koen, Hieronymus-TG, Verlinde-Erik, Hopper-WJ, Lodge-O, Very-Frank - B3 (run wf_324f0b56-83d, done: 8 PASS, Sykes-Chris and Tolman-R OPEN 1 each; 25.7M tokens, 2h21m): Sigma-Ro, Larson-Tyson, Langmuir-Irving, Minderle-Thomas, Sykes-Chris, Haisch-Bernard, StClairJohn, Kennard, Tolman-R, Kilner-Walter - Stopped after B3 (owner, 2026-10-04) for the HTML format change; resumed with B4 on 2026-10-06. - 2026-10-06: STEP I time rule tightened (search to at least 9.5 min; "not reached (stopped early)" label). Applies from B5. - 2026-10-05: deliverables moved to HTML. STEP 5 now writes data (work/rankings/), STEP R (tools/render_site.py) builds site/; rankings widened to a top 20 with fixed-format entries. Rankings run 01 is the first in the new format. - B4 (run wf_95309568-7c5, done 2026-10-06: 4 PASS, 6 OPEN (Willams-Crookes 1, Mendeleev-D 2, Su-CC 3, Hilgenberg-OC 1, Linerman-Peter 2, Rahn-Otto 2); 29.8M tokens, 3h05m; rankings run 03): Poincare-Henri, Willams-Crookes, Mendeleev-D, Su-CC, HiIQ-Youtube, Hilgenberg-OC, Parallax, Peek-Frank, Linerman-Peter, Rahn-Otto - Eskridge-Amy (single-project run wf_19b4ea62-bac, done 2026-10-07: QE OPEN 2 minor, both applied by the orchestrator, now PASS; 4.2M tokens, 1h10m; rankings run 04, rank 12, card under Selected scientists/inventors of interest with Heaviside and Maxwell). - Whittaker-ET (single-project run wf_1e63011a-868, done 2026-10-07: QE OPEN 2 minor, both applied by the orchestrator, now PASS; 7.1M tokens, 1h25m; rankings run 05, card in Selected; one deep read (1951 History PDF pp.108-112) blocked by the content filter, its Laplace/Tisserand pages were read from page images in QE; added to Selected; 1951 revised History Vol 1 added to the source folder with a README; follow-up Q07 on gravity as a longitudinal standing wave). - Whittaker-ET Vol. 2 update (2026-10-07, owner approved the download): History Vol. 2 (1953, rexresearch1) added to the source folder; Step 0 re-run (Vol. 2 = 15 chunks, other slugs unchanged; old _text backed up in the session scratchpad); altsci-qe.js run wf_eecb115b-6e7 (digest Vol. 2, QE, remediation, Curator); done: QE MINOR, remediation applied, recheck OPEN 3; orchestrator applied the A07 update -> PASS; sites.md/Investigator queue (Copson 1928, Wiechert 1920, Lorentz 1900) deferred to the next Investigator run; Vol. 2 deep reads 2-5 not run (disclosed); 2.5M tokens, 32 min; rankings run 06. - Ives (Herbert E. Ives; new source folder by owner, sources + README collected by orchestrator 2026-10-07; added to Selected; single-project run wf_134bcc69-3fd done 2026-10-07: QE OPEN 4 minor, applied by the orchestrator, now PASS; 3.8M tokens, 1h13m; rankings run 07, card in Selected; Investigator saved 13 Ives JOSA papers in review/Ives/research/files/. Follow-up Q10 (obtain remaining papers legally) run wf_714d28f6-d02). Moray link: Ruderfer's appendix in Sea of Energy 5th ed. - Tesla (2026-10-08: owner chose the FULL run; altsci-review-batch.js run wf_2ab7f404-0fa done: STEP F answered Q11, Q12 first; QE MINOR -> remediation -> recheck OPEN 1 (stale 27-52% turbine band in claims.md/facts.json), fixed by orchestrator -> PASS; 3 deep reads (The Problem of Increasing Human Energy pp.1-33 and 15-33, Free Energy Secrets pp.137-156) blocked by the content filter, disclosed; 20.7M tokens incl. follow-ups, 2h20m: the context-economy rules did NOT bring it under the 9-12M estimate; rankings run 09, rank 5, card in Selected). Was ON HOLD 2026-10-07: added to Selected; Step 0 done (30 documents, 156 chunks; duplicates: Strange Life txt+pdf, Famous Scientific Illusions x2). Owner to choose full run (est. 9-12M tokens with the new context-economy rules) or trimmed run (~90-100 chunks: Tesla's writings and lectures, Ratzlaff, power-transmission/radiant-energy patents, Prodigal Genius, Valone, FBI files; skip Hayes turbine book, duplicates, Free Energy Secrets; est. 6-8M). Investigator focus: Earth transmission, Wardenclyffe, standing waves, Colorado Springs measurements, Corum analyses, Morgan-Tesla correspondence (the 'where do we put the meter' quote has no known primary source). First run under the context-economy rules: measure tokens. - 2026-10-08 review update (altsci-investigate reviewOnly wf_0b48f4a7-e47, 2.9M tokens): Morray-TH PASS (Le Bon-copy leads, Q09 Foster lead weighed; 'Radiant Energy 1926' printing now dated after 1945); TernionAG claim 11 reworded (standing field / displacement current, or radiated RF), recheck OPEN 1 (stale copies in sites.md leads 31, 34 and claims.md) fixed by orchestrator -> PASS; rankings run 10 (Tesla card refreshed too). - B5 DONE 2026-10-08 (run wf_4c1b25c3-e59, 448 agents, 47.3M tokens, 3.7 h): Lorentz-HA, Stiffler, RuderferM, Soddy, Lakhovsky George, Smith-Don, Morton-Wall, Handwerker-Roberto, Meyl-Konstantin. All QE MINOR; no verdict changes. Recheck PASS: Stiffler, RuderferM, Morton-Wall. OPEN (stale copies, Lakhovsky side-predictions, Smith-Don 3 Investigator files, Meyl Neutrino Power pp.133-222): orchestrator fixer agents 2026-10-08 -> ALL 9 PASS. No verdict changes. Corrections: Lakhovsky side-predictions weighed (all Never properly tested; soil claim is Claim 5); Meyl Neutrino Power now read in full (Moon factor 36 also on NP p.191 and Objektivitaetstheorie 1; Annalen referee attribution narrowed); Smith-Don Investigator files read (Utkin quote is 'is produced', Utkin's caption not Smith's words; Boulder Dam passage unsigned, not Smith's voice; Kelly reports ~200 replications claimed, 4 gamma-to-volt conversions). Rankings run 11; top 20 now includes Lakhovsky George (11), Meyl-Konstantin (16), Smith-Don (18). Site rendered: 1462 pages, 85 projects. - B6 DONE 2026-10-09 (run wf_8aebbe21-df7, 831 agents, 90.9M tokens, 5.6 h). STEP F answered Q15-Q20. All QE MINOR. Recheck PASS: Kelly-Patrick, Tousey-Sinclair, Steinmentz-Charles. OPEN (stale copies; Wheeler sites.md; Thompson reads 7-8 not carried to arbiter): 2 orchestrator fixer agents 2026-10-09 -> ALL 10 PASS, no verdict changes (Thompson claim 3 strong form stays Tested, not supported; grounds now sharp lines and Duane-Hunt cutoff, not slit diffraction). Site rendered: 1691 pages, 95 projects. Rankings run 12: Vassilatos-Gerry 6th, Kelly-Patrick 10th. - Q16 nuance: Sea of Energy p.139 also says the assembly is used 'for the rectification and amplification' of radio signals; record both uses in the Morray-TH update. Q20: crossing beams DO superpose (fringes hidden by fineness, phase drift, crossed polarization). - B7 DONE 2026-10-09 (run wf_b1a4a169-68d, 419 agents, 48.1M tokens, 2.8 h): Mills-Randal, Summerfield, Rutherford-Ernest, Meyers-Stan. All QE MINOR, no verdict changes. Rankings run 13: Mills-Randal 7th, Meyers-Stan 14th. Mills and Rutherford stale copies fixed by orchestrator -> PASS (UK appeal [2008] EWHC 2763 allowed and remitted, not dismissed; Rutherford gravity sentence already in 1904 1st ed. sec. 206). Fixer agents 2026-10-09 -> ALL 4 PASS: Summerfield deep reads 2-7 done (arbiter section 12; Moseley stays Established; half-integer 'later vindicated' corrected to Lande's; withdrawals list fixed); Meyers Bell chart is January 10, 1983 (1963 was OCR), No. 10 tap 11 / rain 16.1 / ocean 26.1K ppm (No. 11a copy still has OCR figures rain 36.17, ocean 2.1K). Site rendered: 1784 pages, 99 projects, run 13, links OK. Meyers-Stan has NO Step 1a digests (84 chunks): process gap, deep reads cover the load-bearing documents; owner to decide. - LeBon-G added to Selected (owner, 2026-10-08); moved from B7 into B6 so its card comes sooner. Morray-TH research/pages/ holds rexresearch Le Bon texts (A11, A12) worth pointing its Investigator to. - B6: LeBon-G, Wheeler-Ken, Larken, Vassilatos-Gerry, Tyndall-John, Kelly-Patrick, Jeans-JH, Tousey-Sinclair, Steinmentz-Charles, Thompson-JJ - B7: Mills-Randal, Summerfield, Rutherford-Ernest, Meyers-Stan (Tesla done separately; LeBon-G moved to B6)

• 2026-10-10 (owner): THEME FOLLOW-UPS added. followups.py add theme:ID "question" -> work/themes/followups/{ID}.json (global Q ids); shown as "User follow-ups" on the theme card; STEP F (altsci-followups.js, now keyed by target) answers them from the members' review files plus web research; notes in work/themes/research/, rendered to site/themes/research/. prompt.txt STEP F updated. Tested end to end with a throwaway question (removed). First real theme follow-up: Q21 (flat-vs-round-earth, first-principles experiment and assumption register), run wf_222fdf32-f20. A21 answered; corrections applied by orchestrator: plane offset is -k x 6.0 ft, not zero (arbiter claim 1, physicist, both cards); South Pole 2001 pendulum has data (24 h +/- 50 min; arbiter anecdote note). Open from A21: a clean low-budget southern-hemisphere pendulum run with controls does not exist yet. Owner judged A21 a fail (no clear step-by-step): follow-up answer format gained (a) 'procedure' box after Known/Inferred/Open linking to a separate step-by-step page with expected results (required for how-to/experiment questions; prompt.txt STEP F, followups.py, render_site.py, style.css) and (b) 'orchestrator_notes' rendered as Note(s). A21 now has its procedure page (work/themes/research/2026-10-10-followup-Q21-procedure.md) and five notes; the first-draft bracket moved from the question into Notes.

How to resume

• Workflow scripts are saved in tools/workflows/ (run with the Workflow tool via scriptPath, passing args):
• altsci-review-batch.js — full review of new projects in prompt.txt order: Steps 0, 1a, 1b, I, 2, 3, 4, 4b, then Curator (args {date, projects: [name | {name, src: folder or [sources]}]}; a plain name is a folder under Energy/Researchers; a list of sources makes a combined project; deep: true lifts the Step I 10-minute search cap).
• altsci-investigate.js — Investigator run on request plus the review update that weighs its findings (Steps I, 2-4b, Curator) (args {date, projects: [name | {name, focus}], reviewOnly?}; reviewOnly: true skips Step I and folds in findings not yet weighed; deep: true on a project lifts the 10-minute search cap when the owner asks for a deep dive).
• altsci-evidence-rules.js — revise arbiter/primer under the four evidence rules, then a compliance check in qe.md (args {date, projects: [name | {name, notes}]}).
• altsci-qe.js — retrofit for projects reviewed before digests existed: Step 1a digests + Step 4b QE + remediation + Curator (args {date, projects: [{name, src, docs: [[slug, chunkCount]]}]}, built from _text/manifest.json).
• altsci-followups.js — STEP F on its own: answers open owner follow-ups one at a time (Chrome, 20-minute search cap each), then renders site/ (args {date, ids?: ['Q03']}; ids for a speedy answer to specific questions). altsci-review-batch.js and altsci-investigate.js run it first, so any batch clears every open follow-up.
• altsci-curate.js — STEP 5 and STEP R on their own: ranking run file, rank entries (top-20 and selected cards), render site/ (args {date, changed: [names] | 'all', note?}). The three scripts above call it as their last step.
• altsci-qe-fix2.js — second remediation round from a qe.md recheck (args {projects: [{name, src}]}).
• history/ — one-off run records, superseded; do not re-run (altsci-ternionag.js, altsci-fix-and-research2.js, altsci-evidence-rules-2026-10-03.js).
• The Investigator needs the claude-in-chrome extension connected for sites that block scripts; only one agent uses Chrome at a time.
• Recording a follow-up: .venv/bin/python tools/followups.py add PROJECT "question" prints its id (Q01, ...); open ones show as pending in the card until the next run answers them.
• Work-in-process follow-ups: status "wip" (set in followups.json; the answer is the orchestrator's summary of an owner discussion, the full note in research/). STEP F skips them until the owner says to open one. Q09 (Morray-TH) was wip until 2026-10-08, when the owner removed the label; its answer is still the unresearched discussion summary.
• Follow-up summary (2026-10-08): each answered question has a two-sentence "summary" (what was asked; the takeaway), shown while the question is collapsed on the site; STEP F writes it, followups.py check enforces it.
• Python venv: .venv (sympy, numpy, markdown). Rebuild the site at any time with .venv/bin/python tools/render_site.py (no tokens).
• 2026-10-09 (owner): every card shows "Card updated YYYY-MM-DD HH:MM" (entry field updated now carries local date and time; render_site.py --check enforces it; existing entries backfilled from file times). Meyers-Stan added to Selected; Selected is kept and shown alphabetically.
• 2026-10-09 (owner): THEMES added (prompt.txt STEP 5c; work/themes/). Sections renamed Ranked Projects / Selected Projects; new Ranked Themes (20) and Selected Themes (owner: Aether Theory, Disintegration of Matter, Free Energy). Owner chose 16 of my candidates; I added Radiant energy from the environment, Wireless power and earth resonance, Alternative gravity theories, Permanent-magnet motors and generators. Aether and Free Energy are separate umbrella themes. New roles: Theme tagger, Theme curator, Theme writer, Theme QE. altsci-curate.js runs 5c after 5b (themes:false skips; themesOnly; tagProjects; themeCards). Prototype run wf_35483331-612: tag all 99, map 01, card for disintegration-of-matter only.
• 2026-10-09 themes prototype done (wf_35483331-612, 13 agents, 1.0M tokens): map 01 (all 99 projects tagged; ranked themes 1-3: pulsed over-unity, radiant collectors, gas discharge). Disintegration of Matter card: Theme QE OPEN 7 (card overstated or dropped arbiter verdicts) -> fixer -> PASS. Fixed: doubled file paths in altsci-curate.js agent prompts (also affected project entry writers); timeline CSS. Other 22 theme cards await owner go-ahead (themesOnly run, themeCards = the 22 ids).
• 2026-10-09: other 22 theme cards launched (wf_1d578904-b7b; reuses map 01 via themeMap arg; new automatic Theme fix pass for cards QE leaves OPEN).
• 2026-10-09 all 23 theme cards done (wf_1d578904-b7b, 67 agents, 9.2M tokens). Every first draft failed Theme QE (1-10 issues each, 106 in all: softened or dropped verdicts and caveats, misattribution, interested-party material stated flatly, arbiter recomputations shown as source figures); the Theme fix pass cleared all, so every QE file ends PASS. The PASS lines after fixes were written by the fixer itself, not an independent recheck. Recurring errors added to prompt.txt STEP 5c as a pre-finish checklist for writers.
• 2026-10-09 independent theme recheck (wf_790daab3-46a): 15 PASS, 8 OPEN with 10 small items (dropped split verdicts, flat attributions, missing qualifiers); fixed by orchestrator -> all 23 PASS. Two orchestrator slips, both corrected: (1) index error overwrote two gravity-control items, restored from work/rankings/archive/2026-10-09-run13/themes/ and re-edited by content match; (2) the recheck-only run also ran 5a/5b (script ordering bug, now fixed), producing ranking run 14 (no moves) and refreshing 10 project cards whose review files had changed (Eskridge-Amy, Lakhovsky George, Maxwell, Meyers-Stan, Meyl-Konstantin, Mills-Randal, Morray-TH, TernionAG, Tesla, Whittaker-ET); all pass --check. Site: 1785 pages, run 14, links OK.
• PARKED (owner, 2026-10-09, 'at least for now'): bound-charge aether as a support for longitudinal waves (longitudinal mode where eps(omega) = 0, D = 0; constraints: Lyddane-Sachs-Teller, static vs optical c, no vacuum dispersion; strongest in ground and ionosphere media). Not recorded as a follow-up; theme follow-ups not built. Options if revived: theme follow-up on scalar-waves, or Q21 on van Vlaenederen Koen.
• 2026-10-09 (owner): new project Naudin-JL. Source folder created at MEGA/.../Energy/Researchers/Naudin-JL (owner-requested write to the source area; only that folder), filled by a gatherer agent (JLN Labs pages, papers or patents, independent tests in Related-Research/, README.txt manifest). Added to Selected. Next: full review (altsci-review-batch.js, projects ['Naudin-JL']). Existing projects that mention Naudin: Bearden, Linerman-Peter, Stiffler.
• 2026-10-09: Naudin-JL source folder built (55 files, 12 MB; 37 primary, 17 secondary in Related-Research/; README.txt; live site jnaudin.free.fr, jlnlabs.online.fr; Gray and Mueller not on his sites). Full review launched wf_c1200250-8af.
• Owner 2026-10-09: Naudin investigation should be a survey (per research project: name, summary, significant outcomes; omit uninteresting ones). Run wf_c1200250-8af was already past the primer; after it finishes, add a 'Naudin's research projects at a glance' section to the primer (orchestrator fix), and re-render.
• 2026-10-09 (owner): STEP P privacy review added (prompt.txt; Privacy reviewer role, Opus). Runs last in altsci-curate.js after Render (privacy:false skips); reviews site/ for the owner's attributable personal opinions or information and for source-document disclosure; writes only aaa-privacy.md (never rendered). First review will come at the end of the Naudin-JL run.
• 2026-10-09 Naudin-JL review done (wf_c1200250-8af, 173 agents, 16.9M tokens; QE recheck PASS; rank 16 in run 15; Selected). Theme map 02: Naudin core in gravity-control, free-energy, magnet-motors, related in 5 more; 18 theme cards rewritten (8 for Naudin, 10 because member entries were refreshed in run 14); 5 left OPEN after independent recheck, fixed by an orchestrator agent -> all 23 PASS (one member note edited in map02 for pathological-science). Primer section 1a 'Naudin's research projects at a glance' added (11 projects) at owner request; a dangling 'section 5, step 6' reference fixed. Privacy step did not run in this batch (script loaded before STEP P was added); first STEP P review launched separately. Site: 1805 pages, 100 projects, run 15, links OK.
• 2026-10-09 first STEP P privacy review (aaa-privacy.md): 15 findings (5 H, 4 M, 6 L), owner approved fixes 1-4: (1) render_site.py safeguards: local-path scrubbing, private facts.json entries not listed, build fails on work/privacy.json terms; (2) source cleanup by two agents (attribution and personal details only; claims and verdicts unchanged); (3) Q08 long quotes trimmed; (4) prompt.txt owner-privacy rule for every role. aaa-progress.md cleaned by orchestrator. Re-run STEP P after render.
• 2026-10-09 second STEP P review: 0 H, 1 M, 12 L. Orchestrator cleared the owner-attribution residue (P1 run files' Q07 label, P2 TernionAG example.py, P3, P7, P8 Eskridge README private, P5/P10/P11 progress wording). Left for owner decision: D2 long quotations (Q02 1,237 chars, Perreault 758), D3 third-party e-mails, the Q09 question text (P4), MEGA folder-name mentions (P9), third-party sensitive details.
• 2026-10-09 (owner): criteria chips restyled. Short names A - Untested, B - Open question, C - Testable, D - Influential; three styles: letter only in tables (unchanged), 'A - Untested' on project and theme card headers, and the ranking key as chip + full definition (wraps inside the box; fixes the chip overflowing the border). render_site.py (CRITERIA_SHORT, crit_key), style.css, prompt.txt STEP 5a note. Site re-rendered, all run pages included.
• 2026-10-09 (owner): site title 'Interesting Rankings' renamed 'Alternative Science (Alt-Sci)' (index heading, page title, sidebar brand on every page).