Open: the † literature battery (OP10)

The literature battery is the planned check of every reference that the program’s papers cite from model memory, marked †, against the published source. A † citation was recalled by Claude rather than retrieved, so its authors, title, journal reference, or the statement it is cited for may be wrong. Checkpoint #11 lists the battery as open problem 10 of its §14 and calls itself “not for submission until the † audit (open item 10) completes”. Some of the works involved have been checked in other contexts, and one suspected error in #11’s list has been flagged, but no complete battery is on record. It is part of the first item of the current work queue.

Status: Open — [O] (#11 §14 item 10) — no completed battery is recorded in the repository; it gates any submission (#11 footer) and sits in item 1 of the current queue (root CLAUDE.md). It qualifies the † mark defined on Provenance tags and citation marks and is listed in Open problems and the queue.

The open item#

“Literature battery completion; every † reference source-checked [O].” (#11 §14 item 10). In #11’s header, [S] marks a reference “retrieved by live search during this project” and † one “cited from model memory — verify against the source before external use” (#11, header; the legend of its reference list says the same more briefly). The battery turns each † into a checked reference, or finds what is wrong with it.

It matters for Tier 2 of #11’s claim tiers: literature results are “correct at the level of well-worn results; exact numbers, conventions, and attributions are flagged † pending source checks” (#11 §2). A memory citation can be right about the physics and wrong about where it was published, who proved it, or what exactly the paper says.

Why it gates external use#

#11’s footer: “Working checkpoint — not for submission until the † audit (open item 10) completes.” HANDOFF.md §7: “All †-flagged citations must be source-verified before anything goes external.” The root CLAUDE.md gives † the same condition: “from model memory (verify against the source before any external use)”.

What is to be checked#

• #11. Entries [1]–[18] and [23] carry †; [19]–[22] and [24]–[29] carry [S]. By subject, the † entries are: Born’s two reciprocity papers ([1], [2]); Freidel–Leigh–Minic on Born reciprocity in string theory ([3]); curved momentum space, from Snyder to relative locality ([4]–[6]); the Born connection ([7]); spin-2 theory, from the multi-graviton no-go to Hinterbichler–Rosen ([8]–[12]); wave dark matter ([13]–[15]); Weyl semimetals ([16]); the Dirac–Schwinger–Zwanziger (DSZ) quantization and the electric–magnetic dualities ([17], [18]); and bigravity dark matter ([23]). The entries are collected on References; the document is the checkpoint.
• #11’s † citation without a list entry. #11 §11.2 credits the dual-superconductor picture of confinement, in which a condensate of magnetic monopoles squeezes the field between quarks into strings, to ’t Hooft and Mandelstam: “[T, ’t Hooft–Mandelstam†]”. The reference list has no entry for it, so the papers meant have to be identified before they can be checked.
• #13. All four references of conjugate special relativity carry †: Gibbons and Schiller on maximum force, Caianiello on maximal acceleration, and Born’s two papers, which are #11’s [1] and [2] (maximum-force conjectures).
• An unreferenced phrase (not a † entry; the wiki lists it here). #11 §11.3 says that choosing a duality frame “is choosing a Lagrangian polarization of charge space, in the literature’s own words”, without naming the source of the words (The symplectic charge lattice).
• An unreferenced limit (not a † entry; the wiki lists it here). #11 §9 excludes the lightest masses of the wave island by observation: “the light-mass escape is closed independently [15†, Lyman-α]”. [15] is Dalal and Kravtsov’s paper on ultra-faint dwarf galaxies, among the † entries above. The Lyman-α forest limit cited beside it, a bound from the absorption lines of intergalactic hydrogen in quasar spectra, has no reference in #11 or in the earlier statements of the same closure, calc note #4 (notes/debt3c-ufd-endpoint.md) and the v1 paper (#9 §7) (bounds on fuzzy dark matter).

What a check involves#

The blind referees of rounds 3 and 4 used a protocol that shows what a source check is. Its first step: “Verify every citation against INSPIRE, arXiv or the publisher, including that each one supports what it is cited for.” (review/round3-preregistration.md, ‘Protocol’). There are two parts. The reference must exist with the stated authors, title, journal, volume, page, year and preprint number; and the paper must say what the citing sentence uses it for. The round-3 referee did the second by searching the full text for each cited statement (review/round3/referee/audit-notes.md, Step 1).

Citation errors found so far#

The program’s reviews and audits have found errors of each kind the battery targets. All but the last were in other manuscripts; no review has audited #11’s list.

• A citation inverted. Round 1, on the soliton manuscript #14: “Preskill citation inverted (fixed)” (HANDOFF.md §6).
• Cited for more than it says. Round 3 found #14’s reference [4] “over-generalized”, Zwanziger’s quantization stated with 4π where the manuscript wrote 2π, and Schwinger named but uncited (ruling 12). The referee also claimed that Baumgart et al. do not contain an operator #14 cited them for; the adjudicator overturned this from a footnote, quoted verbatim, that the referee had missed (ruling 8). The referee had searched the full text by keyword and quoted the same page (review/round3/referee/audit-notes.md, Step 1); in the wiki’s reading, a check has to read the cited passage, not only search it for terms (Round-3 verdict, rulings and citation audit).
• Wrong details. Round 4, on the toy-model note: “A. Duncan, not N.”; a missing erratum; and, short of an error, one reference’s specific claim that “cannot be checked (paywalled)” (review/round4-adjudication.md, ‘Rulings’, Citations; Review round 4).
• One entry pairing two papers. #11’s [23]† gives the title and journal of “Bigravitational origin of dark matter” (Phys. Rev. D 94, 084055, which is arXiv:1604.08564) with arXiv:1607.03497, the number of the companion paper “Heavy spin-2 Dark Matter”. A verifier of the wiki build found this by live search; it is flag 20 in review/wiki-build-flags.md, a suspected error awaiting Paul’s ruling, recorded “For the † battery (OP10)” (bigravity dark matter).

A related case concerned a convention rather than a citation. Working on the toy models, Claude found that its recalled Fourier eigenvalues contradicted Grosse–Wulkenhaar’s stated invariance: “So a recalled convention is wrong somewhere.” A direct numerical integral settled it, “not a recalled eigenvalue” (notes/toy-lagrangian.md, ‘Chunk 1 results’, ‘Chunk 2 results’).

HANDOFF.md §1 summarized the record at the handoff as “Zero fabricated citations across all checks”, and the round-3 referee found all 23 of #14 v3’s references, “37 individual items”, resolving with matching metadata. The errors above are errors of detail and of support, not invented papers.

Progress so far#

The battery is open; what follows is partial progress, none of which closes it.

• One † upgraded. #11 v2.0 marked reference [21], the second data release of the Dark Energy Spectroscopic Instrument (DESI), with †. v2.1 retrieved it and marked it [S], and added references [26]–[29], “all INSPIRE-checked” (repository history, commit e1c1f67, the v2.1 amendment).
• Works checked through other manuscripts. Several works that #11 cites with † were also cited by #14 v3 or by the toy-model note, and the referees of rounds 3 and 4 resolved them there: Born 1938 and 1949 (#11’s [1], [2]; metadata confirmed, the primary text not read), Boulanger–Damour–Gualtieri–Henneaux ([8]), Hassan–Rosen ([10], [11]), Dirac 1931 and Zwanziger 1968 (two of the three works in [17]), Montonen–Olive (one of the two in [18]), and the two 2014 papers of Freidel, Leigh and Minic ([3]) (review/round3/referee/audit-notes.md and review/round4/referee/audit-notes.md, Step 1, matched against notes/knot-dark-matter-standalone.md and papers/reciprocity-toy-models.md, ‘References’). Those checks confirmed the bibliographic data and, except for Born’s papers, the support for what the other manuscripts cited each work for. Whether each supports what #11 cites it for remains part of the battery.
• The wiki build. The verifiers of this wiki checked pages against their sources, and at least one † reference was checked by live search, which produced flag 20. Two further flags concern [S] references, outside the battery: [24][S] has no authors and possibly a descriptive title (flag 9), and a live search found that the release #11 cites as [21][S] already covers three years of operations, which bears on the handoff’s label “three-year data” for the next release (flag 26).

Scope and place in the queue#

HANDOFF.md §8 item 6 defined the task: “Literature battery completion: every † reference in #11 and #13 source-checked; resolve the gravity-induced kinetic mixing question.” The second half was settled by round-3 Decision 1, “Kinetic mixing: strict sector separation”, now part of postulate P4 (Round-3 decisions and lessons; kinetic mixing).

The root CLAUDE.md places the rest in the first item of the current queue: “1. HANDOFF §8 items 5–7: the partially-massless comparison, the † literature battery, the phase-space-diameter note.” See Open: the partially-massless comparison and the phase-space diameter relation. The references of the partially-massless comparison carry [S] and are outside the battery.

History#

• v1. The v1 paper’s open problems ended with “Literature verifications: symmetric-bimetric Λ = 0; LPT coefficient calibration; Dalal–Kravtsov exact bound [O]” (#9 §10 item 8; LPT is Levkov–Panin–Tkachev), and its provenance note asked that “All citations flagged [T — verify] should be checked against sources before any submission” (#9 §11). The Λ = 0 comparison is now #11’s open problem 2. The quintessence fits #9 tagged “[T — refresh citations before publication]” (§8) are [S] in #11.
• v2.0. #11 introduced the two marks: the handoff summarizes it as having “25 references flagged † (memory — verify) vs [S] (live-searched)” (HANDOFF.md §3).
• Since the move. New citations are retrieved, not recalled. The literature report for #14 v3 pulled each record “with no citations from memory” (review/v3-worklog.md). The toy-model work’s prior-art searches retrieved every citation by live search, and round 4’s revisions required that “Every new citation is to be retrieved, not recalled”, which version 2 applied (“five new references, all retrieved”, notes/toy-lagrangian.md). The pre-registered citation forecasts both scored hits: “All 23 citations resolve” in round 3 and “All 12 citations resolve” in round 4 (pre-registration). Neither round covered #11 or #13 (Review round 3).

Sources#

• papers/broken-born-checkpoint-v2.md §14 item 10, footer, header, ‘References’, §2, §11.2, §11.3, §9 — the open item, the submission gate, the definitions of † and [S], the † and [S] entries, Tier 2, the † citation without a list entry, the unreferenced phrase, the unreferenced Lyman-α limit
• papers/conjugate-special-relativity.md ‘References’ — #13’s † entries
• Root CLAUDE.md ‘Since the handoff’, ‘Method’ — the current queue; the meaning of †
• HANDOFF.md §7, §8 item 6 — the rule for external use; the battery’s original scope; §1 — the record at the handoff; §3 — v2.0’s marks; §6 — the inverted Preskill citation
• review/round3-preregistration.md ‘Protocol’ — what a citation check involves
• review/round3-adjudication.md ‘Rulings’ 8 and 12, ‘Decisions’, ‘Pre-registration scored’ — citation findings, Decision 1, the round-3 forecast
• review/round4-adjudication.md ‘Rulings’, ‘Pre-registration scored’, ‘Revisions for v2 of the note’ — citation findings, the round-4 forecast, “retrieved, not recalled”
• review/round3/referee/audit-notes.md and review/round4/referee/audit-notes.md Step 1 — the referees’ citation audits
• notes/knot-dark-matter-standalone.md and papers/reciprocity-toy-models.md ‘References’ — the reference lists those audits covered
• Repository history, commit e1c1f67 — reference [21] upgraded to [S]
• review/wiki-build-flags.md items 9, 20, 26 — the wiki build’s citation findings
• review/v3-worklog.md preamble and notes/toy-lagrangian.md — retrieval since the move; the recalled convention
• notes/broken-born-program-v1.md §8, §10 item 8, §11 — the v1 form of the item; §7 — the Lyman-α limit named without a reference
• notes/debt3c-ufd-endpoint.md ‘The attractive endpoint: criticality and the drain’ — the Lyman-α limit’s first statement, without a reference