[zero-aska]

One file, notes/bh-lensing.md. These are the optional findings from #30's re-review, plus what #31's own review found.

• SDP.81 had two reported anomalies, and the note named one.
  • Hezaveh et al. (2016): a 6.9σ dark lump in the lens.
  • Inoue et al. (2016; MNRAS 457, 2936, doi:10.1093/mnras/stw168): a 2 to 3σ flux-ratio and astrometric anomaly, which they attribute to a dark clump along the line of sight. They argue against a sub-halo in the lens.
  • Stacey et al. (2025) find no evidence for either and conclude "these previous results are not robust". Hezaveh et al.'s lump was most likely spurious, from poorly fitted multipoles. Inoue et al.'s anomaly "can be explained" by limits of the earlier method and the lens's angular structure.
  • The summary, the section 5 table row and the section 5 paragraph now say this, and Inoue et al. joins the Sources.
  • Section 4 names all three models of SDP.81's anomalies. None includes a black hole mass, so the negative claim still holds.
• One value for one mass. The pseudo-Jaffe plus point mass point mass is 1.9×10⁵ M☉ in both places.
• Wording closer to Vegetti et al. The pseudo-Jaffe plus point mass model is one "which the data disfavour (Δln E = −30 against the best fit)", following their "significantly disfavored".
• "Bound", not "weigh". In section 4, central images "bound" lens-galaxy black holes, as section 5 already says.

Reviews. Three independent reviews, none by the drafting agent:

• 1b84a94: no blocking problems (verdict merge). Four non-blocking findings in text this PR had added, all fixed in 508e41d against Inoue et al.'s and Stacey et al.'s own text.
• 508e41d: three wording fixes asked for, made in 023c60d.
• 023c60d: no problems, blocking or non-blocking; verdict merge.

Checked: npm test passes; node notes/bh-lensing.mjs runs; the new DOI resolves and matches Crossref; CI green on 023c60d (Node 18, 20, 22).

🤖 Generated with Claude Code

https://claude.ai/code/session_01XfgH6WcwTueUFZKb6fS2s2

[zero-aska]

This is an independent review of #31 at 1b84a94. The drafting session spawned separate agent instances that had not drafted the branch and did not get its drafting context: one reviewer, plus one refuter for each finding (two for each blocking one).

What was verified

• Stacey et al. 2025 (arXiv:2508.02776, main.tex): the abstract and lines 421 and 452 support the note's "not robust" (lines 24 and 208). They also support "most likely spurious, from poorly fitted multipoles" (lines 200 and 208-209). Line 452 attributes the H16 result to "a poorly fitting multipole component" and concludes "these previous results are not robust".
• Vegetti et al. 2026 (arXiv:2601.02466, ms.tex line 362, Table 1): PJ+PM has Δln E = −30 and M_PM = 1.93 ± 0.36 ×10⁵ M☉. Line 299 says it is "significantly disfavored", and the best fit is UD+PM (line 294). So lines 162-163 are correct.
• Zhang and Yu 2026 (arXiv:2606.12909, lines 99 and 194): they adopt M_BH = 1.9×10⁵ M☉ with M_200 = 10¹¹ M☉, and the SIDM snapshot best matches PJ+PM (line 143). So line 165 now agrees with line 163.
• New Sources entry (line 280): it matches Crossref and the arXiv API field by field. That covers authors, title, MNRAS 457(3) 2936-2950, 2016, DOI 10.1093/mnras/stw168 (which resolves) and arXiv 1510.00150. It is in the right alphabetical place.
• Inoue et al.'s model has no black hole mass. It is an SIE plus external shear plus a compensated clump, and the lens AGN (image G) is used only as a position constraint (prepri2.tex lines 248, 356-357 and 1005-1013). Section 4's "no such study" claim therefore still holds.
• Agreement within the note: CORRECTIONS.md item 16 (lines 110-117) agrees with the new summary and section 5. The "bound" wording in section 4 (line 170) agrees with section 5 (line 213).
• Diff hygiene: only notes/bh-lensing.md changed (25+/22-). Every changed line is 115 characters or fewer. Line 25 does not parse as a list marker. There are no tabs or trailing spaces, and no stray pipes in the table.
• Checks: npm test reports "all 4 check file(s) pass", and node notes/bh-lensing.mjs exits 0.

Blocking findings

None.

Non-blocking findings

• F1, notes/bh-lensing.md:147. Section 4 still says "SDP.81's lump (Hezaveh et al. 2016; Stacey et al. 2025)". That is one lump and two papers, while lines 24, 200 and 208 now describe two reported lumps that include Inoue et al. 2016. The PR left this line alone, but it no longer matches the lines the PR changed. Inoue et al.'s model has no black hole mass (see above), so the fix is safe. Suggested wording: "the models of SDP.81's lumps (Hezaveh et al. 2016; Inoue et al. 2016; Stacey et al. 2025)". The PR body's caveat "I have not checked Inoue et al.'s model myself" can then be dropped. Upheld by its refuter.
• F2, notes/bh-lensing.md:24, 200, 207-209. "A second reported lump" and "two reported dark lumps" put Inoue et al. beside Hezaveh et al.'s 6.9σ detection in the lens. Inoue et al. actually report a 2-3σ flux-ratio and astrometric anomaly (prepri2.tex 101-102, 642, 1213). They attribute it to a compensated clump along the line of sight, possibly at a different redshift from the lens (line 1226), and argue against a sub-halo in the lens galaxy (lines 645-651 and 1233-1245). Nothing in the note is wrong, and the conclusion holds either way, but these points are missing. Suggested wording: "a line-of-sight dark clump suggested at 2-3σ (Inoue et al. 2016)". Upheld by its refuter.
• F3, notes/bh-lensing.md:209. "Inoue et al.'s explained by limits of the earlier method and the lens's angular structure" states as settled what Stacey et al. offer as a possibility. Their line 452 says "can be explained", and line 423 says "it is very plausible". Suggested wording: "Inoue et al.'s can be explained by ...". Upheld by its refuter.
• F4, notes/bh-lensing.md:162-163. The edit made the sentence ungrammatical. In "which the data disfavour (...) and has a 1.9×10⁵ M☉ point mass", "which" is the object of the first clause but has to serve as the subject of the second. The old "which fits less well ... and has" was correct. The facts are correct. Suggested wording: "which the data disfavour (...) and which has ...", or "a model the data disfavour (...), with a 1.9×10⁵ M☉ point mass". Upheld by its refuter.
• Refuted: F5, notes/bh-lensing.md:229-230. The finding said section 6's "anomaly is now disputed" contradicts the "not robust" wording. Its refuter rejected it: the PR did not touch section 6, "disputed" is a fair summary of Stacey et al., and #30's re-review already treated this wording as a judgment call.

Verdict: merge

---

Generated by Claude Code

[zero-aska]

Re-review of fix commit 508e41d. It was done by separate agent instances that did not draft the branch; the drafting session spawned them without its drafting context.

Findings from the earlier review

• F1: fixed. notes/bh-lensing.md:147 now says the models of SDP.81's anomalies (Hezaveh et al. 2016; Inoue et al. 2016; Stacey et al. 2025) do not include its black hole. Inoue et al.'s model is SIE plus external shear plus a compensated spherical clump, and image G appears only as a position constraint. Neither Inoue et al. nor Stacey et al. mentions a black hole, and the PR body no longer says "I have not checked Inoue et al.'s model myself".
• F2: fixed. notes/bh-lensing.md:200 and 208-209 now call Inoue et al.'s anomaly a 2–3σ line-of-sight dark clump and set it against Hezaveh et al.'s lump in the lens. This matches Inoue et al.: an anomaly at 2σ to 3σ, attributed to intergalactic dark structures along the line of sight, modelled as a homogeneous spherical clump.
• F3: fixed. notes/bh-lensing.md:209 now says the clump "can be explained by limits of the earlier method and the lens's angular structure". This matches Stacey et al. (arXiv:2508.02776, main.tex line 452), and the "most likely spurious, from poorly fitted multipoles" attribution still matches their abstract.
• F4: fixed. notes/bh-lensing.md:161-163 now use an appositive ("a model the data disfavour (Δln E = −30 against the best fit), with a 1.9×10⁵ M☉ point mass"), which is grammatical, and the facts are unchanged. node tests/run.mjs passes, node notes/bh-lensing.mjs exits 0, and apart from reflow the diff touches only lines 24, 147, 162-163, 200 and 208-209.

New problems

No blocking problems. Three non-blocking ones:

• notes/bh-lensing.md:24, 207-208 (non-blocking): "both" now has two meanings in one sentence: "SDP.81 had both until Stacey et al. (2025) found both reported anomalies not robust". The first "both" means an anomaly plus a black hole bound. The second means two anomalies, which nothing earlier mentions; the old text said "its two reported dark lumps". Suggested: "found its two reported anomalies not robust" in both places.
• notes/bh-lensing.md:162-163 (non-blocking): two "with" phrases that do different jobs sit side by side across the colon ("..., with a 1.9×10⁵ M☉ point mass: with a self-interacting halo ..."). The first describes the model, and the later ones complete "match ... with". A reader can attach the later ones to the point mass instead. An optional rewording: "which the data disfavour (...) and which has a 1.9×10⁵ M☉ point mass:".
• notes/bh-lensing.md:200, 209 (non-blocking): "2–3σ" writes a range with an en dash. The rest of the repo's Markdown prose writes ranges as "X to Y" (for example "18 to 63 pc" and "5 to 12%"). Suggested: "2 to 3σ".

Verdict: merge

---

Generated by Claude Code

[zero-aska]

This is a confirmation of commit 023c60d by a separate agent instance that did not draft the branch, spawned by the drafting session without its drafting context.

What was checked

• Scope. git diff --word-diff=porcelain 508e41d 023c60d touches only notes/bh-lensing.md (12+/12-). The only word changes are: "both" → "its two" (lines 24 and 208); "a model" → "which" and "fit), with" → "fit) and which has" (lines 162-163); "2–3σ" → "2 to 3σ" (lines 200 and 209). The rest is reflow of lines 24-26 and 162-167. No other file changed, and the commit message lists the same three fixes.
• "its two reported anomalies" (lines 24, 208). "its" can only mean SDP.81, and each "both" in these sentences now has a single meaning: an anomaly plus a black hole bound. The phrase agrees with section 4 at line 147 ("the models of SDP.81's anomalies (Hezaveh et al. 2016; Inoue et al. 2016; Stacey et al. 2025)"). It also agrees with the table row and lines 208-210, which name the two anomalies. It agrees with CORRECTIONS.md item 16 too, which says Stacey et al. find "no evidence for either previously reported lump" and that "no lens in the note now has both a small-scale anomaly and a bound on its black hole".
• "which the data disfavour (Δln E = −30 against the best fit) and which has a 1.9×10⁵ M☉ point mass:" (lines 162-163). These are two joined non-restrictive relative clauses, with "which" repeated: it is the object of the first and the subject of the second. The sentence is grammatical. The relative clause now ends at the colon, so the "with ... or with ..." phrases after it attach to "match", not to the point mass. The facts are unchanged; the first review checked Δln E = −30 and the 1.9×10⁵ M☉ point mass against Vegetti et al. 2026, Table 1.
• "2 to 3σ" (lines 200, 209). This matches the repo's "X to Y" range style ("18 to 63 pc", "5 to 12%"). On line 209 the range comes before a noun, unhyphenated, as the note's other quantity modifiers are ("10⁶ M☉ dark lump", "36 milliarcsecond core"). No "2–3σ", "both reported" or "a model the data" remains anywhere in the repo.
• Markdown and line length. The SDP.81 table row (line 200) still has 5 pipes and four cells. No changed line starts with a list marker: line 25 starts "5). ", and a marker needs a space right after ")". No added line has tabs or trailing spaces. Counted in Unicode code points, the longest changed prose lines are 115 characters (lines 165 and 209), and the rest are 63 to 114.
• Checks. node tests/run.mjs ends with "all 4 check file(s) pass" and exits 0. node notes/bh-lensing.mjs also exits 0.

New problems

None, blocking or non-blocking.

Verdict: merge

---

Generated by Claude Code