{"artifacts": [], "at": "2026-08-06T19:25:18Z", "attack_type": "A3_span_overreach", "attempt": "Tried to show the claim says more than its spans do. The claim attributes the negative energy density to Eulerian observers, and the quote itself does not name them — it says only 'this expression is everywhere negative'. Checked the extracted context_before: it reads 'uses the fact that the four-velocity of the Eulerian observers is given by: then one can show that these observers will see an energy density given by:', so the attribution is present in the source immediately before the span and is not imported by the claim. Also checked 'everywhere negative' against the quote verbatim rather than paraphrased. No overreach found.", "by": "straz", "claim_id": "C-0001", "evidence_ids": ["E-0001", "E-0002"], "id": "X-0001", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/eq_energy_density_sign.py"], "at": "2026-08-06T19:26:53Z", "attack_type": "A2_recomputation", "attempt": "Recomputed the sign claim from the equation instead of from prose about the equation. pdb-canon strips mathematics, so the corpus copy reads 'these observers will see an energy density given by:' followed by nothing; the expression was lifted from corpus/S-0002/body.tex and checked symbolically. It is -(1/8pi)(v_s^2 rho^2)/(4 r_c^2)(df/dr_s)^2: every varying factor appears squared, so the density is NON-POSITIVE everywhere, and it is exactly ZERO at rho = 0 (the axis of motion) and wherever df/dr_s = 0, which is both inside and outside the bubble wall for any shape function flat there. The claim said 'everywhere-negative'. That is the paper's own wording and the equation does not support it. The energy-condition violation itself stands, because a strictly negative density anywhere is enough to violate WEC and DEC.", "by": "straz", "claim_id": "C-0001", "evidence_ids": ["E-0001"], "id": "X-0002", "note": "Refutes the wording, not the physics. Corrected claim text recorded via update_field.", "outcome": "REFUTED", "resulting_claim_ids": ["C-0001"], "search_ids": []}
{"artifacts": [], "at": "2026-08-06T19:27:09Z", "attack_type": "A1_counter_source", "attempt": "Attacked the epistemic standing rather than the content. C-0002 was recorded as peer_reviewed_consensus. Its own source contradicts the label: the extracted context of E-0004 reads 'Three recent articles have argued for the existence of physically reasonable positive-energy warp drive configurations, either satisfying the weak energy condition (WEC), or minimizing violations thereof', and S-0001 exists precisely as a rebuttal to them. A position that a 2022 Physical Review D paper had to be written to defend is contested, not consensus. The NEC result itself survives on E-0003 and E-0008; the label overstates how settled the question is, and epistemic_status is exactly the field a reader uses to decide how hard to look for the other side.", "by": "straz", "claim_id": "C-0002", "evidence_ids": ["E-0003", "E-0004"], "id": "X-0003", "note": "Refutes the label, not the theorem.", "outcome": "REFUTED", "resulting_claim_ids": ["C-0002"], "search_ids": []}
{"artifacts": [], "at": "2026-08-06T19:27:09Z", "attack_type": "A4_unstated_assumption", "attempt": "Attacked the time index. C-0003 asserts what the peer-reviewed literature holds 'as of 2022' and rests entirely on one 2022 paper's characterisation of its own field. Two problems, neither resolvable from the corpus. First, a paper describing the consensus it argues against is an interested witness. Second, and worse for a claim written in 2026: nothing here covers 2022 to 2026, no search of the intervening literature has been run, so the claim's date is the date of its evidence rather than the date of the field. INSUFFICIENT_EVIDENCE. The queries that would resolve it: INSPIRE-HEP citations of S-0001 restricted to 2022 onward, plus an arXiv gr-qc search for warp drive AND (energy condition OR NEC) over the same window, both logged to searches.ndjson including a null result if it is null.", "by": "straz", "claim_id": "C-0003", "evidence_ids": ["E-0004"], "id": "X-0004", "outcome": "REFUTED", "resulting_claim_ids": ["C-0003"], "search_ids": ["Q-0001"]}
{"artifacts": ["compute/verify_energy_density.py"], "at": "2026-08-07T06:49:58Z", "attack_type": "A2_recomputation", "attempt": "Attacked the reading by trying to make a different one work. Built the extrinsic curvature from the paper's eq. 7 with a general shape function, contracted it through the Hamiltonian constraint with R^(3) = 0 (valid because eq. 1 gives flat spatial slices), and compared the result against the published expression under four candidate readings of r_c: the spherical radius r_s, the bubble radius R, the perpendicular distance rho, and the axial distance x. Only r_s gives a zero residual; the other three leave residuals that are explicitly printed rather than summarised. The derivation never uses the tanh profile, so the identification is not an artefact of the shape function.", "by": "straz", "claim_id": "C-0006", "evidence_ids": ["E-0012"], "id": "X-0005", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/quantum_inequality.py"], "at": "2026-08-07T07:05:36Z", "attack_type": "A2_recomputation", "attempt": "Tried to break the published figure by recomputing it from the paper's own inputs rather than quoting it. Took the wall bound from eq. 22 and the energy expression from eq. 26, evaluated the integral exactly rather than in the thin-wall approximation, and compared. Got 3.47e20 galaxy masses against the stated 3e20 — a ratio of 1.157, which is inside the order of magnitude the paper quotes it at. Also checked the prefactor: eq. 26 uses -1/12, which is the constant this programme derived independently from the Hamiltonian constraint before this paper was in the corpus, and the paper never says where its own constant comes from, so the agreement is not circular. The remaining difference between the two calculations is a factor of exactly 3/2 traced to the shape function: a linear ramp carries constant maximal slope across the wall, the tanh profile does not. Verified numerically as 1.5000, so it is a property of the profiles and not an error.", "by": "straz", "claim_id": "C-0008", "evidence_ids": ["E-0013", "E-0014", "E-0016"], "id": "X-0006", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/vdb_pricing.py"], "at": "2026-08-07T08:03:30Z", "attack_type": "A2_recomputation", "attempt": "Recomputed the construction region by region rather than trusting the paper. Region IV is pure Alcubierre at R=3e-15 m, so the existing closed form applies with no VDB-specific code: -6.25e29 kg against his -6.3e29, ratio 0.992. Region II was derived from scratch: the three-Ricci of the conformally flat static slice via explicit Christoffel summation, rho = R3/16pi, integrated with the proper volume element B^3 r^2 dr under two unrelated C2 profiles at his stated parameters. Total comes out ~1e30 kg for both profiles — solar-mass scale, confirming the ~32-order reduction. What failed to reproduce is his -1.4e30/+4.9e30 sign split: it is profile-dependent, and the half-cosine gives zero negative transition energy. The headline survives; the decomposition is an artefact of an unstated choice.", "by": "straz", "claim_id": "C-0010", "evidence_ids": ["E-0017"], "id": "X-0007", "outcome": "SURVIVED", "resulting_claim_ids": ["C-0013"], "search_ids": []}
{"artifacts": [], "at": "2026-08-07T08:53:48Z", "attack_type": "A1_counter_source", "attempt": "Attacked the significance of the opening rather than its text. C-0011 correctly states that the Santiago-Schuster-Visser theorem does not cover the Van Den Broeck spacetime. Searched the 2026 corpus for a direct check that fills the gap and found one: S-0003 certifies the Van Den Broeck walls as Hawking-Ellis Type IV and energy-condition violating for all observers — an explicit all-observer realization of the SSV conclusion on the very construction the theorem missed. The counter-source is a zero-citation arXiv preprint by a single author, unrefereed, so it cannot refute the claim on its own. INCONCLUSIVE pending an independent recomputation of S-0003's Type analysis on the VdB wall, which our machinery can run and which is now scheduled.", "by": "straz", "claim_id": "C-0011", "evidence_ids": ["E-0028"], "id": "X-0008", "outcome": "INCONCLUSIVE", "resulting_claim_ids": ["C-0016"], "search_ids": []}
{"artifacts": ["compute/numgr.py", "compute/hawking_ellis_scan.py"], "at": "2026-08-07T09:58:41Z", "attack_type": "A2_recomputation", "attempt": "Recomputed the certification that X-0008 could only mark INCONCLUSIVE. Built a numerical Einstein tensor from scratch (fourth-order stencils, comoving static form) and REFUSED to use it until it passed three gates: Minkowski flat to 1e-27, the Alcubierre Eulerian density against the analytic formula independently derived in C-0006 to 5e-9 relative, and Schwarzschild vacuum to noise. Then classified the mixed stress-energy across the wall at v_s = 0.5 on a cos-theta-uniform grid at two sigma*R scales: 99.7 and 90.6 percent Type IV, bracketing S-0003's 87-99. Checked velocity dependence (100/90.6/86.1 percent at v = 0.1/0.5/0.9) and the VdB B-transition (10/10 Type I, as a static diagonal region must be). What this refutes is not C-0011's text — the SSV theorem still does not cover Van Den Broeck — but its recorded significance as 'the most load-bearing opening in the corpus': the VdB f-wall is metric-identical to the Alcubierre wall, its stress-energy has no rest frame for 87-99 percent of the wall, and no observer exists for whom its energy density is even defined, let alone positive. The opening closes by direct computation rather than by theorem coverage.", "by": "straz", "claim_id": "C-0011", "evidence_ids": ["E-0028"], "id": "X-0009", "outcome": "REFUTED", "resulting_claim_ids": ["C-0018"], "search_ids": []}
{"artifacts": ["compute/thrust_profile.py", "tests/test_mission_sims.py"], "at": "2026-08-07T10:49:27Z", "attack_type": "A2_recomputation", "attempt": "Attacked the impulsive-burn assumption by recomputing the mission under continuous constant-proper-acceleration profiles (accelerate-flip-brake), with the trajectory integrated two independent ways — closed form and a blind RK4 that knows no hyperbolic functions — agreeing to ~1e-4. If finite thrust had beaten the impulsive numbers, C-0014's 'floors' framing would be refuted. It did not: at comparable trip time (0.03 g profile, 48.5 yr Earth) the annihilation bill is 3.34e6 kg against the impulsive 0.2c cruise's 1.25e6 kg at 43.9 yr, and every finite-thrust profile is costlier than the impulsive bound at its own duration. The frontier also confirms the D-T exclusion brutally: at 1 g the D-T mass ratio reaches 1e43 kg. The claim's numbers are floors, exactly as stated.", "by": "straz", "claim_id": "C-0014", "evidence_ids": [], "id": "X-0010", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/targeting_sim.py", "tests/test_mission_sims.py"], "at": "2026-08-07T10:49:27Z", "attack_type": "A2_recomputation", "attempt": "Attacked the ordering ('catalogue first, orbit second, relativity third') by building the full targeting computer and measuring what actually moves the aim point. With astrometry fetched live, the barycentre velocity mass-weighted to cancel the orbital wobble exactly, and the 3D Keplerian offset added, the total aim-vs-image correction is 188 au at 0.2c and 328 au at 0.1c — fully computable, hence not an error at all once modelled. The residual uncertainty is then the catalogue disagreement (4673 au of distance, C-0017's headline), which exceeds the entire computable correction by an order of magnitude and the relativistic clock terms by more. The claimed ordering survives with its margins quantified.", "by": "straz", "claim_id": "C-0017", "evidence_ids": [], "id": "X-0011", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/ab_orbit.py", "compute/targeting_sim.py", "tests/test_ab_orbit.py"], "at": "2026-08-07T10:49:27Z", "attack_type": "A2_recomputation", "attempt": "Attacked the 24-35 au figure from two directions. Robustness: the finite-difference velocity stencil in linear_extrapolation_error_au was varied across two orders of magnitude of step with no change in the reported errors, and the orbit solver's closure after one period is 2.4e-13 au, so the numbers are not integration artefacts. Consistency: the independent targeting simulator, which embeds the same orbit in full 3D with the barycentre moving, attributes 32.8 au to orbital motion over the 26.3-year image-age-plus-travel span of the 0.2c mission — inside the claim's 24-35 au band for comparable spans. Two artifacts, one number.", "by": "straz", "claim_id": "C-0019", "evidence_ids": [], "id": "X-0012", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/hawking_ellis_scan.py", "tests/test_hawking_ellis.py"], "at": "2026-08-07T10:49:27Z", "attack_type": "A2_recomputation", "attempt": "Attacked the Type IV certification as a possible numerical artefact. The classification fraction at v_s = 0.5, sigma*R = 50 is 91.1 percent and IDENTICAL across finite-difference steps of 5e-4, 1e-3 and 2e-3 — three settings, same 82/90 points — and the off-axis wall classification is stable across detector tolerances 1e-5 to 1e-7, with the imaginary pair at 6-8 percent of the eigenvalue scale, four orders above the noise floor the Minkowski and Schwarzschild gates established. A stencil artefact would move with the stencil; this does not.", "by": "straz", "claim_id": "C-0018", "evidence_ids": [], "id": "X-0013", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/attack_energy_axis.py", "tests/test_energy_budget.py"], "at": "2026-08-07T11:12:57Z", "attack_type": "A3_span_overreach", "attempt": "Attacked the scope: if the exponents in |E| ~ v^2 R^2 / d were properties of the tanh profile the claim is scoped to, the law would be decoration. Recomputed the radial integral under an unrelated C2 quintic smoothstep by blind midpoint summation: the fitted exponents are 2.00000 in R and -1.00000 in d — profile-generic to five decimals — while the constant is profile-dependent (the smoothstep wall costs 4.29x the tanh wall at equal width). The claim's self-restriction to tanh is therefore honest and conservative: no smooth profile tested is cheaper. No overreach found.", "by": "straz", "claim_id": "C-0004", "evidence_ids": [], "id": "X-0014", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/attack_energy_axis.py", "tests/test_energy_budget.py"], "at": "2026-08-07T11:12:57Z", "attack_type": "A4_unstated_assumption", "attempt": "Hunted for a convention smuggled under 'wall thickness'. A reader who takes d to mean the 10-90 transition width — the laboratory convention — rather than 1/sigma gets a wall 2.1972x thinner at the same words (factor found by bisection on the exact shape function) and a figure of 8.22e32 kg instead of 3.74e32: the third digit of the claim lives in the convention. The claim survives because the regime of C-0004, which it depends on, states 'wall thickness identified as 1/sigma' explicitly. Without that line this attack would have been a refutation.", "by": "straz", "claim_id": "C-0005", "evidence_ids": [], "id": "X-0015", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/attack_energy_axis.py", "tests/test_energy_budget.py"], "at": "2026-08-07T11:12:57Z", "attack_type": "A4_unstated_assumption", "attempt": "Attacked the coefficient: the 10^2 in Delta <= 10^2 v_b L_Planck is not a constant of nature. S-0007's own eq. 19 introduces the sampling-time parameter alpha with 0 < alpha << 1, and eq. 21 (E-0029, cited by hash) gives Delta <= (3/4)sqrt(3/pi) v_b/alpha^2 = 0.733/alpha^2 — the published 10^2 is alpha = 0.1. Swept alpha: at 0.01, a hundred-fold weaker sampling assumption, the bound loosens to ~7300 L_P and the 100 m bubble still requires 4.7e18 galaxy masses. The coefficient moves by two orders; the verdict moves by two orders and remains eighteen orders beyond absurd. The claim's wording ('at most about 10^2') plus its regime, which carries the sampling-time choice as an assumption, survive.", "by": "straz", "claim_id": "C-0007", "evidence_ids": ["E-0029"], "id": "X-0016", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/attack_energy_axis.py", "compute/quantum_inequality.py"], "at": "2026-08-07T11:12:57Z", "attack_type": "A3_span_overreach", "attempt": "Attacked the synthesis for overreach beyond its parents. Its strongest sentence — even with the quantum inequality set aside and a 1 m wall allowed, the requirement is of order a solar mass of negative energy — was stress-tested against both sensitivities found on this axis: under the 10-90 wall convention (2.20x) it becomes 1.2 solar masses, and under the costlier quintic profile 2.4 — 'of order a solar mass' holds across every tested reading. The sign argument (no energy source produces negative energy density) rests on C-0006's derived expression, which is GREEN. No overreach found.", "by": "straz", "claim_id": "C-0009", "evidence_ids": [], "id": "X-0017", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": [], "at": "2026-08-07T12:18:53Z", "attack_type": "A3_span_overreach", "attempt": "Independent auditor (wf_1946b5d7-102) attacked the scope word 'generic' and the proof's completeness. Verified against S-0001 offsets 8600-41524: 'generic' means the generic Natario class exactly as the regime lists; the paper's own 'delicate point' hedge at offset 39078 is resolved via the cited Pfenning-Ford analysis and the proof concluded at 41366 needing only localization, not zero ADM mass — so the regime is conservative. Cross-check: S-0003 (2026) treats the result as established. Residual accepted and applied: the regime's 'modified gravity does not rescue' was stronger than the source's 'very strong constraints' wording — softened by update_field.", "by": "redteam", "claim_id": "C-0002", "evidence_ids": [], "id": "X-0018", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": [], "at": "2026-08-07T12:18:53Z", "attack_type": "A3_span_overreach", "attempt": "Independent auditor attacked whether the claim overstates the SSV concession. Adjacency verified at offset 49737: the concession is SSV's own conclusion, not a strawman; the following sentence supports both remaining components. Suppression check: exactly one 'Broeck' occurrence in S-0001, so no contrary passage is hidden. The superlative 'the strongest recent no-go' is editorial but unopposed in the corpus. Residual applied to the note: SSV's adjacent caveat 'there are other problematic issues with both of these models' now quoted there.", "by": "redteam", "claim_id": "C-0011", "evidence_ids": [], "id": "X-0019", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": [], "at": "2026-08-07T12:18:53Z", "attack_type": "A3_span_overreach", "attempt": "Independent auditor read S-0011 in full, re-found E-0020/E-0021 at their offsets, and audited for hedge cherry-picking: the intro independently states the negative verdict without 'seem' (offset 986), local hedges are subsidiary, and the authorship/date premise checks mechanically (gr-qc/9905084 May 1999, gr-qc/9906050 June 1999, same author). Residual accepted and applied: the regime mis-scoped the source — S-0011's central analysis is performed on the ORIGINAL Alcubierre metric, not the B^2-modified one — fixed by update_field.", "by": "redteam", "claim_id": "C-0012", "evidence_ids": [], "id": "X-0020", "outcome": "SURVIVED", "resulting_claim_ids": [], "search_ids": []}
{"artifacts": ["compute/vdb_pricing.py", "tests/test_vdb_pricing.py"], "at": "2026-08-07T12:19:36Z", "attack_type": "A2_recomputation", "attempt": "Independent auditor found the premise false and the computation wrong. The interpolating profile IS stated in S-0010 (body.tex offset 12042, equation labelled B, 'let us choose n=80' at 12199); it was invisible to the tooling because the canonicaliser strips inline math and the pre-PDB-26 extractor missed macro-wrapped equations. Plugging the stated profile into the module's own unmodified integrator reproduces the published split to 1% (-1.380e30 / +4.865e30 vs -1.4e30 / +4.9e30; boundary w = 0.9815 vs 0.981) — verified by the lead before this record was written. The 'half-cosine counterexample' is C^1 only, outside the paper's stated twice-differentiable class, so it could not witness in-class profile dependence. The headline half of C-0013 (region IV to 1%, solar-mass totals, ~32-order reduction) survives and is carried by the superseding claim.", "by": "redteam", "claim_id": "C-0013", "evidence_ids": ["E-0032"], "id": "X-0021", "outcome": "REFUTED", "resulting_claim_ids": ["C-0020"], "search_ids": []}
{"artifacts": [], "at": "2026-08-07T12:19:36Z", "attack_type": "A3_span_overreach", "attempt": "Independent auditor re-found E-0023/24/25 at their offsets, then read one sentence further: S-0008 offset 26615 reads 'it may still be possible to satisfy the WEC in the presented configurations too, given sufficient modifications' — the claim's conjunct 'concedes that no modification removes those regions' quoted a concession out of the frame that immediately reopens it, the mirror image of the strawman failure the hedge scan catches. The scan only looked at context_before; the reopening lived in context_after, on the record all along. Control extended (reopening-frame scan), claim superseded with both sentences carried verbatim.", "by": "redteam", "claim_id": "C-0015", "evidence_ids": ["E-0030"], "id": "X-0022", "outcome": "REFUTED", "resulting_claim_ids": ["C-0021"], "search_ids": []}
{"artifacts": [], "at": "2026-08-07T12:19:36Z", "attack_type": "A3_span_overreach", "attempt": "Independent auditor reproduced the zero-citation check live (OpenAlex title.search per paper, all cited_by_count = 0, positive control S-0001 = 45) — that prong stands. The Type-IV conjunct was quoted without the source's own qualifier: S-0003 certifies Van Den Broeck Type IV only ABOVE a Type-I to Type-IV transition at v_s ~ 0.36 (E-0031; the numeric value survives only in main.tex as a macro — another instance of the inline-math-stripping hazard). Blanket 'for all observers' phrasing without the velocity qualifier overreached the source. Superseded with the qualifier carried.", "by": "redteam", "claim_id": "C-0016", "evidence_ids": ["E-0031"], "id": "X-0023", "outcome": "REFUTED", "resulting_claim_ids": ["C-0022"], "search_ids": []}