{"id":"790d54ac-99b1-439a-ae29-7162b14977d8","arxiv_id":"2605.24512","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":0,"one_line_summary":"A fourth-order null-surface computation is presented as completing the tree-level 2-to-2 graviton amplitude and establishing all-loop UV finiteness, but the verification is left conditional and one partial-fraction identity is algebraically wrong.","lead":"This paper extends the null-surface formulation of gravitational scattering to fourth order and claims that this completes the tree-level graviton amplitude and shows ultraviolet finiteness at every loop order. The promised verification is incomplete: the identities that would prove the amplitude is reproduced are left as conditions, and a key algebraic check does not close.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Advanced-cone corrections to σ+4 are identified but never computed, so M^(24) and the claimed tree-level amplitude are not established.","rationale":"The reader's REJECT is correct, and my reading keeps that verdict unchanged. The most load-bearing problem is not merely that a consistency identity is unverified; it is that the paper's own corrected fourth-order kernel is incomplete. The advanced-cone corrections, advertised as the main new contribution, are written only schematically: terms like δ¯K^(4),adv_(II),(2,2) in Eq. (102) are never given as explicit functions of the momenta, and the manuscript states that their explicit computation 'remains.' Because Theorem V.2 reduces the tree-level amplitude to exactly three terms, an unevaluated M^(24) means Eq. (39) is not a result of this paper. This is a missing derivation, not a presentation issue. The reader's identified assumption about cancellation of the scattering-correction cut terms in Appendix A Step 2b is important and is also not rigorously proved, but even granting that cancellation, the advanced corrections are uncomputed. The partial-fraction check in Eq. (114) is algebraically wrong—for a simple test kinematics it does not reduce to the claimed identity—so the only explicit consistency check offered in the text is invalid. There is no machine-checked proof or reproducible code to offset these gaps, and the paper depends on unpublished companion papers for the inputs δσ+2 and δσ+3. The advertised central claims are therefore not demonstrated in the text provided.","tokens_in":36873,"tokens_out":8124,"duration_ms":89746,"concrete_test":"Compute δK^(4),adv_(II),(2,2) and δK^(4),adv_(III),(2,2) by substituting the output δσ+2 from Ref. [5] and δσ+3 from Ref. [6] into Eqs. (19) and (26), performing the affine-parameter integral and the S^2 Green-function convolution explicitly, and adding the results to the retarded kernels in Eq. (87). Then evaluate M_t and M_u via Eqs. (84)–(85), and check whether the corrected numerators satisfy the on-shell identities (101)/(107) and the corrected partial-fraction decomposition of −κ²s³/(4tu). If the δK terms cannot be evaluated from the given formulas, or the identities fail, the tree-level claim is not demonstrated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim, Eq. (39), requires the complete fourth-order kernel (29), including the advanced-cone corrections δK^(4),adv. But the paper never computes these corrections. Section II E–F gives only their general structure: Eq. (28) contains an unspecified kernel F^(δ); Eq. (87) defines the channel-III kernel in terms of an unevaluated δK^(4),adv_(III),(2,2); and Eq. (102) writes δM_t in terms of δ¯K^(4),adv_(II),(2,2) without substituting δσ+2, δσ+3. The paper concedes this explicitly: the Remark on corrected kernels says the operators δa_out4,± are 'complete once these corrections are computed explicitly,' and Sec. V.E states 'only the explicit computation of the advanced cone corrections δM_t,u remains to fully determine the numerical values of N_t,u+M_t,u.' After Eq. (101) it adds that δM_t 'must be verified against the DeWitt amplitude separately.' Since M^(24) is one of exactly three tree-level pieces in Theorem V.2, the advertised equality M_tree = −κ²s³/(4tu) is conditional on a computation absent from the text. This is the load-bearing gap. The reader's cut-cancellation assumption in Appendix A Step 2b is a related secondary unproven step, and Eq. (114)'s partial-fraction algebra is also inconsistent, but the missing δK computation alone invalidates the present derivation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This third paper in a trilogy claims to complete the null-surface formulation (NSF) computation of the tree-level 2→2 graviton scattering amplitude by determining the fourth-order Bondi shear σ_4^+. The central assertions are: (i) the tree amplitude is M_tree = M^(22)+M^(33)+M^(24) = −κ² s³/(4tu), with M^(24) derived from the new fourth-order computation; (ii) the NSF expansion is UV-finite at all loop orders, with the L-loop integrand scaling as dq/q^{4L} (Theorems V.2 and V.4); and (iii) a generalized Jordan–Pauli relation determines σ_n^+ recursively from the free incoming data, with new advanced-cone corrections first appearing at n=4. The paper provides a detailed channel classification, explicit retarded-cone kernel assembly in Appendices B–G, and consistency identities (101) and (107) that the form factors must satisfy.","tokens_in":37247,"tokens_out":4752,"duration_ms":53858,"significance":"If the advertised equality with the Weinberg–DeWitt amplitude and the all-order UV finiteness were established, this would be a significant result: an on-shell, manifestly UV-finite formulation of graviton scattering without off-shell propagators would be a major alternative to covariant perturbative quantum gravity. The paper contains useful structural ideas — the recursive factor formula, the generalized Jordan–Pauli decomposition, the loop-counting classification, and the retarded-cone kernel assembly. However, the central result is conditional: the advanced-cone corrections that are claimed to complete σ_4^+ are never explicitly computed, and the paper itself repeatedly states that this computation remains. In addition, Eq. (114) contains an algebraic inconsistency in the DeWitt-amplitude check. The significance is therefore prospective rather than demonstrated.","major_comments":[{"comment":"The advanced-cone corrections δK^{(4),adv}_{(1,3)} and δK^{(4),adv}_{(2,2)} are introduced in Eq. (29) and claimed to be part of the complete fourth-order kernel, but they are never explicitly evaluated. After Eq. (87) the text says the correction 'needs to be added to complete M^{(24)}'; after Eq. (101) it says the advanced correction 'must be verified against the DeWitt amplitude separately'; and Sec. V.E states that 'only the explicit computation of the advanced cone corrections δM_{t,u} remains to fully determine the numerical values of N_{t,u}+M_{t,u}.' Since M^{(24)} is one of the three tree-level pieces in Eq. (39), the advertised equality M_tree = −κ² s³/(4tu) is conditional on a computation that is absent from the manuscript. This is a load-bearing gap, not a presentation issue.","section":"Secs. IIE–IIF, Eq. (29), Eq. (87), Eq. (102), Sec. V.E"},{"comment":"The derivation of the master equation depends on the assertion that all scattering-correction cut terms, Σ_{j=2}^{n−1} Z^+_{n,cut}|_{δσ^+_j}, cancel as boundary/total-u-derivative terms. The text provides only a verbal explanation ('they appear as ˙σ^+·Z^+_j, which is a total u-derivative') and concludes that they cancel. No explicit computation of these terms at n=4 is shown. If this cancellation is not exact, σ_4^+ receives additional contributions not present in Eq. (18), and the computed kernels and the tree-level identification are incomplete. This is a second unproven step in the central derivation.","section":"Appendix A, Step 2b, Eq. (A15)/Eq. (11)"},{"comment":"The partial-fraction identity used to confirm the DeWitt amplitude is algebraically false as written. With the numerators of Eq. (113), the sum P_s/s + (N_t+M_t)/t + (N_u+M_u)/u equals −κ²/4 [s³/(tu)+s³/t²+s³/u²], not −κ²/4 s³/(tu). The passage to the final expression requires the extra terms s³/t²+s³/u² to vanish, which they do not for generic t,u. Moreover, the t-channel residue quoted in Eq. (90) is of order s²/t (after using u≈−s), whereas Eq. (113) assigns N_t+M_t ~ s³/t², which is more singular. The DeWitt-amplitude verification in Sec. V.H therefore fails as written.","section":"Eq. (114), Sec. V.H"}],"minor_comments":[{"comment":"The text refers to 'Case 2 of Sec. 15.4' and 'Sec. 15.4' which appears to be a nonexistent section number; the intended reference is likely Sec. V.D.","section":"Sec. V.D"},{"comment":"The phrase 'absent in the original computation of this paper' is confusing. It should clarify that the reference is to the retarded-cone computation alone, not to an earlier version of the same paper.","section":"Eq. (29)"},{"comment":"The loop-counting formula L=(n1+n2−6)/2 gives L=−1 for the (δa_2^out)^2 term, which is then treated as a special tree-level case. The theorem is stated with 'if and only if n1+n2 ∈ {4,6}', but the formula itself does not cover the case n1+n2=4. The presentation should state the exceptional case explicitly rather than presenting Eq. (51) as universal.","section":"Theorem V.2, Eq. (51)"},{"comment":"The notation σ_j^+[σ^-] = −σ̄_j^-(ζ̂) + δσ_j^+[σ^-] is introduced, but the retarded mirror term σ̄_j^-(ζ̂) is not defined until later. A short definition or reference would improve readability.","section":"Sec. II.A and Eq. (4)"}],"recommendation":"reject","confidential_remarks":"I concur with the reader's rejection. The manuscript's central claim is explicitly conditional on advanced-cone corrections that are never computed, and Eq. (114) contains an algebraic error that undermines the stated DeWitt-amplitude check. The paper also relies heavily on unpublished companion papers (Refs. [1,2,5,6]) for inputs that are described as 'known'. These issues are load-bearing and cannot be resolved by minor editorial changes. The structural ideas may be worth pursuing, but the present manuscript does not establish the advertised results."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nYou should know two things about this one. First, the paper does contain something new: the generalized Jordan–Pauli relation for n≥4, which says the standard cancellation between advanced and retarded cones breaks down and leaves explicit corrections δK^(4),adv built from the lower-order scattering fields σ+2, σ+3. The loop-counting formula L=(n1+n2−6)/2 and the channel decomposition of σ+4 are also genuine. Second, the central claim—that σ+4 completes the tree-level 2→2 amplitude to M_tree = −κ²s³/(4tu)—is not demonstrated in the text. The advanced-cone corrections are identified but never computed. The paper says so itself: after Eq. (36) the operators are \"complete once these corrections are computed explicitly,\" and Sec. V.E states that \"only the explicit computation of the advanced cone corrections δM_t,u remains.\" Since M^(24) is one of the three tree-level pieces, the equality is conditional on a computation that is absent.\n\nThe final consistency check also has a concrete error. Eq. (114) claims s/(tu) + 1/t² + 1/u² = s³/(tu) after using s+t+u=0. That identity is false; test s=1, t=1, u=−2. So the partial-fraction match to the DeWitt amplitude is not established. There are weaker spots too: the cut-cancellation in Appendix A Step 2b is asserted as a boundary term rather than shown; the all-order UV-finiteness theorem V.4 is a power-counting sketch that assumes the kernel scaling K^(n)∼ω_ext/q^{n−2} rather than proving it at each order; and the main inputs σ+2, σ+3 come from two unpublished companion papers. Several appendix tables are empty placeholders, which makes the draft feel incomplete.\n\nNone of this means the work is empty. The generalized Jordan–Pauli structure is worth taking seriously; if the advanced-cone terms can be computed and the partial-fraction identity repaired, the paper could make a real contribution. What I cannot do is accept the claim that the Weinberg–DeWitt amplitude has been reproduced by NSF at fourth order, because the calculation that would show it is not in the manuscript.\n\nI would send this to a referee—there is enough new structure to justify referee time—but I would not recommend publication in anything like its present form. The authors should be asked to carry out the missing δK computation, fix the algebra, and make the companion papers available. If they can do that, the result will be worth revisiting.","headline":"The paper has a genuinely new structural idea—the generalized Jordan–Pauli relation with advanced-cone corrections—but the tree-level amplitude is not actually derived: the corrections are left uncomputed and the final partial-fraction check is algebraically wrong.","tokens_in":37694,"tokens_out":4090,"would_cite":false,"duration_ms":43489,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["04.60.-m"],"model":"deepseek-v4-flash","headline":"Fourth-order recursion completes the graviton tree amplitude","keywords":["null surface formulation","graviton scattering","tree-level amplitude","ultraviolet finiteness","asymptotic shear","loop counting","helicity amplitudes","unitarity"],"falsifier":"Evaluate, at order n=4, the sum of scattering-correction cut terms Σ_{j=2}^{3} Z+_{4,cut}|_{δσ+_j} in the master equation; if it is not a boundary term that vanishes after integration over the sphere, then the master equation misses contributions to σ+_4 and the tree-level identification fails. Alternatively, compute the one-loop (δaout_4)^2 integrand in the uniform radial UV region and check whether it scales as dq/q^4 rather than dq/q^{4L}; a mismatch would falsify the all-orders UV-finiteness theorem.","tokens_in":36765,"feed_emoji":"🌌","tokens_out":9629,"duration_ms":100386,"temperature":0.7,"pith_summary":"The paper completes a three-part program that computes quantum graviton scattering directly from null-cone data at past and future infinity, without off-shell propagators or loop-momentum integrals. It derives the fourth-order correction to the outgoing shear and shows that the tree-level 2→2 amplitude is exhausted by contributions from the second-, third-, and fourth-order fields; after imposing four-momentum conservation externally, those three pieces add to the standard result −κ²s³/(4tu). It also proves a loop-counting formula, L=(n1+n2−6)/2, which says no field beyond fourth order contributes at tree level, and a power-counting theorem, K(n)∼ω_ext/q^{n−2}, which makes the L-loop integrand scale as dq/q^{4L}. If these hold, the expansion is ultraviolet finite at every loop order without regularization. A sympathetic reader would care because the result points to a version of perturbative quantum gravity in which UV divergences never arise, and in which the S-matrix is recursively determined by a single free datum.","feed_headline":"Fourth-order recursion completes the graviton tree amplitude","feed_subtitle":"The three lowest-order shear fields give the full tree amplitude, with every loop order ultraviolet finite.","key_machinery":"The generalized retarded-advanced cone matching equation — the master equation — is the central object. It equates the advanced and retarded null-cone integrals with the linear cut term, after the free-field parts cancel between the two cones; the surviving terms are the advanced-cone corrections from previously determined scattering fields. The same equation yields the kernel-scaling relation K(n)∼ω_ext/q^{n−2} inductively, while the operator-counting identity L=(n1+n2−6)/2 classifies loop order. The inverse of the angular derivative operator on the sphere converts cone sources into shear coefficients and produces the denominators that become s, t, u on shell.","core_discovery":"The central claim is that the perturbative null-surface formulation determines every coefficient of the outgoing shear from the free incoming datum, and at fourth order the computation closes the tree-level 2→2 graviton amplitude. Concretely, σ+_4 is obtained by inverting the second-order angular derivative operator on the sphere, using three retarded cone pairs and, for the first time, advanced-cone corrections built from the nonzero second- and third-order scattering fields; the full kernel is twice the retarded contributions plus those advanced corrections. The resulting matrix element M(24), together with M(22) and M(33), equals −κ²s³/(4tu) after the kinematic substitution that maps the","pith_inferences":["If the cancellation of scattering-correction cut terms holds at every order, the entire perturbative S-matrix is fixed by the single free incoming datum at past null infinity; no additional asymptotic input would be needed at any loop order.","The null-denominator structure that produces UV convergence is also what generates soft limits; checking whether the recursion reproduces the universal soft-graviton factor order by order would be a direct test of the framework beyond the tree amplitude.","Because all internal gravitons are on-shell and fixed by frequency deltas, the expansion reorganizes the amplitude as a sum over causal histories; evaluating the one-loop (δaout_4)^2 integrand explicitly would test the claimed dq/q^4 scaling before the missing fifth- and sixth-order fields are computed.","The paper uses flat null cones; if the cones are replaced by the self-corrected ones determined by the recursion, the inductive denominator growth could change, which would alter the UV power counting — a direction the paper itself names as future work."],"forward_implications":["Tree-level 2→2 graviton scattering is fully accounted for by the second-, third-, and fourth-order shear fields; no higher-order field can contribute at zero loop order.","Imposing four-momentum conservation externally on the null-cone denominators reproduces the standard amplitude −κ²s³/(4tu), with the three pole channels s, t, u individually identified.","Every L-loop contribution to 2→2 scattering scales as ∫ dq/q^{4L}, so the ultraviolet part vanishes as the regulator is removed and convergence improves with loop order; no renormalization is needed.","A complete one-loop amplitude requires the fifth- and sixth-order shear fields in addition to the fourth; this paper identifies the required operator sectors but leaves those fields for future work.","The S-matrix is unitary order by order, because each order's generator is Hermitian as a consequence of the real kernels and the exponential expansion of the unitary operator."],"fun_headline_variants":["Graviton tree amplitude closes at fourth order via null-surface recursion","Null-surface recursion yields full graviton tree amplitude","Fourth-order shear closes graviton tree amplitude","All-order recursion for graviton scattering from null surfaces","Graviton scattering: tree amplitude complete at fourth order"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is that all scattering-correction cut terms, built from the earlier nontrivial fields δσ+_j, cancel as boundary or total-derivative terms in the master equation; if that cancellation is not exact, σ+_4 picks up additional contributions and the computed kernels, and hence the tree-level identification, are incomplete.","fun_headline_variants_meta":{"raw":{"variants":["Graviton tree amplitude closes at fourth order via null-surface recursion","Null-surface recursion yields full graviton tree amplitude","Fourth-order shear closes graviton tree amplitude","All-order recursion for graviton scattering from null surfaces","Graviton scattering: tree amplitude complete at fourth order"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000595,"raw_usage":{"total_tokens":2696,"prompt_tokens":891,"completion_tokens":1805,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":635,"completion_tokens_details":{"reasoning_tokens":1728}},"tokens_in":635,"tokens_out":1805,"duration_ms":14347,"temperature":1.0,"reasoning_tokens":1728,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T13:14:30.191463+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate, at order n=4, the sum of scattering-correction cut terms Σ_{j=2}^{3} Z+_{4,cut}|_{δσ+_j} in the master equation; if it is not a boundary term that vanishes after integration over the sphere, then the master equation misses contributions to σ+_4 and the tree-level identification fails. Alternatively, compute the one-loop (δaout_4)^2 integrand in the uniform radial UV region and check whether it scales as dq/q^4 rather than dq/q^{4L}; a mismatch would falsify the all-orders UV-finiteness theorem.","supporting_citations":[],"review_version":2}