{"id":"6fe00830-f058-4d52-a951-a2e2d7a51846","arxiv_id":"2511.16062","paper_version":2,"verdict":"CONDITIONAL","confidence":"LOW","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"GESC is a gauge-equivariant GNN that uses self-interference cancellation via phase connections and rank-1 projections to improve performance on heterophilous graphs.","lead":"The paper introduces GESC, a graph neural network that adds a U(1) phase connection and rank-1 projection to cancel self-interference before attention, replacing standard additive aggregation. This targets heterophily and oversmoothing problems that standard gauge-equivariant GNNs leave unaddressed.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"Attribution of oversmoothing to missing self-interference handling is not isolated from other GESC components","rationale":"The proposed test directly probes the reader's weakest assumption by isolating the projection step. Because the reader's verdict is already CONDITIONAL on the basis of abstract-only review and flags exactly this attribution issue, confirming or refuting the ablation would either strengthen or further qualify the claim without requiring a verdict change at this stage. The concern is internal to the argument's causal isolation rather than external consensus.","tokens_in":1728,"tokens_out":368,"duration_ms":27365,"concrete_test":"Implement an ablation variant of GESC that retains the U(1) phase connection and sign-aware gate but replaces the rank-1 projection with standard additive aggregation; retrain and evaluate on the same heterophilous benchmarks (e.g., Texas, Cornell, Chameleon) used in the paper. If the performance gap to the original GESC exceeds 3-5% on average while the gap to prior gauge models shrinks substantially, the interference mechanism is not shown to be the primary driver.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires that redundant low-frequency self-interference (rather than attention dilution, normalization, or the sign-aware gate) is the dominant oversmoothing driver in prior gauge-equivariant GNNs. The abstract invokes this when stating that existing models 'lack interference handling' and that this 'drives oversmoothing under gauge transport,' yet the reported gains compare full GESC against external baselines. No ablation is described that holds the U(1) phase connection and sign-aware gate fixed while removing only the rank-1 projection, leaving open the possibility that performance differences arise from those other elements or from implementation details rather than interference cancellation per se.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript introduces Gauge-Equivariant Graph Network with Self-Interference Cancellation (GESC). It replaces additive aggregation in gauge-equivariant GNNs with a projection-based mechanism consisting of a U(1) phase connection, a rank-1 projection that suppresses self-parallel low-frequency components before attention, and a sign-aware gate for negatively aligned neighbors. The central claim is that the absence of interference handling in prior gauge-based models is a primary driver of oversmoothing under gauge transport; the paper reports consistent outperformance over recent state-of-the-art models on diverse graph benchmarks and releases code.","tokens_in":1870,"tokens_out":508,"duration_ms":36942,"significance":"If the attribution of gains specifically to self-interference cancellation holds, the work supplies a concrete mechanism for mitigating oversmoothing in heterophilous and gauge-transport settings and unifies message passing under an interference-aware lens. The public code release supports reproducibility.","major_comments":[{"comment":"The central claim that redundant low-frequency self-interference (rather than attention dilution, normalization, or the sign-aware gate) is the dominant oversmoothing driver requires isolation. The reported gains compare the full GESC model against external baselines, but no ablation is described that holds the U(1) phase connection and sign-aware gate fixed while removing only the rank-1 projection. This is load-bearing for the claim in the abstract that 'the absence of interference handling ... is a primary driver of oversmoothing under gauge transport.'","section":"Experimental results / ablation studies"},{"comment":"The exact definition and derivation of the rank-1 projection operator (including how it is applied after the U(1) phase connection and before attention) must be stated with explicit equations. Without this, it is not possible to verify that the projection indeed cancels self-parallel components by construction rather than through additional learned parameters.","section":"Method section (projection formula)"}],"minor_comments":[{"comment":"Notation for the sign-aware gate threshold or scaling factor should be introduced once and used consistently; the abstract refers to it only descriptively.","section":"Notation and preliminaries"},{"comment":"Figure captions should explicitly state which baselines are gauge-equivariant versus non-equivariant to aid comparison with the interference-cancellation narrative.","section":"Figures"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for the constructive and detailed comments. We address each major point below and will revise the manuscript accordingly to strengthen the presentation and empirical support for our claims.","responses":[{"response":"We agree that an internal ablation isolating the rank-1 projection is necessary to substantiate the central claim. In the revised manuscript we will add an ablation that fixes the U(1) phase connection and sign-aware gate while comparing performance with and without the rank-1 projection. This will directly test whether the projection, rather than the other components, is responsible for mitigating self-interference and oversmoothing.","revision_made":"yes","referee_comment":"[Experimental results / ablation studies] The central claim that redundant low-frequency self-interference (rather than attention dilution, normalization, or the sign-aware gate) is the dominant oversmoothing driver requires isolation. The reported gains compare the full GESC model against external baselines, but no ablation is described that holds the U(1) phase connection and sign-aware gate fixed while removing only the rank-1 projection. This is load-bearing for the claim in the abstract that 'the absence of interference handling ... is a primary driver of oversmoothing under gauge transport.'"},{"response":"We acknowledge that the current exposition would benefit from greater mathematical precision. In the revised manuscript we will add explicit equations in the Method section defining the rank-1 projection operator, deriving its action on the phase-adjusted features, and showing its application immediately after the U(1) phase connection and before the attention step. The derivation will demonstrate that the operator projects out the self-parallel component by construction.","revision_made":"yes","referee_comment":"[Method section (projection formula)] The exact definition and derivation of the rank-1 projection operator (including how it is applied after the U(1) phase connection and before attention) must be stated with explicit equations. Without this, it is not possible to verify that the projection indeed cancels self-parallel components by construction rather than through additional learned parameters."}],"tokens_in":1381,"tokens_out":420,"duration_ms":46249,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The main point to know is that GESC introduces an explicit self-interference cancellation mechanism into gauge-equivariant graph networks. It uses a U(1) phase connection followed by a rank-1 projection to suppress self-parallel components, plus a sign-aware gate, aiming to address oversmoothing on heterophilous graphs better than prior additive approaches. This combination looks new compared to the cited magnetic and gauge GNN literature. The projection step before attention is a concrete architectural choice that isn't just re-applying existing operators. The paper does well by reporting consistent outperformance on various benchmarks and releasing the code, which allows checking the implementation. The softer part is the attribution of oversmoothing primarily to missing interference handling. The abstract states this as a key insight, but the results compare the complete model against baselines without describing an ablation that isolates the rank-1 projection while holding the phase connection and gate constant. That leaves room for the gains to come from other elements instead. This work is aimed at graph learning researchers dealing with heterophily or equivariance. A reader interested in new message passing variants will find practical value here. It has a solid enough idea and results to deserve a serious referee. I would recommend sending it to peer review.","headline":"GESC adds a U(1) phase plus rank-1 projection to cancel self-interference in gauge-equivariant GNNs, but the oversmoothing attribution rests on full-model comparisons without isolating ablations.","tokens_in":2384,"tokens_out":333,"would_cite":false,"duration_ms":39649,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":{"model":"grok-4.3","evidence":[{"relation":"unclear","rs_module":"IndisputableMonolith/Cost/FunctionalEquation.lean","rs_theorem":"washburn_uniqueness_aczel","paper_passage":"We introduce a U(1) phase connection followed by a rank-1 projection that suppresses self-parallel components before attention, and a sign-aware gate that regulates negatively aligned neighbors."},{"relation":"unclear","rs_module":"IndisputableMonolith/Foundation/AbsoluteFloorClosure.lean","rs_theorem":"absolute_floor_iff_bare_distinguishability","paper_passage":"Proposition 5.1 (Effect of SIC on message decomposition) ... (I−η_sic Π_ε(h_i)) h̃ = (1−η_sic λ) h̃_∥ + h̃_⊥"},{"relation":"unclear","rs_module":"IndisputableMonolith/Foundation/AlexanderDuality.lean","rs_theorem":"alexander_duality_circle_linking","paper_passage":"Theorem 5.5 (Gauge equivariance) ... local U(1) gauge transformation"}],"headline":"GNN architecture with U(1) phase transport and rank-1 SIC projection unrelated to RS distinction-forcing chain","alignment":"orthogonal","rationale":"Paper's core machinery (complex embeddings, gauge-equivariant U(1) transport, Tikhonov-regularized rank-1 projector Π_ε for self-parallel cancellation, sign-aware gating, Lipschitz bounds on message passing) operates in the domain of heterophily-aware graph learning and spectral filtering. It invokes neither J-cost, ratio symmetry, φ-ladder, 8-tick periodicity, nor any theorem from the reality_from_one_distinction forcing chain. No parameter-free derivation of constants or cosh-type recognition cost appears; the work is a standard ML architecture paper whose claims about oversmoothing are evaluated on graph benchmarks.","tokens_in":64449,"confidence":"high","tokens_out":443,"duration_ms":17257,"cache_read_input_tokens":38528,"cache_creation_input_tokens":0},"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"A projection-based mechanism cancels self-interference to reduce oversmoothing in gauge-equivariant graph networks.","keywords":["gauge-equivariant GNN","self-interference cancellation","oversmoothing","heterophily","graph neural networks","U(1) gauge","rank-1 projection","message passing"],"falsifier":"Running the same benchmarks with and without the rank-1 projection and sign-aware gate to check if performance gains disappear and oversmoothing metrics increase when interference handling is removed.","tokens_in":2617,"feed_emoji":"📉","tokens_out":570,"duration_ms":65475,"temperature":0.7,"pith_summary":"Graph neural networks often struggle with heterophilous graphs because signals become self-reinforcing and phase-inconsistent during message passing. The paper shows that existing gauge-equivariant GNNs lack mechanisms to handle self-interference from redundant low-frequency components, which drives oversmoothing under gauge transport. GESC addresses this by using a U(1) phase connection followed by a rank-1 projection that suppresses self-parallel components before attention, plus a sign-aware gate for negatively aligned neighbors. This replaces standard additive aggregation with an explicit interference-aware approach. Readers should care if they work with graphs where standard methods fail to propagate useful information across dissimilar nodes.","feed_headline":"Self-interference cancellation curbs oversmoothing in gauge GNNs","feed_subtitle":"A U(1) phase connection and rank-1 projection suppress redundant low-frequency components, allowing consistent outperformance over state-of-","key_machinery":"A rank-1 projection applied after a U(1) phase connection that suppresses self-parallel components in the message passing process.","core_discovery":"The paper claims that the absence of interference handling in gauge-based GNNs is a primary driver of oversmoothing, and introduces GESC to explicitly model and cancel self-interference from redundant low-frequency components using a projection mechanism, leading to consistent outperformance on diverse graph benchmarks.","pith_inferences":["This could inspire interference-aware designs in non-gauge GNNs to address oversmoothing.","Connections to phase inconsistency in signals might link to other heterophily solutions.","Scalability tests on larger graphs would show if the projection remains efficient."],"forward_implications":["Absence of interference handling drives oversmoothing in gauge-based GNNs.","GESC outperforms recent state-of-the-art models across diverse graph benchmarks.","The approach provides a unified interference-aware view of message passing.","Replacing additive aggregation with projection suppresses self-parallel components effectively."],"fun_headline_variants":["Self-interference cancellation reduces oversmoothing in gauge GNNs","Gauge GNNs reduce oversmoothing by canceling self-interference","Rank-1 projection suppresses self-parallel components in GNNs","U(1) gauge connection cancels low-frequency interference in GNNs","Projection suppresses redundant low-frequency signals before attention"],"cache_read_input_tokens":64,"weakest_assumption_plain":"Self-interference from redundant low-frequency components is the dominant cause of oversmoothing in prior gauge-equivariant GNNs.","fun_headline_variants_meta":{"raw":{"variants":["Self-interference cancellation reduces oversmoothing in gauge GNNs","Gauge GNNs reduce oversmoothing by canceling self-interference","Rank-1 projection suppresses self-parallel components in GNNs","U(1) gauge connection cancels low-frequency interference in GNNs","Projection suppresses redundant low-frequency signals before attention"]},"model":"grok-4.3","cost_usd":0.013269,"raw_usage":{"total_tokens":5728,"prompt_tokens":626,"num_sources_used":0,"completion_tokens":74,"cost_in_usd_ticks":132687000,"prompt_tokens_details":{"text_tokens":626,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":5028,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":626,"tokens_out":74,"duration_ms":58840,"temperature":1.0,"reasoning_tokens":5028,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-05-21T17:44:30.892476+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Running the same benchmarks with and without the rank-1 projection and sign-aware gate to check if performance gains disappear and oversmoothing metrics increase when interference handling is removed.","supporting_citations":[],"review_version":1}