The halo effect: how academic hierarchy undermines peer review and enables fraud
Letter to the Editor
Academic Intelligence · Curated Daily
AcademicHub aggregates real-time literature from top journals and preprint platforms. Build your personal research radar and let large language models compile cross-disciplinary analysis briefings automatically.
Letter to the Editor
arXiv:2606.15064v1 Announce Type: new Abstract: Manipulation demonstrations have temporal phase structure, and a natural hypothesis is that demonstration-curation metrics should be applied within phases rather than globally. The idea is to segment each trajectory into phases, score each phase with the metric that is locally most informative, and then aggregate. This follows directly from prior work showing that a single global metric can be the best detector of a defect and yet the worst curator of the resulting policy. We test the per-phase hypothesis on three contact-rich LIBERO pick-and-place tasks with a controlled early-release structural defect, comparing phase-gated curation against the same metrics applied uniformly and against a strong single global metric. Across all three tasks and five random seeds per condition, phase-gated curation is never the best curation strategy, and it is the worst of the three on two of the three tasks (Task 1: 86.0 vs. 92.0 for global; Task 3: 22.7 vs. 48.0 for uniform). We trace the failure to a concrete mechanism. When the defect signal is concentrated in a single phase, rank-aggregating across phases dilutes that signal with uninformative scores from defect-free phases, selecting a worse demonstration subset than simply applying the defect-informative metric everywhere. We further show that the per-phase metric selection does not transfer across tasks, since no phase shares a winning metric between any two tasks, so the selection cannot be reused and must be re-derived per task from a noisy sweep. These results bound a plausible and previously untested method, and they argue that practitioners should prefer identifying a single defect-informative metric over decomposing curation by phase. We release the full pipeline, all metric implementations, and per-seed results.
The automated analysis of heterogeneous natural textures is frequently hindered by physical damage and data loss, presenting a significant challenge to computer vision. While deep learning has shown success in controlled environments, its application to complex geological materials under conditions of incomplete information remains underexplored. This study presents an integrated framework for the inpainting and classification of high-resolution core sample images. We propose an end-to-end pipeline that utilizes object detection for sample segmentation, followed by image inpainting using Generative Adversarial Networks (GANs) with Contextual Residual Aggregation (CRA) to reconstruct missing high-frequency details. Subsequently, we evaluate the performance of modern Transformer-based (Swin, ViT) and CNN architectures on the reconstructed data. Our experiments revealed a critical divergence between reconstruction quality and downstream utility: despite high structural fidelity (PSNR 28.7~dB, FID 74.01), classification accuracy plateaued at 53\%. To improve minority-class detection, we propose a confidence-based hybrid ensemble that raises MCA from 48\% to 58\%. These results highlight the limitations of current state-of-the-art generative models, which may produce visually plausible but semantically ambiguous features ("hallucinations") that confound classifiers. This work provides insights into the dependencies between image reconstruction quality and classification performance, offering a reproducible baseline for future research in non-destructive testing and material science. Given that cross-well accuracy remains in the 49–53\% range, we position the resulting system as a decision-support and screening tool for lithofacies interpretation rather than as a fully autonomous classifier. The code is available at https://github.com/GalymzhanAbdimanap/Lithology_recognition
arXiv:2606.19664v1 Announce Type: new Abstract: We reconstruct unitary quantum dynamics from a minimal axiomatic foundation built on Hilbert-space observables and isospectral evolution. The only dynamical assumption is that physical time evolution is a continuous one-parameter flow of Hermitian observables that preserves their spectra, i.e. the possible outcomes of measurement. We show that this assumption is already sufficient to force the Lax form of quantum dynamics. The Heisenberg equation, the time-dependent and time-independent Schrödinger equations, conservation laws, and good quantum numbers then follow as theorems rather than postulates. In this formulation, Lax pair theory supplies the missing dynamical bridge between the measurement structure of a Hilbert space and standard quantum evolution: the Hamiltonian is not assumed, but emerges as the generator required for an isospectral observable flow.
Subject-driven image customization aims to generate images that not only follow textual instructions but also preserve the identity of a given reference subject. Existing approaches, including test-time fine-tuning, encoder-based methods, and token competition in shared attention spaces, suffer from limited efficiency, misalignment between extracted reference features and the generative process, and interference from irrelevant information. To address these limitations, we formulate the customization task as a distribution shift induced by incorporating reference images into text-to-image generation, and derive a Conditional Attention Distribution Shift formulation grounded in maximum entropy theory. Building on this formulation, we propose CustomShift, a dual-branch architecture based on Stable Diffusion 3. The Reference-Alignment Branch leverages self-attention between reference images and subject names to achieve layer-wise alignment with latent representations, while the Cross-Guidance Branch integrates textual and reference cues to guide generation. Experiments on the DreamBooth and Custom101 benchmarks demonstrate that our method consistently outperforms state-of-the-art approaches, achieving a better balance between semantic fidelity and subject consistency.
arXiv:2606.13799v1 Announce Type: cross Abstract: Finding the shortest program that generates a sequence is uncomputable, and for six decades that fact has been mistaken for a wall around finding any generating program. It is not a wall but a price, and this paper measures it. For every algorithm that learns about a candidate program only through its score, a class spanning Levin search, evolutionary methods, simulated annealing, and the cross-entropy method, we define the coupling width of a search problem and prove an unconditional worst-case lower bound, exponential in that width with base one less than the domain size. From it follows a conservation law: structural knowledge injected into a search trades one for one against the search it removes, and their sum can never fall below the length of the program sought. Levin's 1973 upper bound and the lower bound proved here are the two ends of one conserved quantity, closing on each other as the instruction set grows. The only escape is to read a candidate's structure rather than its score, and its price, which we prove for generic targets, is incompleteness. A deterministic engine built on this theory recovers a generating program, certified by compressing its data and predicting an unseen continuation, for 2,383 of 3,914 sequences across four independent populations, including 244 of the 256 elementary cellular automata, with measured discovery cost rising along program length more than an order of magnitude inside the score-oracle worst case.
While significant progress has been made in perinatal outcomes over recent decades in low- and middle-income countries (LMICs), maternal and newborn quality improvement initiatives often fail to account for the spatial conditions in which they are implemented. Health systems are increasingly deploying evidence-based care models into built environments that are not optimally structured to meet the needs of its patient population. As the principal users, patients and health care workers can offer pragmatic insights about improving these structural designs. Our objective was to gather insights from patients, providers, and companions about how the physical design of their health facilities influenced their experience receiving or delivering perinatal care. We conducted a prospective observational study using a human-centred design (HCD) approach to analyse perceptions of the quality of perinatal care across two low resource settings: Ethiopia and Bangladesh. Using engagement and assessment tools, we conducted interviews, focus groups, facility walk-throughs, co-design workshops, and infrastructural assessments with patients, companions, providers, and Ministry of Health representatives. Descriptive statistics and thematic analysis were used to identify key learnings and develop recommendations. Across both countries, participants identified the need for facility layouts that better support privacy, mobility during labour, alternative birth positions, companion involvement, cultural and religious practices, sanitation, and provider visibility. Based on these insights, we developed six recommendations to better align health facility infrastructure with maternal and newborn care delivery needs. Our findings suggest that investments in health facility infrastructure may improve care experiences and help enable respectful, safe, and evidence-based maternal and newborn care. Alongside targeted spatial improvements, government authorities responsible for health facility planning should incorporate participatory design processes to ensure infrastructure reflects the needs of patients, companions, and providers and supports high-quality care delivery.
arXiv:2606.17665v1 Announce Type: cross Abstract: This paper builds a hierarchy of explicit, non-asymptotic tail bounds for the supremum of the Kostlan–Shub–Smale (KSS) random field on the sphere, and applies it to two problems: Spiked Tensor PCA and the landscape of the spherical $k$-spin model. For Tensor PCA, we study the non-asymptotic statistical limits of estimating a rank-$R$ symmetric signal tensor of order~$k\ge 3$ and dimension~$d\ge 3$ from a single Gaussian observation at signal-to-noise ratio~$\lambda$, through the profile maximum likelihood estimator, the MLE restricted to normalized rank-$R$ tensors of coherence at least~$\kappa$. Our analysis uses a single reduction: a deterministic geometric inequality (the Tube Method) and a rank-reduction step bound the estimation error by the supremum of the canonical KSS field, which the Kac–Rice formula turns into a Gaussian integral against the expected absolute characteristic polynomial of a shifted Gaussian Orthogonal Ensemble, controlled in turn by the four explicit tail bounds of our hierarchy (three from a Mehta–Fyodorov representation, one from a Ben Arous–Dembo–Guionnet large deviation). The same reduction yields two results, each with explicit constants. For estimation, a finite-$(k,d)$ error bound recovers the asymptotically optimal rate~$\sqrt{d\log k}$ of Perry, Wein and Bandeira, with explicit dependence on the rank~$R$ and the coherence~$\kappa$. For the landscape, a two-sided non-asymptotic bracketing of the annealed complexity of the spherical $k$-spin Hamiltonian recovers the Auffinger–Ben Arous–\v{C}ern\'y complexity function in the high-dimensional limit.
Neutrino oscillations (see refs. 1,2 and references therein), a quantum effect manifesting at macroscopic scales, are governed by lepton flavour mixing angles and neutrino mass-squared differences3 that are fundamental parameters of particle physics, representing phenomena beyond the Standard Model. Precision measurements of these parameters are essential for testing the completeness of the three-flavour framework, determining the mass ordering of neutrinos and probing possible new physics. The Jiangmen Underground Neutrino Observatory (JUNO)4 is a 20-ktonne liquid-scintillator detector located 52.5 km from multiple reactor cores, designed to resolve the interference pattern of reactor neutrinos with sub-percent precision5,6. Here we report, using the first 59.1 days of data collected since detector completion in August 2025, the first simultaneous high-precision determination of two neutrino oscillation parameters, $${\sin }^{2}{\theta }_{12}=0.3092\,\pm \,0.0087$$ and $$\Delta {m}_{21}^{2}=(7.50\,\pm \,0.12)\times 1{0}^{-5}\,{\mathrm{eV}}^{2}$$ for the normal mass ordering scenario, improving the precision by a factor of 1.6 relative to the combination of all previous measurements. These results advance the basic understanding of neutrinos, validate the design of the detector and indicate the readiness of JUNO for resolving the neutrino mass ordering with a larger dataset. The rapid achievement with a short exposure highlights the potential of JUNO to push the frontiers of precision neutrino physics and paves the way for its broad scientific programme. The first data of the Jiangmen Underground Neutrino Observatory deliver high-precision neutrino oscillation parameters, improving measurements and demonstrating readiness to determine neutrino mass ordering.
Predicting single-cell transcriptional responses to genetic, chemical and cytokine perturbations is a fundamental challenge in computational biology and AI Virtual Cell (AIVC) modeling, with direct implications for drug discovery and the elucidation of gene regulatory networks. Existing approaches often rely on auxiliary cell-state encoders, hierarchical variational autoencoders, dedicated Transformer encoder-decoder modules, or gene-interaction priors to compress high-dimensional expression profiles into latent representations. While effective, these designs increase architectural complexity and may limit scalability and generalizability. This paper introduces OCOO-T, a minimalist flow-matching-based AIVC model for transcriptional perturbation response prediction. OCOO-T utilizes a vanilla Transformer stack that operates directly on continuous gene expression profiles and formulates perturbation response prediction as a continuous-time denoising process. Perturbation embeddings, dosage information, and cell-line/cell-type specificity are integrated through adaptive layer normalization and in-context tokens. Comprehensive evaluations on Tahoe100M, Replogle, and PBMC benchmarks demonstrate that OCOO-T achieves state-of-the-art performance across diverse perturbations and cell types while effectively scaling to long transcriptional profiles through patching and depatching of cellular contexts. By leveraging the simplicity of Transformer-based denoising for single-cell omics, OCOO-T provides an effective and scalable framework for in-silico cellular simulation.
Property Graphs are rapidly being adopted as database frameworks for representing heterogeneous data sources. To enable precise access to the information contained in them we need conversational interfaces based on Text-To-Cypher (Text2Cypher) parsers. This paper presents an automatic synthetic data generation method that can be leveraged to fine-tune small LLMs for this task. We conduct experiments on all the major Text-To-Cypher benchmarks, demonstrating that with our synthetic data generation approach we can significantly increase the performance of small LLMs, allowing them to compete with much larger proprietary models. This means that in settings in which models must be locally deployed we can ensure data-sovereignty without sacrificing accuracy and without costly annotation campaigns.
Turn-taking in multi-party spoken conversations remains a fundamental challenge for voice-based agents, particularly under dynamic floor competition and varying user expectations. We propose ModeratorLM, a role-playing voice agent that conditions turn-taking behavior on an explicitly assigned role in multi-party settings. The system is built on a speech large language model operating in chunk-wise streaming manner. We further introduce a reasoning-augmented variant that incorporates chain-of-thought reasoning over conversational context and the assigned role. We construct RolePlayConv, a large-scale synthetic dataset of spoken multi-party conversations with diverse assistant roles. Experiments on real-world meeting data and RolePlayConv show improved turn-taking precision by over 40% and recall by more than 70%, while substantially reducing false-positive interruptions compared to non-role-conditioned baselines.
The quadratic complexity of attention poses a critical bottleneck for long-context processing, spurring interest in hybrid attention designs. Most open-source hybrid models adopt a layer-wise strategy. Yet, prior work has noted the inherent difficulty of integrating Linear Attention (LA) with Full Attention (FA), suggesting that the design space of attention hybridization remains underexplored. To probe this space, we conduct interpretability analysis and observe that layers exhibit block-wise functional similarity, while individual heads within the same layer display distinct functional specialization despite sharing input features. This head-level heterogeneity suggests that the head dimension provides a natural and principled granularity for fusing heterogeneous attention signals. Building on this insight, we introduce HydraHead, a novel architecture that hybridizes FA and LA along the head axis. HydraHead features two key innovations: (1) an interpretability-driven selection strategy that identifies retrieval-critical heads and preserves FA only for them, and (2) a scale-normalized fusion module that reconciles the distributional gap between FA and LA head outputs. By leveraging a three-stage transfer pipeline with parameter reuse and distillation, we achieve high-performance hybrid models with minimal training overhead. Under a unified training setup, HydraHead outperforms other hybrid designs in long-context tasks while maintaining strong general reasoning. With interpretability-driven head selection, it matches a 3:1 layer-wise hybrid's long-context performance at a 7:1 LA-to-FA ratio. Crucially, trained on only 15B tokens, HydraHead achieves over 69% improvement over the baseline at 512K context length, approaching Qwen3.5, a leading model of comparable size with a native context length of 256K. This highlights the significant scaling potential of head-level hybridization.
arXiv:2606.04602v3 Announce Type: replace Abstract: As agents grow more capable, legal-domain LLM agents promise to turn document-heavy matters into reviewable work products – yet reliable deployment faces three obstacles: no large-scale evidence on how today's strongest model-and-harness combinations behave on end-to-end legal matters; no agent architecture adapted to the legal vertical, only general-purpose harnesses; and, in a setting that keeps shifting with new facts, authorities, and deadlines, no mechanism for systems to learn from their own outcomes. We address each. A large-scale empirical study on Harvey LAB – $12{,}510$ agent trajectories – shows that even frontier agents remain far from completing matters in a single pass: per-criterion accuracy climbs with stronger models while strict matter completion stalls. We then introduce \textsc{Parthenon}, a self-evolving legal-agent framework that factors Model, Harness, Agent roles, legal Knowledge, deterministic Tools, and procedural Skills into auditable surfaces for source traceability, date and number grounding, deliverable compliance, and issue closure. Finally, an anti-leakage learning loop converts scored failures into task-agnostic edits to skills, tools, and knowledge, letting the system improve with experience – as a firm refines its checklists and playbooks after each matter – without touching model weights. Across our large-scale empirical analysis, \textsc{Parthenon} substantially improves the performance of state-of-the-art models and harnesses on legal-matter tasks.
Diffusion LLMs (dLLMs) have recently emerged as a powerful alternative to autoregressive LLMs (AR-LLMs) with the potential to operate at significantly higher token-generation rates. To unlock this potential, we present Spiffy, a speculative decoding algorithm to accelerate dLLM inference while provably preserving the model's output distribution. This work addresses the unique challenges involved in applying ideas from speculative decoding of AR-LLMs to dLLMs. Spiffy performs auto-speculation to eliminate the overheads of an independent draft model, structuring draft states in the form of a novel directed draft graph to take advantage of the bidirectional, blockwise nature of dLLM generation. These draft graphs are calibrated offline to maximize acceptance rates and are dynamically pruned during inference for improved computational efficiency. We present a detailed formulation of Spiffy and demonstrate its ability to accelerate LLaDA, Dream, and SDAR models in combination with KV caching and threshold-based dynamic unmasking leading to up to $8.6\times$ reduction in model inferences and $6.3\times$ acceleration in token rate.
While the shift from cascaded dialogue systems to end-to-end (E2E) speech Large Language Models (LLMs) improves latency and paralinguistic modeling, E2E models often exhibit a significant performance degradation compared to their text-based counterparts. The standard Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) training methods fail to close this gap. To address this, we propose X-OPD, a novel Cross-Modal On-Policy Distillation framework designed to systematically align the capabilities of Speech LLMs to their text-based counterparts. X-OPD enables the Speech LLM to explore its own distribution via on-policy rollouts, where a text-based teacher model evaluates these trajectories and provides token-level feedback, effectively distilling teacher's capabilities into student's multi-modal representations. Extensive experiments across multiple benchmarks demonstrate that X-OPD significantly narrows the gap in complex tasks while preserving the model's inherent capabilities.
With the development of video understanding, there is a proliferation of tasks for clip-level temporal video analysis, including temporal action detection (TAD), temporal action segmentation (TAS), and generic event boundary detection (GEBD). While task-specific video understanding models have exhibited outstanding performance in each task, there remains a dearth of a unified framework capable of simultaneously addressing multiple tasks, which is a promising direction for the next generation of AI. To this end, in this paper, we propose a single unified framework, coined as Temporal2Seq, to formulate the output of these temporal video understanding tasks as a sequence of discrete tokens. With this unified token representation, Temporal2Seq can train a generalist model within a single architecture on different video understanding tasks. In the absence of multi-task learning (MTL) benchmarks, we compile a comprehensive co-training dataset by borrowing the datasets from TAD, TAS, and GEBD tasks. We evaluate our Temporal2Seq generalist model on the corresponding test sets of three tasks, demonstrating that Temporal2Seq can produce reasonable results on various tasks and achieve advantages compared with single-task training on this framework. We also investigate the generalization performance of our generalist model on new datasets from different tasks, which yields superior performance to the specific model.
Visual place recognition in natural forest environments remains challenging due to repetitive vegetation, weak structural cues, and significant appearance variation across traversals. To address this limitation, this paper proposes a lightweight depth-aware distillation framework that injects geometric cues into a DINOv2-based place recognition model, while maintaining its pre-trained descriptor space. Evaluated on the recent WildCross benchmark, the proposed approach yields gains over an appearance-only counterpart, providing robustness to appearance variations. These results demonstrate the importance of depth as a strong complementary modality for place recognition in natural environments and identify depth-aware distillation as a promising direction for more robust forest perception.
Systematic failures of vision models on semantically coherent subsets, known as error slices, reveal limitations in robustness and evaluation. Existing slice discovery approaches largely model slices as clusters in representation space or combinations of predefined attributes. While effective for image-level classification, such formulations are insufficient for instance-level tasks such as object detection and segmentation, where failures often arise from contextual relational and spatially grounded visual patterns. We propose GH-ESD (Grounded Hypothesis-Driven Error Slice Discovery), a generate and verify framework that reformulates slice discovery as grounded hypothesis generation and statistical verification. GH-ESD constructs relational failure hypotheses using LLM priors and grounded visual evidence, discovers hypothesis slices at the instance level via Vision Language Models, and verifies them through statistical trend analysis over instance-level errors. We also introduce GESD (Grounded Error Slice Dataset), a new benchmark for instance-level error slice discovery, providing expert-defined and spatially grounded slices derived from detection and segmentation failures. Extensive experiments demonstrate that GH-ESD consistently outperforms baselines, improving Precision@10 by 0.10 (0.73 vs. 0.63) on the GESD benchmark for detection tasks, while also supporting segmentation scenarios. GH-ESD identifies interpretable slices that facilitate actionable model improvements. The GESD dataset will be made publicly available upon acceptance.
arXiv:2606.15899v1 Announce Type: cross Abstract: Open-source LLM agent ecosystems are growing rapidly, yet the security of community-contributed skills - modular tool definitions that extend agent capabilities - remains largely unvetted. The gap we fill: existing scanners operate at the code layer and are structurally blind to instruction-layer and multi-agent risk - natural-language directives that hijack an agent, exfiltrate data through encoded side channels, or chain harm across pipelines - so what is needed is a semantic, multi-dimensional vetting system rather than another signature matcher. We present SKILLVETBENCH, a live public leaderboard on Hugging Face that uses an LLM-as-Judge to vet agent skills. What is new: SARS (Skill Agentic Risk Score), a five-dimensional agentic-risk metric with a principled weighted formula for instruction-following systems. What is integrated: full CVSS v4.0 vector decomposition and a ClawHub dual-view that places our LLM-generated review beside the official marketplace verdict. What is demonstrated: drawing on our companion benchmark paper [ 1], the LLM-as-Judge stage achieves zero false negatives across 78 confirmed-malicious skills and zero false positives across 22 benign controls, while the best static baseline (SKILLSIEVE) still misses 15%; for instruction-layer categories such as Prompt Injection and Memory Poisoning, conventional tools miss between 89% and 100% of threats (e.g., CODEBERT detects none of nine memory-poisoning skills). Detection rates vary from 35% to 95% across four LLM evaluators, motivating ensemble scoring in production deployments.
arXiv:2606.18000v1 Announce Type: cross Abstract: Optical networks need intent-driven, closed-loop agentic management, a key enabler for higher autonomy levels. We present the first T-API-compliant reasoning and act (ReAct) loop. We show that domain-specific composite tools achieve 90% oracle-validated correctness with threefold token savings compared to generic tools.
Contrastive vision-language models like CLIP exhibit remarkable zero-shot generalization. However, prompt tuning remains highly sensitive to label noise, as mislabeled samples generate disproportionately large gradients that can overwhelm pre-trained priors. We argue that because CLIP already provides a near-optimal initialization, adaptation should be inherently conservative, particularly against the extreme gradient updates common in noisy settings. To this end, we propose Double-Softmax Prompt Tuning (DSPT), a hyperparameter-free method for intrinsic gradient suppression. By applying a sequential probabilistic normalization, DSPT induces a self-adaptive saturation zone that suppresses gradients from high-error noisy samples while maintaining informative updates. We also provide both theoretical analysis and empirical evidence about how this mechanism achieves adaptive suppression. This design transforms ``gradient vanishing'', traditionally a training bottleneck, into a principled noise-filtering shield for label-noise prompt tuning. Extensive experiments confirm that this simple, drop-in design achieves state-of-the-art robustness across various noisy benchmarks, outperforming methods with complex architectures and handcrafted hyperparameters.
IMPACTeen is a dataset of textual social influence scenarios spanning interpersonal, media-based, and digital settings in an adolescent context. It contains 1,021 texts, 5,100 individual annotation records, and gold labels for social influence techniques, with each text annotated from five distinct perspectives: teenagers, parents, psychologists, communication experts, and teachers. The resource was constructed through constrained LLM generation, followed by a two-step human editing and validation phase aimed at ensuring youth-context realism. A multi-dimensional annotation covered influence presence, techniques, intentions, consequences, resistance, reactions, and annotation confidence. The dataset supports research on social influence detection, annotator disagreement, cross-lingual modeling, and the training and evaluation of language models. The dataset was created in Polish and is accompanied by a corresponding English version.
arXiv:2511.04514v2 Announce Type: replace Abstract: The phenomenon of linear mode connectivity (LMC) links several aspects of deep learning, including training stability under noisy stochastic gradients, the smoothness and generalization of local minima (basins), the similarity and functional diversity of sampled models, and architectural effects on data processing. In this work, we experimentally study LMC under data shifts and identify conditions that mitigate their impact. We interpret data shifts as an additional source of stochastic gradient noise, which can be reduced through small learning rates and large batch sizes. These parameters influence whether models converge to the same local minimum or to regions of the loss landscape with varying smoothness and generalization. Although models sampled via LMC tend to make similar errors more frequently than those converging to different basins, the benefit of LMC lies in balancing training efficiency against the gains achieved from larger, more diverse ensembles. Code and supplementary materials are available at https://github.com/DLR-KI/LMC. This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible.
arXiv:2606.19603v1 Announce Type: new Abstract: Linear probes are widely used in interpretability research and often compared by cosine similarity. The Mahalanobis cosine similarity (MCS) between two directions, which reweights the inner product by test data covariance, is a natural task-aware refinement. Ying et al. (2026) report that a probe's MCS to a reference probe trained on the out-of-distribution (OOD) data near-perfectly linearly predicts the probe's OOD AUROC (R^2 = 0.98). Here, we extend this empirical finding across models, layers, and concept domains, and prove this general phenomenon in closed form: For balanced classes whose projections are Gaussian, OOD AUROC and MCS to the reference probe are linear because both are sigmoid-shaped functions of the probe's signal-to-noise ratio (SNR) on the test data. The theory also predicts when this linearity fails, which we verify empirically. MCS offers a theoretically grounded and empirically effective alternative to Euclidean cosine similarity for comparing linear probes.