REVIEW 4 major objections 1 minor 19 references
Performances and Correlations of Centrality Measures in Complex Networks
T0 review · 4 major / 1 minor · reviewed 2026-08-05 · deepseek-v4-flash
Pith's one-line read Across 80 real-world networks, 16 centrality measures split into two strongly correlated communities; the five idiosyncratic measures each capture distinct aspects of node importance, and performance rankings for identifying influential sin
desk verdict A paper only in its abstract: all the supporting science is absent, so the honest call is desk rejection, though the core question is worth a proper look. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing machinery is an epidemic spreading model used as the ground-truth scoring function for node influence, together with the correlation matrix of the 16 centrality measures' node rankings on 80 networks. A centrality measure assigns each node a score; the paper then ranks nodes by that score, takes the top-ranked nodes as the candidate most influential nodes, and evaluates the candidate(s) under the epidemic process. The second analytic object is the topological distance between a measure's top-ranked nodes: the paper claims this spatial dispersion predicts performance in the node-set task. The abstract does not state the epidemic model's parameters (transmission, recovery, see
What would settle it
Read the supplied full text looking for the 80-network, 16-measure centrality study. The provided text is about robotic cable routing and contains no epidemic model, no correlation matrix, and no centrality rankings. To settle the abstract's claim, obtain the actual paper and check (a) whether the epidemic parameters are stated, and (b) whether re-running the two task evaluations with different parameters preserves the negative correlation between the two rankings.
Extended reading notes
Core claim
The discovery the abstract advances is that centrality measures are not interchangeable, and their differences matter concretely when a network analyst must decide which nodes to target. Pairwise correlations of node rankings across 16 measures are moderate to high overall, but the correlation structure has two tight clusters—one with 4 measures, one with 7—whose members rank nodes nearly identically, while 5 measures (including, per the abstract's performance results, the top performers for the node-set task) behave idiosyncratically, correlating weakly with everything. The paper further claims that where a measure places its top-ranked nodes has predictive value: measures whose top nodes a
Load-bearing premise
The load-bearing premise is that the epidemic spreading model—with specific but unreported transmission, recovery, and seed-selection settings—is a valid ground truth for node influence in both tasks; if those settings change, the measure rankings and their negative correlation can change, and the supplied full text (an unrelated robotics paper) provides no way to inspect them.
Editorial extensions
If this is right
- If the negative correlation holds, an analyst optimizing for a single influential node will systematically mis-select measures for the node-set task; choosing a top single-node measure such as LocalRank will tend to underperform on set selection.
- The two-community correlation structure implies strong redundancy within each community: for many purposes, one representative measure per community may carry nearly the same ranking information at far lower computation cost.
- Because the five idiosyncratic measures are weakly correlated with all others, any consensus or ensemble ranking that drops them loses information the other 11 measures do not provide.
- The spatial-dispersion result gives a cheap heuristic: before running an expensive spreading simulation, measure the average pairwise distance among the top-ranked nodes of a candidate centrality; higher dispersion predicts better node-set performance.
Reading between the lines
- One immediate robustness test the paper leaves implicit: vary the epidemic parameters (transmission rate, recovery rate, seed fraction) and re-run the two task rankings; if the negative correlation is not preserved, the headline is an artifact of the chosen parameter regime rather than a structural property of centrality measures.
- The same correlation-community analysis could be extended to 'effective rank' estimation—finding the minimal subset of the 16 measures that reproduces the full battery's ranking information—which would make the practical guidance actionable for large networks.
- The dispersion–performance link suggests a possible connection between centrality measures and network backbone structure: in networks with strong modularity, distant top nodes may cover more communities, which could explain the node-set advantage; this is testable by conditioning on modularity in the 80-network corpus.
- The supplied full text being an unrelated robotics paper, the first verification step is to obtain the actual manuscript and confirm that the 80 networks, the measure list, and the epidemic parameters accompany the abstract's numbers.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The manuscript by arXiv:2508.09563 is submitted as a statistical study of centrality measures. Based on the abstract, it claims an empirical comparison of 16 centrality measures on 80 real-world networks, finding two communities of strongly correlated measures (sizes 4 and 7) and five idiosyncratic measures, plus an analysis of influential-node spatial distributions and a comparison of two epidemic-spreading-based tasks. The abstract's headline result is that the performance rankings of the 16 measures for identifying the most influential single node versus the most influential node set are negatively correlated. The supplied full text, however, is an unrelated robotics manuscript ('CaRoBio: 3D Cable Routing with a Bio-inspired Gripper Fingernail', arXiv:2508.09558v1), containing no mention of centrality, networks, epidemic models, or correlation analysis. Therefore, none of the empirical claims can be inspected, and the central argument is entirely unsupported in the provided material.
Significance. If substantiated, the abstract's claims would be of interest to network science: a systematic 80-network comparison of 16 measures, the identification of two correlated communities, and the striking negative correlation between single-node and node-set influence rankings would all be useful empirical contributions. The potential contribution is also notable because the paper appears to offer no parametric curve-fitting; the central comparisons are direct empirical correlations, which if fully documented would be straightforward to reproduce. However, with no accompanying methods, data, or analysis in the supplied full text, the significance cannot be evaluated. The result may be correct, but the manuscript as provided offers no evidence for it.
major comments (4)
- [Full Text (entire manuscript)] The supplied full text is arXiv:2508.09558, 'CaRoBio: 3D Cable Routing with a Bio-inspired Gripper Fingernail,' a robotics paper by different authors. It contains no centrality measures, no networks, no epidemic model, and no correlation analysis. Every load-bearing component of the abstract—the 80-network corpus, the definitions of the 16 measures, the correlation statistic, the community-detection procedure, the epidemic parameters, and the node-set identification—is absent. This is not a missing derivation that can be patched in revision; the complete submission is mismatched.
- [Abstract, quantitative claims] The abstract reports specific numbers (two communities of 4 and 7, five idiosyncratic measures, 80 networks) but no methodology. There is no definition of the rank-correlation statistic, no threshold for 'moderate to high correlation' or for 'exceptionally strong pairwise correlations,' no description of how communities were detected, and no listing of the 16 centrality measures. Without these, the reported partition is not reproducible.
- [Abstract, epidemic spreading model] The abstract says 'Using the epidemic spreading model' but gives no model family (e.g., SIR/SIS), no transmission or recovery rates, no seed-selection procedure, and no definition of 'most influential node set' (e.g., size of the set, whether it is selected greedily or by combinatorial optimization). The headline negative correlation between the two task rankings could plausibly depend on these choices, as the reader's stress-test notes. Since no parameter values or sensitivity analysis are provided, the claim cannot be checked.
- [Abstract, node-set task] The key novelty is the negative correlation between rankings for 'identifying the most influential single node' versus 'the most influential node sets.' The abstract never states how a node set's influence is measured or how the 'most influential node sets' are defined. This is central because one obvious explanation of the negative correlation would be that the single-node task and the set task use incompatible or arbitrarily chosen set sizes; without a clear definition, the result is uninterpretable.
minor comments (1)
- [Abstract, wording] The phrase 'the resulting rankings are negatively correlated' could be clarified by naming the correlation coefficient (e.g., Spearman's rho) and the aggregation across 80 networks (e.g., average of per-network ranks or rank of averaged scores).
Circularity Check
No circularity detectable: the supplied full text is an unrelated robotics paper, so no derivation chain, equations, or fitted parameters from the centrality-measure abstract are present to analyze.
full rationale
The circularity pass requires quoting a specific reduction in the paper's own derivation chain: an equation that equals its input by construction, a fitted parameter renamed as a prediction, or a load-bearing claim justified only by a self-citation. The supplied manuscript, however, is not the paper described in the abstract. The abstract for arXiv:2508.09563 reports an empirical comparison of 16 centrality measures on 80 networks, two correlated communities, and a negative correlation between single-node and node-set influence rankings under an epidemic spreading model, but the provided full text is CaRoBio, a robotics paper on cable routing with a bio-inspired fingernail. It contains no equations defining centrality measures, no epidemic model parameters, no correlation computations, and no citations to prior work by the authors that might be load-bearing. Consequently, there is no quoted text from which one could exhibit a self-definitional reduction, a fitted-input-called-prediction step, or an imported uniqueness theorem. The central claims are unverifiable from the supplied material, but unverifiability is a completeness or integrity problem, not a circularity problem. Per the hard rules, a circularity finding must be grounded in the paper's own equations or self-citation chain; no such evidence exists here. The honest finding is therefore no significant circularity, score 0.
Assumptions & free parameters
free parameters (2)
- Epidemic spreading parameters (infection rate, recovery rate, seed set size)
- Rank correlation statistic and community-detection threshold
assumptions (3)
- domain assumption Epidemic spreading is a valid ground-truth model of node influence for both single nodes and node sets
- domain assumption Rank correlations across 80 heterogeneous networks can be pooled into one comparison
- standard math A rank correlation statistic (e.g., Kendall's tau) underlies the reported 'correlation between node rankings'
Cite this review
Pith. "Pith review of Performances and Correlations of Centrality Measures in Complex Networks." pith.science (2026). https://pith.science/paper/GY4IBNR6
@misc{pith2026250809563,
author = {Pith},
title = {Pith review of: Performances and Correlations of Centrality Measures in Complex Networks},
year = {2026},
howpublished = {\url{https://pith.science/paper/GY4IBNR6}},
note = {Machine review of arXiv:2508.09563}
}
read the original abstract
Numerous centrality measures have been proposed to evaluate the importance of nodes in networks, yet comparative analyses of these measures remain limited. Based on 80 real-world networks, we conducted an empirical analysis of 16 representative centrality measures. In general, there exists a moderate to high level of correlation between node rankings derived from different measures. We identified two distinct communities: one comprising 4 measures and the other 7 measures. Measures within the same community exhibit exceptionally strong pairwise correlations. In contrast, the remaining five measures display markedly different behaviors, showing weak correlations not only among themselves but also with the other measures. This suggests that each of these five measures likely captures unique properties of node importance. Further analysis reveals that the distribution patterns of the most influential nodes identified by different centrality measures vary significantly: some measures tend to cluster influential nodes closely together, while others disperse them across distant locations within the network. Using the epidemic spreading model, we found that LocalRank, Subgraph Centrality, and Katz Centrality perform best in identifying the most influential single node, whereas Leverage Centrality, Collective Influence, and Cycle Ratio excel in identifying the most influential node sets. Overall, measures that identify influential nodes with larger topological distances between them tend to perform better in detecting influential node sets. Interestingly, despite being applied to the same dynamical process, when using two seemingly similar tasks, identifying influential nodes versus identifying influential node sets, to rank the performances of the 16 centrality measures, the resulting rankings are negatively correlated.
Reference graph
Works this paper leans on
-
[1]
Offline-online learning of deformation model for cable manipulation with graph neural networks,
C. Wang, Y . Zhang, X. Zhang, Z. Wu, X. Zhu, S. Jin, T. Tang, and M. Tomizuka, “Offline-online learning of deformation model for cable manipulation with graph neural networks,”IEEE Robotics and Automation Letters, vol. 7, no. 2, pp. 5544–5551, 2022
2022
-
[2]
J. Buzzatto, J. Chapman, M. Shahmohammadi, F. Sanches, M. Ne- jati, S. Matsunaga, R. Haraguchi, T. Mariyama, B. MacDonald, and M. Liarokapis, “On robotic manipulation of flexible flat cables: Em- ploying a multi-modal gripper with dexterous tips, active nails, and a reconfigurable suction cup module,” in2022 IEEE/RSJ International Conference on Intelligen...
work page 2022
-
[3]
I. Armengol, A. Suarez, G. Heredia, and A. Ollero, “Design, integration and testing of compliant gripper for the installation of helical bird diverters on power lines,” in2021 Aerial Robotic Systems Physically Interacting with the Environment (AIRPHARO), pp. 1–8, 2021
work page 2021
-
[4]
Contact-aware shaping and maintenance of deformable linear objects with fixtures,
K. Chen, Z. Bing, F. Wu, Y . Meng, A. Kraft, S. Haddadin, and A. Knoll, “Contact-aware shaping and maintenance of deformable linear objects with fixtures,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 1–8, IEEE, 2023
work page 2023
-
[5]
Cable routing and assembly using tactile-driven motion primitives,
A. Wilson, H. Jiang, W. Lian, and W. Yuan, “Cable routing and assembly using tactile-driven motion primitives,” in2023 IEEE International Conference on Robotics and Automation (ICRA), pp. 10408–10414, IEEE, 2023
work page 2023
-
[6]
Multi-stage cable routing through hierarchical imitation learning. arxiv pre-print, 2023,
J. Luo, C. Xu, X. Geng, G. Feng, K. Fang, L. Tan, S. Schaal, and S. Levine, “Multi-stage cable routing through hierarchical imitation learning. arxiv pre-print, 2023,”URL https://arxiv. org/abs/2307.08927, vol. 22
arXiv 2023
-
[7]
Robotic cable routing with spatial representation,
S. Jin, W. Lian, C. Wang, M. Tomizuka, and S. Schaal, “Robotic cable routing with spatial representation,”IEEE Robotics and Automation Letters, vol. 7, no. 2, pp. 5687–5694, 2022
work page 2022
-
[8]
A soft-robotic gripper for ultra-high-voltage transmission line operations,
Y . Chen, Z. Fang, S. Liu, Y . Wang, C. Zhong, C. Cai, Y . Zhang, Y . Wei, and Z. Wang, “A soft-robotic gripper for ultra-high-voltage transmission line operations,” in2020 IEEE 4th Conference on Energy Internet and Energy System Integration (EI2), pp. 788–793, 2020
work page 2020
Show all 19 references
-
[9]
Development of a gripper for manipulation of soft line-shaped object,
S. Ishikawa, T. Nishimura, and T. Watanabe, “Development of a gripper for manipulation of soft line-shaped object,” in2024 IEEE/SICE Inter- national Symposium on System Integration (SII), pp. 225–226, 2024. Fig. 10. Cable routing results with cable of different initial state, ...
2024
-
[10]
Grasping without squeezing: Shear adhesion gripper with fibrillar thin film,
E. W. Hawkes, D. L. Christensen, A. K. Han, H. Jiang, and M. R. Cutkosky, “Grasping without squeezing: Shear adhesion gripper with fibrillar thin film,” in2015 IEEE International Conference on Robotics and Automation (ICRA), pp. 2305–2312, 2015
2015
-
[11]
In- hand following of deformable linear objects using dexterous fingers with tactile sensing,
M. Yu, B. Liang, X. Zhang, X. Zhu, X. Li, and M. Tomizuka, “In- hand following of deformable linear objects using dexterous fingers with tactile sensing,”2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 13518–13524, 2024
2024
-
[12]
An approach for the bimanual manipulation of a deformable linear object using a dual-arm industrial robot: cable routing use case,
P. M. Fresnillo, S. Vasudevan, and W. M. Mohammed, “An approach for the bimanual manipulation of a deformable linear object using a dual-arm industrial robot: cable routing use case,” in2022 IEEE 5th International Conference on Industrial Cyber-Physical Systems (ICPS), pp. 1–8, 2022
2022
-
[13]
Assembly of randomly placed parts realized by using only one robot arm with a general parallel- jaw gripper,
J. Zhao, X. Wang, S. Wang, X. Jiang, and Y . Liu, “Assembly of randomly placed parts realized by using only one robot arm with a general parallel- jaw gripper,” in2020 IEEE International Conference on Robotics and Automation (ICRA), pp. 5024–5030, 2020
2020
-
[14]
Planning and control for cable-routing with dual-arm robot,
G. A. Waltersson, R. Laezza, and Y . Karayiannidis, “Planning and control for cable-routing with dual-arm robot,” in2022 International Conference on Robotics and Automation (ICRA), pp. 1046–1052, IEEE, 2022
2022
-
[15]
Feel the tension: Manipulation of deformable linear objects in environments with fixtures using force information,
F. S ¨uberkr¨ub, R. Laezza, and Y . Karayiannidis, “Feel the tension: Manipulation of deformable linear objects in environments with fixtures using force information,” in2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 11216–11222, IEEE, 2022
2022
-
[16]
Integration of robotic vision and tactile sensing for wire-terminal inser- tion tasks,
D. De Gregorio, R. Zanella, G. Palli, S. Pirozzi, and C. Melchiorri, “Integration of robotic vision and tactile sensing for wire-terminal inser- tion tasks,”IEEE Transactions on Automation Science and Engineering, vol. 16, no. 2, pp. 585–598, 2018
2018
-
[17]
Yolov8: A novel object detection algorithm with enhanced performance and robustness,
R. Varghese and S. M., “Yolov8: A novel object detection algorithm with enhanced performance and robustness,” in2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS), pp. 1–6, April 2024
2024
-
[18]
Tracking deformable objects with point clouds,
J. Schulman, A. Lee, J. Ho, and P. Abbeel, “Tracking deformable objects with point clouds,” in2013 IEEE International Conference on Robotics and Automation, pp. 1130–1137, 2013
2013
-
[19]
The motion coherence theory,
A. Yuille and N. Grzywacz, “The motion coherence theory,” in[1988 Proceedings] Second International Conference on Computer Vision, pp. 344–353, 1988
1988
Reviewed August 5, 2026 · model on record in the stance chip above.
Discussion (0). Sign in to comment.