Pith.

Physics and Society · arXiv:2611.16326 v1 · submitted 12 Nov 2026 · 66 references

Tracing dependency networks explains how origins shape downstream vulnerability

Ruiwu Niu, Muhammad Azhar, Michaël Antonie van Wyk, Xiaoqun Wu

The paper says

“Many systems sustain outcomes through chains of prerequisites, yet these dependencies are often partially observed. We trace backward from sampled outcomes to recover dependency structure and measure edge importance by the fraction of outcomes lost after an edge fails.”

more of the abstract

Across 1,620 synthetic networks, revealing 90% of origins required tracing 7.24% of outcomes on average. We define pair thickness as the number of origin and outcome pairs connected through an edge. Under strict dependency, where every input is necessary, thickness counts each affected outcome once per upstream origin, whereas loss counts it once. Edges therefore form origin groups with distinct thickness and loss relations, also seen in software, citation and river topologies. Within a network, greater origin diversity corresponds to lower strict loss at equal complete thickness. We develop a growth model that predicts how edges populate these groups and how losses are distributed under partial observation, tested on 2,560 independent networks. Separating origin counts from downstream reach improves identification of edges causing the greatest losses among unsampled outcomes in strict synthetic tests. Substitutable inputs reduce losses and change which importance scores perform best. Edge importance thus measures a dependency's contribution to sustaining later outcomes, guiding the protection of critical connections.

from the abstract

Pith says

Machine review · deepseek-v4-flash · 13 Nov 2026 · what was read was not recorded · not checked by a person

It can be wrong. Challenge this review or reply to it.

The passes disagreed. Reader: holds up, with conditions. Skeptic: does not hold up. Referee: major revision.

Proposition 1 is a definitional identity, but the paper's real contribution is the systematic analysis of how sampling and root multiplicity affect edge importance—a useful framework, not a discovery. This line is wrong

4 major objections · 6 minor

How often objections from reviews like this one are right: not yet measured

Are you an author? What you can do.

Not checked: figures, code, what the 66 references say.

The paper says, at Proposition 1 (Eq. 3)

The review did not quote this passage.

Pith says: Objection 1 of 4 · major

The central theoretical result (Proposition 1) is a direct algebraic consequence of the definitions of pair thickness and strict dependency. The paper presents the grouping of edges by n(e) as a discovery, but this grouping follows tautologically from the product definition C_G(e) = n(e)U_G(e) and the strict loss formula I_e(0) = U_G(e)/T_G.

more of objection 1

The slope of the ray is n(e)T_G, which differs when n(e) differs. There is no additional structural or empirical content in this claim beyond the definitions. The empirical demonstrations merely confirm that the definitions hold in the data, which they must by construction. This undermines the claim that the framework 'explains how origins shape downstream vulnerability'—it re-describes the input assumptions. The paper does not test any nontrivial hypothesis about dependency networks; it only illustrates the consequences of its own definitions.

The paper says, at §5.1, Eq. (S3)-(S4)

The review did not quote this passage.

Pith says: Objection 2 of 4 · major

The paper states that pair thickness and strict loss are related by a constant factor within root groups, and that this explains the distinct rays. However, this is a mathematical identity, not an empirical finding. The paper should clarify that the grouping is a definitional consequence, not a discovered property of real networks.

more of objection 2

The empirical tests (Figs. 4, 5) show that the data conform to this identity, but that is expected if the definitions are applied consistently. The paper would benefit from testing a hypothesis that goes beyond the definitions, such as whether the growth model predictions hold in real networks where the generative process is unknown.

The paper says, at Fig. 4 and Fig. 5

The review did not quote this passage.

Pith says: Objection 3 of 4 · major

The paper claims that root groups explain distinct thickness–loss relations, but this is a direct consequence of the definitions. The empirical observation that edges with different root counts lie on different rays is not a test of the framework; it is a restatement of the definitions.

more of objection 3

The paper should either provide a non-trivial prediction that can be falsified, or acknowledge that the main theoretical result is a definitional identity. The growth model predictions (Figs. 8, 9) are more substantive, but they are tested on synthetic networks generated by the same model, so they are essentially consistency checks.

The paper says, at §3.1, Eq. (S3)-(S4)

The review did not quote this passage.

Pith says: Objection 4 of 4 · major

The paper states that Proposition 1 is proved in Supplementary Note 3.1, but the proof is a direct algebraic manipulation of the definitions. The paper should be explicit that this is a definitional identity, not a derived insight. The claim that 'Edges therefore form origin groups with distinct thickness and loss relations' is a restatement of the definitions, not a discovery.

more of objection 4

This does not invalidate the paper, but it should be presented as a mathematical consequence, not as an empirical finding.

6 minor objections

The paper says, at Abstract

Pith says: minor objection 1 of 6

The abstract states that 'Edges therefore form origin groups with distinct thickness and loss relations, also seen in software, citation and river topologies.' This phrasing implies that the grouping is an empirical observation, but it is a definitional consequence. Consider rephrasing to clarify that the grouping follows from the definitions and is then observed in data.

The paper says, at Introduction

Pith says: minor objection 2 of 6

The introduction could better distinguish between the definitional identity (Proposition 1) and the empirical contributions (growth model, sampling effects). The current framing may overstate the novelty of the theoretical result.

show 4 more

The paper says, at Fig. 4 caption

Pith says: minor objection 3 of 6

The caption says 'Dashed rays give the observed graph’s strict response scale RS/n.' This is not a fit but a reference line. Clarify that these are not fitted slopes but theoretical lines based on the definitions.

The paper says, at §5.1

Pith says: minor objection 4 of 6

The paper cites 'Endpoint counting 26 combined with strict failure propagation proves the relation (Supplementary Note 3.1).' The reference [26] is Batagelj (2003) on citation network analysis. It is not clear how that reference proves the relation; the proof appears to be a direct derivation from the definitions. Please clarify the role of the citation.

The paper says, at Supplementary Note 3.1

Pith says: minor objection 5 of 6

The proof of Proposition 1 is essentially a restatement of the definitions. The paper should acknowledge that this is a definitional identity, not a substantive theorem.

The paper says, at General

Pith says: minor objection 6 of 6

The paper uses the term 'discovery' to describe the grouping of edges by root count, but this is a mathematical consequence. Consider using 'consequence' or 'implication' instead.

What would settle it

Re-derive Proposition 1 from the definitions of pair thickness (Eq. 1) and strict loss (Eq. 2) without using any additional assumptions. If the derivation is a direct substitution (C_G = n U, I_e(0) = U/T_G), then the grouping by n(e) is a mathematical identity, not an empirical finding. The paper should state explicitly that this is a definitional consequence, not a testable prediction.

from the skeptic's reading

How the review went

4 readings by deepseek-v4-flash · 13 Nov 2026

  1. Reader holds up, with conditions confidence high
  2. Skeptic disagrees: says it does not hold up 9 s · 52,405 tokens read, 742 written The strict-dependency Proposition 1 is a definitional identity, not an empirical finding; the paper's central novelty—that root groups explain distinct thickness–loss rays—follows tautologically from the product definition of pair thickness and the strict failure rule, so the claimed explanatory…
  3. Editor wrote the verdict above 9 s · 53,362 tokens read, 598 written
  4. Referee major revision: 4 major objections, 6 minor 15 s · 52,872 tokens read, 1,500 written

Not run on this paper: the simulated authors' defence.

Pith did not check this review's objections against the paper.

The review in full

The review in brief, machine-written

The paper's central theoretical result (Proposition 1) is indeed a direct algebraic consequence of the definitions of pair thickness and strict terminal loss. The stress-test note is correct: edges with different root counts necessarily lie on different rays because C = nU and I = U/T, so C = nT I. This is not an empirical finding; it is a restatement of the definitions. The paper would be stronger if it acknowledged this more clearly rather than presenting it as a derived insight.

That said, the paper does real work beyond Proposition 1. The main contribution is the systematic analysis of how partial observation affects the recovery of origins versus edge rankings, and how root multiplicity interacts with sampling to produce the patterns seen in Figures 4 and 5. The growth model that predicts root group composition and loss distributions under sampling is a genuine extension, and the tests on 2,560 independent networks give it some support. The empirical demonstrations on Bioconductor, HEP-TH, the Rio Grande, and COVID-19 trees are thorough and well-executed. The paper is clearly written, the methods are detailed, and the limitations are stated explicitly (e.g., strict dependency assumption, need for public code/data).

The soft spots are real but not fatal. The lack of publicly available code and data is a significant hindrance to reproducibility, as the reader notes. The paper's claim to 'explain how origins shape downstream vulnerability' is overstated—it describes a relationship that follows from the definitions, not a causal explanation. The growth model predictions are tested on synthetic networks with known parameters, which is appropriate but limits the strength of the claims about real-world applicability. The empirical tests on regulatory perturbations (TF data) show weak correlations, which the paper honestly reports.

Who is this paper for? Researchers working on dependency networks, citation analysis, infrastructure vulnerability, or supply chain risk who want a principled way to separate origin multiplicity from downstream reach when working with sampled data. The framework is useful for thinking about these problems, even if the core result is definitional.

Recommendation: This paper deserves peer review. It is a solid methodological contribution with extensive testing. The main theoretical result is a tautology, but the paper's value is in the systematic exploration of its consequences under partial observation and the growth model. A good referee will push the authors to be more honest about what is definitional versus what is discovered, and to release code and data. I would accept a review invitation.

The strongest objection

Proposition 1 (Eq. 3) states that under strict dependency and complete observation, terminal loss I_e(0) = U_G(e)/T_G and pair thickness C_G(e) = n(e)U_G(e). This is a direct algebraic consequence of the definitions: pair thickness is defined as n(e)U_G(e), and strict loss is defined as U_G(e)/T_G because every reachable terminal fails when a necessary input is cut. The paper then presents the grouping of edges by n(e) as a discovery ('Edges therefore form origin groups with distinct thickness and loss relations'). But this grouping is a mathematical identity: edges with different n(e) necessarily lie on different rays because C_G(e) = n(e)U_G(e) and I_e(0) = U_G(e)/T_G, so C_G(e) = n(e)T_G I_e(0). The slope of the ray is n(e)T_G, which differs when n(e) differs. There is no additional structural or empirical content in this claim—it is a restatement of the definitions. The paper's central theoretical result is therefore a tautology, not a derived insight. The empirical demonstrations (Figs. 4, 5) merely confirm that the definitions hold in the data, which they must by construction. The paper does not test any nontrivial hypothesis about dependency networks; it only illustrates the consequences of its own definitions. This undermines the claim that the framework 'explains how origins shape downstream vulnerability'—it simply re-describes the input assumptions.

The reader's verdict is CONDITIONAL, with the weakest assumption being the strict dependency rule. However, the more fundamental issue is that the paper's central claimed result (Proposition 1 and the root groups) is a tautology. The paper presents it as a key insight ('Edges therefore form origin groups with distinct thickness and loss relations'), but it follows directly from the definitions. The paper does not derive any nontrivial consequence; it only restates the definitions in a different form. This is not a scientific finding—it is a mathematical identity. The paper's empirical demonstrations (Figs. 4, 5) are therefore circular: they confirm what must be true by definition. The paper's claim to 'explain' how origins shape vulnerability is misleading; it merely describes the logical consequences of its own definitions. The strict dependency assumption is not the weakest point; the entire theoretical framework is a tautology. The paper should be rejected because it does not provide any testable or falsifiable hypothesis beyond the definitions. The concrete test would reveal the circularity.

The referee's report

If the framework holds, it provides a principled way to measure edge importance in dependency networks by distinguishing between the number of origins that reach an edge and the number of outcomes that depend on it. This could have practical implications for software dependency security, infrastructure resilience, and supply chain analysis. The paper is thorough in testing on many synthetic and empirical networks, and the growth model is a novel contribution for predicting group composition under partial observation.

Pith's reading of the paper (a machine's summary, not the authors')

Many systems, from software packages to river networks, rely on chains of prerequisites. This paper studies how to measure the importance of a single dependency (an edge) in such a network. The authors define 'pair thickness' as the number of distinct origin-outcome pairs that pass through an edge. Under strict dependency, where every input is required, cutting an edge causes all downstream outcomes to fail. The paper shows that edges with the same number of upstream origins form groups with a specific relationship between pair thickness and the number of outcomes lost. By separating the count of origins from the count of downstream outcomes, the authors develop a better way to identify which edges cause the greatest losses, especially when only some outcomes are observed. They test their framework on synthetic networks, software dependencies, citation networks, river topologies, and COVID-19 transmission trees. A growth model is also developed to predict how edges populate these groups and how losses are distributed under partial observation.

Core claim. Under strict dependency and complete observation, the terminal loss I_e(0) equals U_G(e)/T_G, and pair thickness C_G(e) = n(e)U_G(e). This implies that edges with different upstream root counts n(e) lie on distinct rays when plotting pair thickness against terminal loss, and that at equal complete pair thickness, edges with more upstream roots cause smaller losses. (Proposition 1, Eq. 3)

The premise it rests on. The analysis assumes that the dependency network is a finite, simple, unweighted directed acyclic graph (DAG) with a fixed set of externally sustained roots and a fixed set of designated outcomes (terminals). The strict dependency rule (every input is necessary) is a strong assumption that may not hold in many real-world systems where inputs can be substitutable. The paper does test a tolerant rule, but the core theoretical results (Proposition 1, root groups) are derived under the strict rule.

Check this review

Pith stored no record of this review's passes and took no fingerprint of it. Reviews made by code from 13 Nov 2026 on carry both.

How often reviews like this one are right

Reviews like this one: written by deepseek-v4-flash, where what was read was not recorded.

How often objections from reviews like this one are right: not yet measured

Every value above, in the review's public record (JSON)

Reviews by people

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review

How Pith checks papers · The full audit of this paper

The paper

Abstract

Many systems sustain outcomes through chains of prerequisites, yet these dependencies are often partially observed. We trace backward from sampled outcomes to recover dependency structure and measure edge importance by the fraction of outcomes lost after an edge fails. Across 1,620 synthetic networks, revealing 90% of origins required tracing 7.24% of outcomes on average. We define pair thickness as the number of origin and outcome pairs connected through an edge. Under strict dependency, where every input is necessary, thickness counts each affected outcome once per upstream origin, whereas loss counts it once. Edges therefore form origin groups with distinct thickness and loss relations, also seen in software, citation and river topologies. Within a network, greater origin diversity corresponds to lower strict loss at equal complete thickness. We develop a growth model that predicts how edges populate these groups and how losses are distributed under partial observation, tested on 2,560 independent networks. Separating origin counts from downstream reach improves identification of edges causing the greatest losses among unsampled outcomes in strict synthetic tests. Substitutable inputs reduce losses and change which importance scores perform best. Edge importance thus measures a dependency's contribution to sustaining later outcomes, guiding the protection of critical connections.

The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.

abstract compiled · 2026-11-13

The authors' main claim, as the classifier picked it out 0.80

“Edge importance thus measures a dependency's contribution to sustaining later outcomes, guiding the protection of critical connections.”

How the main claim is stated: asserted 1.00

What the abstract itself states, as probabilities: claims first 0.04 · states limitation 0.57 · human participants 0.02 · controlled comparison 0.09 · randomized assignment 0.02 · states quantitative result 0.99 · compares with prior results 0.31 · mentions code or data availability 0.02

The abstract, sentence by sentence

This table labels the abstract only (arXiv listing abstract v1). The number is the probability the classifier gave the label.

SentenceRoleThe authors' own workStates the main conclusionStrength
Many systems sustain outcomes through chains of prerequisites, yet these dependencies are often partially observed. background 0.87 0.04 0.03 asserted 0.98
We trace backward from sampled outcomes to recover dependency structure and measure edge importance by the fraction of outcomes lost after an edge fails. method 0.95 0.96 0.07 asserted 0.99
Across 1,620 synthetic networks, revealing 90% of origins required tracing 7.24% of outcomes on average. result 1.00 0.95 0.11 asserted 1.00
We define pair thickness as the number of origin and outcome pairs connected through an edge. method 0.51 0.86 0.03 asserted 0.99
Under strict dependency, where every input is necessary, thickness counts each affected outcome once per upstream origin, whereas loss counts it once. method 0.33 0.61 0.05 asserted 1.00
Edges therefore form origin groups with distinct thickness and loss relations, also seen in software, citation and river topologies. result 0.74 0.72 0.08 asserted 0.99
Within a network, greater origin diversity corresponds to lower strict loss at equal complete thickness. result 0.92 0.81 0.10 asserted 1.00
We develop a growth model that predicts how edges populate these groups and how losses are distributed under partial observation, tested on 2,560 independent networks. method 0.64 0.96 0.09 asserted 1.00
Separating origin counts from downstream reach improves identification of edges causing the greatest losses among unsampled outcomes in strict synthetic tests. result 0.95 0.92 0.22 asserted 1.00
Substitutable inputs reduce losses and change which importance scores perform best. result 0.94 0.81 0.08 asserted 1.00
Edge importance thus measures a dependency's contribution to sustaining later outcomes, guiding the protection of critical connections. interpretation 1.00 0.47 0.76 asserted 1.00

What kind of paper the classifier reports this is: method or tool 0.83 · evidence from simulation or computation 1.00

What was examined, and what was not

Sources read

  • arXiv listing abstract v1

Not examined

  • full text
  • figures and tables
  • references and citing sentences
  • relations to other papers

Levels run

  • P abstract

Not run

  • S sections
  • T body sentences
  • C citations
  • R cross-paper relations

How the labels were made

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-11-13: 1 request, 7,992 input tokens.

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON

Figures (14)
Figure 1 of the paper
Figure 1. Backward tracing, pair thickness and terminal loss on a constructed example and the complete Bioconductor
Figure 2 of the paper
Figure 2. shows the gap for networks of 2,000 to 10,000 nodes with three mean predecessor counts and three parent selection weights. All nine combinations of predecessor count and selection weight require exten￾sive observation for reliable ranking, with mean stable success reaching 95% only between 80% and 92.5% sampling. Even at 90% sampling, mean Spearman correlation was 0.991 among discovered edges but 0.925 over all reference edges once undiscovered edges were scored zero. The undiscovered edges account for this difference. The same gap appears in a real citation network. We fixed an…
Figure 3 of the paper
Figure 3. Recovery within one fixed OpenAlex reference of 50 designated terminal works. a
Figure 4 of the paper
Figure 4. shows strict responses on the left and responses at half tolerance on the right. Under strict dependency the edges split into groups by upstream root count. Black hollow circles place each edge at its complete thickness and colored points at its observed thickness on a shared loss axis, exposing the effect of incomplete observation. Dashed rays mark the response within the observed graph. In Growth, relationships within groups are positive, yet the median pooled rank correlation is −0.029. Proposition 1. Pair counts and strict terminal loss. In the finite DAG defined above, with…
Figure 5 of the paper
Figure 5. compares these observed slopes. In every synthetic network the one root slope exceeded the two and three root slopes, with Growth medians of 1.73, 0.664 and 0.503, and all three decreased monotonically in 24/30 Growth, 30/30 ER and 29/30 BA graphs. The river contributes 63 positive and two zero slopes, while 89 groups lack predictor variation. Its observed slopes rise between adjacent groups 18 times. These departures from a simple inverse trend are consistent with unequal calibration, as equation (4) allows. Under complete observation, the groups have a direct physical meaning. At…
Figure 6 of the paper
Figure 6. Input tolerance separates failure initiation, propagation to outcomes and total loss.
Figure 7 of the paper
Figure 7. Predicting loss among unsampled terminals under strict support (a) and tolerance τ = 1/2 (b). Top 5% oracle efficiency divides mean loss from independent cuts of selected edges by the greatest mean loss attainable at the same budget with complete information. Rows compare networks on a common efficiency axis. Orange circles show pair thickness and blue diamonds coverage corrected for upstream root count; connecting lines highlight their difference. Purple squares denote node pair projection count (NPPC), gray ticks the seven other structural comparators, open diamonds observed…
Figure 8 of the paper
Figure 8. tests predictions frozen before generation against 2,560 independent test networks, 512 per prefer￾ence, each with three roots, five generations of 20 nodes and samples of 2, 6 or all 20 terminals. The panels compare conditional means and 90% prediction bands, group slopes, the benefit of grouping and prediction band coverage across preferences and sample sizes. These predictions concern a population of networks, with edges weighted by their expected abundance across that population. We use X = nK/m and Y = U/T, where the random counts K and U range over sampled and complete…
Figure 9 of the paper
Figure 9. examines this question in 1,152 independent graphs with three roots and eight generations of 100 nodes at nine preferences. We report the gain from grouping together with grouped accuracy, the fraction of 14
Figure 14 of the paper
Figure 14. and Supplementary Note 11). The empirical references retain stronger pooled correlations, 0.490 in HEP-TH and 0.592 in Bioconductor at their primary samples. In OpenAlex, the median across sampling orders rose from 0.640 to 0.773 as the sample grew from ten to fifty terminals with tracing limited to two steps. Across full references, root count and coverage correlated at −0.050 in HEP-TH and −0.005 in OpenAlex, and temporal associations could be examined only on dated subsets (Supplementary Note 12 and Supplementary

Showing 10 of the paper's 14 figures. Figures are the authors', reproduced from the paper.

References (66)
  1. [1] Ebenhöh, O., Handorf, T. & Heinrich, R. Structural analysis of expanding metabolic networks. Genome Informatics 15, 35–45 (2004). doi:10.11234/gi1990.15.35 on Pithdoi:10.11234/gi1990.15.35
  2. [2] Han, Y ., Li, W., Filko, A., Li, J. & Zhang, F. Genome-wide promoter responses to CRISPR perturbations of regulators reveal regulatory networks in Escherichia coli. Nature Communications 14, 5757 (2023). 10. 1038/s41467-023-41572-4 on Pith
  3. [3] Pichler, A., Pangallo, M., del Rio-Chanona, R. M., Lafond, F. & Farmer, J. D. Forecasting the propagation of pandemic shocks with a dynamic input-output model. Journal of Economic Dynamics and Control 144, 104527 (2022). doi:10.1016/j.jedc.2022.104527 on Pithdoi:10.1016/j.jedc.2022.104527
  4. [4] R Core Team. Writing R Extensions, Section 1.1.3, Package Dependencies. Official manual. Accessed 14 September 2026 on Pith
  5. [5] CRAN Repository Maintainers. CRAN Repository Policy, revision 6875, Source packages. Official policy. Accessed 14 September 2026 on Pith
  6. [6] Y abe, T., García Bulle Bueno, B., Frank, M. R., Pentland, A. & Moro, E. Behaviour-based dependency networks between places shape urban economic resilience. Nature Human Behaviour 9, 496–506 (2025). 10.1038/s41562-024-02072-7 on Pithdoi:10.1038/s41562-024-02072-7
  7. [8] Ferretti, L. et al. Quantifying SARS-CoV-2 transmission suggests epidemic control with digital contact tracing. Science 368, eabb6936 (2020). 10.1126/science.abb6936 on Pithdoi:10.1126/science.abb6936
  8. [9] Carraro, L., Hartikainen, H., Jokela, J., Bertuzzo, E. & Rinaldo, A. Estimating species distribution and abundance in river networks using environmental DNA. Proceedings of the National Academy of Sciences 115, 11724–11729 (2018). 10.1073/pnas.1813843115 on Pithdoi:10.1073/pnas.1813843115
  9. [10] Wu, X. et al. Synchronization in multiplex networks. Physics Reports 1060, 1–54 (2024). 10.1016/j. physrep.2024.01.005 on Pithdoi:10.1016/j
  10. [11] Parshani, R., Buldyrev, S. V . & Havlin, S. Critical effect of dependency groups on the function of net- works. Proceedings of the National Academy of Sciences108, 1007–1010 (2011). 10.1073/pnas.1008404108. 22 Dependency networks on Pithdoi:10.1073/pnas.1008404108
  11. [12] Bashan, A., Parshani, R. & Havlin, S. Percolation in networks composed of connectivity and dependency links. Physical Review E 83, 051127 (2011). 10.1103/PhysRevE.83.051127 on Pithdoi:10.1103/physreve.83.051127
  12. [13] Watts, D. J. A simple model of global cascades on random networks. Proceedings of the National Academy of Sciences 99, 5766-5771 (2002). 10.1073/pnas.082090499 on Pithdoi:10.1073/pnas.082090499
  13. [14] Buldyrev, S. V ., Parshani, R., Paul, G., Stanley, H. E. & Havlin, S. Catastrophic cascade of failures in interdependent networks. Nature 464, 1025-1028 (2010). 10.1038/nature08932 on Pithdoi:10.1038/nature08932
  14. [15] Di Muro, M. A. et al. Cascading Failures in Interdependent Networks with Multiple Supply-Demand Links and Functionality Thresholds. Scientific Reports 7, 15059 (2017). 10.1038/s41598-017-14384-y on Pithdoi:10.1038/s41598-017-14384-y
  15. [16] Ganin, A. A. et al. Operational resilience: concepts, design and analysis. Scientific Reports 6, 19540 (2016). 10.1038/srep19540 on Pithdoi:10.1038/srep19540
  16. [32] Hamedmoghadam, H., Jalili, M., Vu, H. L. & Stone, L. Percolation of heterogeneous flows un- covers the bottlenecks of infrastructure networks. Nature Communications 12, 1254 (2021). 10.1038/ s41467-021-21483-y . 23 Dependency networks on Pith
  17. [36] Chakraborty, A., Reisch, T., Diem, C., Astudillo-Estévez, P . & Thurner, S. Inequality in economic shock exposures across the global firm-level supply network. Nature Communications 15, 3348 (2024). 10.1038/ s41467-024-46126-w on Pith
  18. [38] Su, S., Ramos, M. & Davis, S. BiocPkgTools: Collection of Simple Tools for Learning about Bioconductor Packages. Bioconductor reference manual, version 1.30.1 (2026). Official manual. Accessed 22 September 2026; cited for dependency definitions on Pith
  19. [42] Dorogovtsev, S. N. & Mendes, J. F. F. Evolution of networks with aging of sites. Physical Review E 62, 1842–1845 (2000). 10.1103/PhysRevE.62.1842 on Pithdoi:10.1103/physreve.62.1842
  20. [43] Karrer, B. & Newman, M. E. J. Random Acyclic Networks. Physical Review Letters 102, 128701 (2009). 10.1103/PhysRevLett.102.128701 on Pithdoi:10.1103/physrevlett.102.128701
show 46 more
  1. [45] Burghart, F. Ancestries and Descendants in a Random DAG. 37th International Conference on Proba- bilistic, Combinatorial and Asymptotic Methods for the Analysis of Algorithms (AofA 2026) , Leibniz Inter- national Proceedings in Informatics 381, 7:1–7:14 (2026). 10.4230/LIPIcs.... on Pithdoi:10.4230/lipics.aofa.2026.7
  2. [46] Mariani, M. S., Medo, M. & Zhang, Y .-C. Ranking nodes in growing networks: When PageRank fails. Scientific Reports 5, 16181 (2015). 10.1038/srep16181 on Pithdoi:10.1038/srep16181
  3. [47] Mao, T., Zhang, S., Hanjalic, A. & Wang, H. Estimating nodal spreading influence using partial temporal networks. Chaos, Solitons & Fractals 201, 117163 (2025). 10.1016/j.chaos.2025.117163 on Pithdoi:10.1016/j.chaos.2025.117163
  4. [50] Y oo, E., Craighead, C. W. & Samtani, S. Dependency Network Structure and Security Vulnerabilities in Software Supply Chains. Journal of Management Information Systems 42, 1149–1176 (2025). 10.1080/ 07421222.2025.2561385 on Pith
  5. [51] Diem, C., Borsos, A., Reisch, T., Kertész, J. & Thurner, S. Quantifying firm-level economic systemic risk from nation-wide supply networks. Scientific Reports 12, 7719 (2022). 10.1038/s41598-022-11522-z on Pithdoi:10.1038/s41598-022-11522-z
  6. [52] Elliott, M., Golub, B. & Leduc, M. V . Supply Network Formation and Fragility. American Economic Review 112, 2701-2747 (2022). 10.1257/aer.20210220 on Pithdoi:10.1257/aer.20210220
  7. [53] Schäfer, B., Witthaut, D., Timme, M. & Latora, V . Dynamically induced cascading failures in power grids. Nature Communications 9, 1975 (2018). 10.1038/s41467-018-04287-5 on Pithdoi:10.1038/s41467-018-04287-5
  8. [54] Erdős, P . & Rényi, A. On random graphs. I. Publicationes Mathematicae Debrecen 6, 290–297 (1959). doi:10.5486/PMD.1959.6.3-4.12. 24 Dependency networks on Pithdoi:10.5486/pmd.1959.6.3-4.12
  9. [55] Gilbert, E. N. Random graphs. The Annals of Mathematical Statistics 30, 1141–1144 (1959). Original article on Pith
  10. [56] Barabási, A.-L. & Albert, R. Emergence of scaling in random networks. Science 286, 509–512 (1999). doi:10.1126/science.286.5439.509 on Pithdoi:10.1126/science.286.5439.509
  11. [57] Stanford Network Analysis Project. High-energy physics theory citation network . Official dataset page . Accessed 12 September 2026. Acknowledgements This work was supported by the National Natural Science Foundation of China under Grants 62573301 and 62503335, the Startup Gran... on Pith
  12. [58] These quantities measure path participation or probabilistic connectivity, whereas equation (3) relates distinct endpoint pairs to the loss of designated outcomes after one necessary relation fails. Serial subdivision and process detail. Replace e = ( u → v) in G by e1 = ( u →... on Pith
  13. [59] We used the glucose-medium responses and the independent, author-organized RegulonDB binding an- notation with explicit TF identifiers and operon mapping. The positive-contribution projection keeps positive annotations between distinct operons, deduplicates di- rected pairs an... on Pith
  14. [60] Priem, J., Piwowar, H. & Orr, R. OpenAlex: A fully-open index of scholarly works, authors, venues, institutions, and concepts. 26th International Conference on Science, Technology and Innovation Indicators (STI 2022) (2022). doi:10.5281/zenodo.6936227 on Pithdoi:10.5281/zenodo.6936227
  15. [61] Taube, J. C., Miller, P . B. & Drake, J. M. An open-access database of infectious disease transmission trees to explore superspreader epidemiology. PLOS Biology 20, e3001685 (2022). 10.1371/journal.pbio.3001685 on Pithdoi:10.1371/journal.pbio.3001685
  16. [62] Hummon, N. P . & Doreian, P . Connectivity in a citation network: The development of DNA theory.Social Networks 11, 39–63 (1989). doi:10.1016/0378-8733(89)90017-8 on Pithdoi:10.1016/0378-8733(89)90017-8
  17. [63] Batagelj, V . Efficient Algorithms for Citation Network Analysis . arXiv:cs/0309023, version 1 (2003). Original preprint on PithPith's reviewarXiv
  18. [64] Brandes, U. On variants of shortest-path betweenness centrality and their generic computation. Social Networks 30, 136–145 (2008). doi:10.1016/j.socnet.2007.11.001 on Pithdoi:10.1016/j.socnet.2007.11.001
  19. [65] Sabrin, K. M. & Dovrolis, C. The hourglass effect in hierarchical dependency networks. Network Science 5, 490–528 (2017). 10.1017/nws.2017.22 on Pithdoi:10.1017/nws.2017.22
  20. [66] Šubelj, L., Waltman, L., Traag, V . & van Eck, N. J. Intermediacy of publications. Royal Society Open Science 7, 190207 (2020). 10.1098/rsos.190207 on Pithdoi:10.1098/rsos.190207
  21. [67] Borgs, C., Brautbar, M., Chayes, J. & Lucier, B. Maximizing Social Influence in Nearly Optimal Time. Proceedings of the Twenty-Fifth Annual ACM-SIAM Symposium on Discrete Algorithms , 946–957 (2014). 10.1137/1.9781611973402.70 on Pithdoi:10.1137/1.9781611973402.70
  22. [68] Tang, Y ., Xiao, X. & Shi, Y . Influence Maximization: Near-Optimal Time Complexity Meets Practical Efficiency. Proceedings of the 2014 ACM SIGMOD International Conference on Management of Data , 75– 86 (2014). 10.1145/2588555.2593670 on Pithdoi:10.1145/2588555.2593670
  23. [69] Kempe, D., Kleinberg, J. & Tardos, É. Maximizing the spread of influence through a social network. In Proceedings of the Ninth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining 137–146 (ACM, 2003). 10.1145/956750.956769 on Pithdoi:10.1145/956750.956769
  24. [70] Abate, P ., Boender, J., Di Cosmo, R. & Zacchiroli, S. Strong dependencies between software components. In 2009 3rd International Symposium on Empirical Software Engineering and Measurement 89–99 (IEEE, 2009). 10.1109/ESEM.2009.5316017 on Pithdoi:10.1109/esem.2009.5316017
  25. [71] Setó-Rey, D., Santos-Martín, J. I. & López-Nozal, C. Vulnerability of Package Dependency Networks. IEEE Transactions on Network Science and Engineering 10, 3396–3408 (2023). 10.1109/TNSE.2023. 3260880. 72 Dependency networks Supplementary information on Pithdoi:10.1109/tnse.2023
  26. [72] Marichal, J.-L. On Sugeno integral as an aggregation function. Fuzzy Sets and Systems 114, 347–365 (2000). 10.1016/S0165-0114(98)00116-X on Pithdoi:10.1016/s0165-0114(98)00116-x
  27. [73] Lengauer, T. & Tarjan, R. E. A fast algorithm for finding dominators in a flowgraph. ACM Transactions on Programming Languages and Systems 1, 121-141 (1979). 10.1145/357062.357071 on Pithdoi:10.1145/357062.357071
  28. [74] Janson, S. The number of descendants in a random directed acyclic graph. Random Structures & Algo- rithms 64, 768–803 (2024). 10.1002/rsa.21195 on Pithdoi:10.1002/rsa.21195
  29. [75] Burghart, F. Ancestries in random d-DAGs. arXiv preprint (2026). arXiv:2606.30475v1. arXiv record on PithPith's reviewarXiv
  30. [76] Mailler, C. & Steiner, R. Multi-drawing Pólya urns via labelled random DAGs. arXiv preprint (2025). arXiv:2508.20592v1. 10.48550/arXiv.2508.20592 on PithPith's reviewarXivdoi:10.48550/arxiv.2508.20592
  31. [77] Janson, S. & Lo, T. Y . Y . The number of descendants in a preferential attachment graph. arXiv preprint (2024). arXiv:2412.13975v1. 10.48550/arXiv.2412.13975 on PithPith's reviewarXivdoi:10.48550/arxiv.2412.13975
  32. [78] Ericson, W. A. Subjective Bayesian Models in Sampling Finite Populations. Journal of the Royal Statis- tical Society: Series B (Methodological) 31, 195–224 (1969). 10.1111/j.2517-6161.1969.tb00782.x on Pithdoi:10.1111/j.2517-6161.1969.tb00782.x
  33. [79] Rodríguez, C. E., Walker, S. G. & Mena, R. H. A finite population parametric Bayesian bootstrap for quick counts. The Annals of Applied Statistics 19, 3394–3415 (2025). 10.1214/25-AOAS2066 on Pithdoi:10.1214/25-aoas2066
  34. [80] Dall’ Asta, L., Alvarez-Hamelin, I., Barrat, A., Vázquez, A. & Vespignani, A. Exploring networks with traceroute-like probes: Theory and simulations. Theoretical Computer Science 355(1), 6–24 (2006). 10. 1016/j.tcs.2005.12.009 on Pith
  35. [81] Achlioptas, D., Clauset, A., Kempe, D. & Moore, C. On the bias of traceroute sampling: Or, power-law degree distributions in regular graphs. Journal of the ACM 56(4), 21:1–21:28 (2009). 10.1145/1538902. 1538905 on Pithdoi:10.1145/1538902
  36. [82] Kim, M. & Leskovec, J. The Network Completion Problem: Inferring Missing Nodes and Edges in Networks. Proceedings of the 2011 SIAM International Conference on Data Mining , 47–58 (2011). 10. 1137/1.9781611972818.5 on Pith
  37. [83] Peixoto, T. P . Reconstructing Networks with Unknown and Heterogeneous Errors. Physical Review X 8, 041011 (2018). 10.1103/PhysRevX.8.041011 on Pithdoi:10.1103/physrevx.8.041011
  38. [84] Peixoto, T. P . Network Reconstruction via the Minimum Description Length Principle. Physical Review X 15, 011065 (2025). 10.1103/PhysRevX.15.011065 on Pithdoi:10.1103/physrevx.15.011065
  39. [85] Y oung, J.-G., Valdovinos, F. S. & Newman, M. E. J. Reconstruction of plant–pollinator networks from observational data. Nature Communications 12, 3911 (2021). 10.1038/s41467-021-24149-x on Pithdoi:10.1038/s41467-021-24149-x
  40. [86] Peel, L., Peixoto, T. P . & De Domenico, M. Statistical inference links data and theory in network science. Nature Communications 13, 6794 (2022). 10.1038/s41467-022-34267-9 on Pithdoi:10.1038/s41467-022-34267-9
  41. [87] Reisch, T., Heiler, G., Diem, C., Klimek, P . & Thurner, S. Monitoring supply networks from mobile phone data for estimating the systemic risk of an economy. Scientific Reports 12, 13347 (2022). 10.1038/ s41598-022-13104-5 on Pith
  42. [88] Fessina, M. et al. Inferring firm-level supply chain networks with realistic systemic risk from industry sector-level data. Scientific Reports 16, 19848 (2026). 10.1038/s41598-026-47883-y on Pithdoi:10.1038/s41598-026-47883-y
  43. [89] Hamedmoghadam, H., Jalili, M., Vu, H. L. & Stone, L. Percolation of heterogeneous flows un- covers the bottlenecks of infrastructure networks. Nature Communications 12, 1254 (2021). 10.1038/ s41467-021-21483-y on Pith
  44. [90] Chulahwat, A., Hassan, E. M., Tariverdi, M., Nunez-del-Prado, M. & Mahmoud, H. Identifying influential assets in higher order interdependent infrastructure networks through population impact. Scientific Reports 15, 33351 (2025). 10.1038/s41598-025-15824-w on Pithdoi:10.1038/s41598-025-15824-w
  45. [91] Jiang, X. & Zhuge, H. Forward search path count as an alternative indirect citation impact indicator. Journal of Informetrics 13, 100977 (2019). 10.1016/j.joi.2019.100977 on Pithdoi:10.1016/j.joi.2019.100977
  46. [92] Han, Y ., Li, W., Filko, A., Li, J. & Zhang, F. Genome-wide promoter responses to CRISPR perturbations of regulators reveal regulatory networks in Escherichia coli. Nature Communications 14, 5757 (2023). 10. 1038/s41467-023-41572-4 . 73 on Pith

Challenges and discussion

Disagree with the review, or know something it missed? Leave a comment, or challenge this review formally: a challenge becomes a permanent entry in this paper's record, and its decision is published here.

No comments yet.

Sign in to comment