Pith. Sign in

REVIEW

Tuning degree distributions of scale-free networks

PaperPDF from arXiv

data files

ReviewThe objections and verdict, as JSON Paper breakdownThe authors' claims, conditions and measurements, as JSON Signed receiptProof this review has not been altered
X Bluesky LinkedIn Reddit Hacker News
C. C. Leary, M. Schwehm, M. Eichner, H. P. Duerr

Data Analysis, Statistics and Probability · Submitted 22 Feb 2006 · 34 references

Pith's verdict: Practical tweak to the Barabasi-Albert algorithm that tunes skewness and clustering; worth citing for modelers, but the calibration is thinner than the abstract claims.

T0 · AI review

How this review was made: An AI referee (deepseek-v4-flash) read the paper on August 28, 2026 and answered a fixed list of questions about it. T0 is the starting level every paper gets. The levels describe how deeply Pith has checked the paper, never the authors.

  1. T0 AI review An AI read the paper and answered a fixed list of questions.
  2. T1 Audit Automatic checks that the references and files the paper relies on exist and match.
  3. T2 Compute The paper's code or calculations re-run with the same results.
  4. T3 Formal statement The claims written out in a formal math language a computer can read.
  5. T4 Proof Those claims proved, and the proof checked by computer.

How Pith checks papers →

· 3 major objections · 6 minor · Ask about this paper · The compiled record · Read the full review →
Challenge this review Comment (0) Read on arXiv ↗
Cite ▾
arXiv IDphysics/0602152 (version 2)
Pith IDJSJZOG4G, Pith's permanent record of this paper and its review
CitationC. C. Leary, H. P. Duerr, M. Eichner, M. Schwehm. "Tuning degree distributions of scale-free networks." arXiv:physics/0602152 (2006). https://pith.science/paper/JSJZOG4G
Cite this reviewBibTeX for Pith's review of this paper
Claim paper

Why Pith says this

Highlight any text to comment on it.

Referee report

How Pith checks papers → · Full audit of this paper →

The paper

The paper claims

A single tuning distribution in preferential attachment sets the skewness of a scale-free network.

The reading

The paper generalizes the standard preferential attachment construction of scale-free networks by drawing the attachment random number from a tuning distribution with expectation $\mu_T$ rather than from a uniform distribution. Lower $\mu_T$ favors high-degree vertices and yields overskewed, hub-dominated degree distributions; higher $\mu_T$ favors low-degree vertices and yields underskewed, flatter distributions. Adding a triad formation step with probability $c$ gives control over the clustering coefficient. The payoff, if the claim holds, is that a modeler can grow a network with a targeted tail shape and a targeted level of local clustering using one intuitive growth process, which matters for simulations of transmission across contact networks that are not perfectly scale-free.

What carries the argument

The machinery is the tuning distribution, a probability density on $[0,1]$ used in place of the uniform random number that selects the attachment target. In the Barabási–Albert rule a vertex $w$ is chosen by comparing $r$ against the cumulative degree sums of vertices sorted by degree, so the expectation $\mu_T$ of the tuning distribution directly biases the choice: $\mu_T < 1/2$ overskews by favoring hubs, $\mu_T > 1/2$ underskews by favoring small-degree vertices. The second ingredient is a triad formation step run with probability $c$, which closes triangles by linking the new vertex to a random neighbor of its target and thereby raises the mean clustering coefficient. Together $\mu_T$ and $c$ are the two knobs the paper turns.

Extended reading notes

Core claim

The paper's central claim is that replacing the uniform draw in Barabási–Albert preferential attachment with a tunable distribution $f_m$ on $[0,1]$ of expectation $\mu_T$ makes the skewness of the resulting degree distribution a controlled, approximately linear function of $\mu_T$. For the linear family the measured cube-root skewness follows $\gamma_D = -9.7\mu_T + 7.4$ at $c=0$, and the slope depends on $c$. Skewness tracks $\log D_{\max}$ with $r^2 = 0.987$, so the tail shape of these networks is essentially summarized by the largest degree. With $\beta$-distributed tuning the method can produce non-monotonic degree distributions, including U-shaped ones that create many low-degree vertices and extreme hubs at the same time.

Load-bearing premise

The load-bearing premise is that the empirically fitted lines, e.g. $\gamma_D = -9.7\mu_T + 7.4$, calibrated on 10,000-vertex networks with $\mu_T$ between 0.4 and 0.6, continue to hold for other network sizes, wider $\mu_T$ ranges, and $\beta$ tuning distributions; the paper does not test this.

Editorial extensions

If this is right

  • Choosing $\mu_T < 1/2$ produces networks with fatter tails and prominent hubs; choosing $\mu_T > 1/2$ produces flatter degree distributions, all within the same growing-network construction.
  • The clustering probability $c$ changes the mean clustering coefficient over a wide range while leaving the inverse $\mu_T$-versus-skewness ordering intact, so tail shape and local clustering can be set somewhat independently.
  • Because $\gamma_D$ and $\log D_{\max}$ are nearly interchangeable ($r^2=0.987$), users can report either the cube-root skewness or the largest degree as a summary of tail variability.
  • Beta tuning distributions allow non-monotonic and U-shaped degree distributions, extending the method beyond the linear family.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the linear calibration were to hold at larger sizes, the algorithm would give a direct way to batch-produce ensembles with prescribed skewness for epidemic modeling; that scaling is not tested.
  • The U-shaped beta case hints at a two-class contact structure, many low-degree vertices plus extreme hubs, which could be a better stylization of superspreader networks; the paper illustrates but does not develop this.
  • The near-equivalence of skewness and log maximum degree suggests the tuning effect may be driven by the single largest hub, meaning degree-distribution skewness here is less a shape property than an extreme-value statistic; that distinction is not pursued.
Original abstract

Scale-free networks are characterized by a degree distribution with power-law behavior and have been shown to arise in many areas, ranging from the World Wide Web to transportation or social networks. Degree distributions of observed networks, however, often differ from the power-law type and data based investigations require modifications of the typical scale-free network. We present an algorithm that generates networks in which the skewness of the degree distribution is tuneable by modifying the preferential attachment step of the Barabasi-Albert construction algorithm. Skewness is linearly correlated with the maximal degree of the network and, therefore, adequately represents the influence of superspreaders or hubs. By combining our algorithm with work of Holme and Kim, we show how to generate networks with skewness gamma and clustering coefficient kappa, over a wide range of values.

The compiled record

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. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-09-23: 1 request, 4,301 input tokens.

What the abstract itself states, as probabilities: claims first 0.04 · states limitation 0.12 · human participants 0.03 · controlled comparison 0.04 · randomized assignment 0.01 · states quantitative result 0.39 · compares with prior results 0.59 · mentions code or data availability 0.03.

The abstract, sentence by sentence

arXiv listing abstract v2. Results and the main claim are underlined; .

Scale-free networks are characterized by a degree distribution with power-law behavior and have been shown to arise in many areas, ranging from the World Wide Web to transportation or social networks. Degree distributions of observed networks, however, often differ from the power-law type and data based investigations require modifications of the typical scale-free network. We present an algorithm that generates networks in which the skewness of the degree distribution is tuneable by modifying the preferential attachment step of the Barabasi-Albert construction algorithm. Skewness is linearly correlated with the maximal degree of the network and, therefore, adequately represents the influence of superspreaders or hubs. By combining our algorithm with work of Holme and Kim, we show how to generate networks with skewness gamma and clustering coefficient kappa, over a wide range of values.

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#sentencerolethe authors' own workstates the main conclusionstrength
s0 Scale-free networks are characterized by a degree distribution with power-law behavior and have been shown to arise in many areas, ranging from the World Wide Web to transportation or social networks. background 1.00 0.04 0.02 asserted 1.00
s1 Degree distributions of observed networks, however, often differ from the power-law type and data based investigations require modifications of the typical scale-free network. background 0.66 0.08 0.03 asserted 0.93
s2 We present an algorithm that generates networks in which the skewness of the degree distribution is tuneable by modifying the preferential attachment step of the Barabasi-Albert construction algorithm. contribution 0.67 0.96 0.46 asserted 1.00
s3 Skewness is linearly correlated with the maximal degree of the network and, therefore, adequately represents the influence of superspreaders or hubs. result 0.60 0.76 0.13 asserted 1.00
s4 By combining our algorithm with work of Holme and Kim, we show how to generate networks with skewness gamma and clustering coefficient kappa, over a wide range of values. contribution 0.65 0.92 0.51 asserted 1.00

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from simulation or computation 0.98.

What was examined, and what was not

sources read
  • arXiv listing abstract v2
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

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

Figures
Figure 1. FIG. 1. Examples of degree distributions of different networks, for a subset of tuning distributions (f(x), column 1) and clustering parameters c (columns 2-4, with c = 0 , c =0.5 , c = 1). Moments for the tuning distribution (subscript T ) and for the degree distributions (subscript D ) are represented by µ : mean, σ : standard deviation and γ : cube root of the skewness. Scalefree degree distributions with power-law behavior result from the special case = 1/ 2 µT , representing a uniform tuning distribution. Underskewed degree distributions result from > 1/ 2 µT and overskewed…
Figure 2. FIG. 2. Skewness distributions of different networks, grouped by the mean µT of the tuning distribution and clustering parameter c . For each of the 25 parameter combinations, 100 network realizations were simulated. The skewness distribution of the corresponding degree distributions is represented by a box & whiskers plot, diamonds (mean ± 1.5 standard error of the mean) and outliers (dots). Quartiles are represented by the lower and upper bounds of the box with the median as the line between them. Whiskers represent the quartiles ± 1.5 interquartile range and outliers are defin
Figure 3. FIG. 3. Relationship between κ , the mean clustering coefficient, and c , the clustering parameter. For each value of c ∈{0, 0.05, 0.1, K , 1} and each value of µT , 10 networks were generated. Tuning distributions with differing expectation µT are indicated by differing markers as shown in the inset. Exponential least squares fits are grouped by µT
Figure 4. FIG. 4. Relationship between κ , the mean clustering coefficient and γ , the root skewness of the degree distribution, over 250 networks generated using the parameter values described in the main text. The horizontal bands correspond to the different values of the clustering parameter c ∈{0, 0.25, 0.5, 0.75, 1.0 } (from bottom to top). Tuning distributions with differing expectation µT are indicated by differing markers as shown in the inset
Figure 5. FIG. 5. Examples for degree distributions (ri

Figures reproduced from arXiv:physics/0602152 by the authors.

References

Discussion

Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.