Scalefree networks are rare
Abstract
A central claim in modern network science is that realworld networks are typically “scale free,” meaning that the fraction of nodes with degree follows a power law, decaying like , often with . However, empirical evidence for this belief derives from a relatively small number of realworld networks. We test the universality of scalefree structure by applying stateoftheart statistical tools to a large corpus of nearly 1000 network data sets drawn from social, biological, technological, and informational sources. We fit the powerlaw model to each degree distribution, test its statistical plausibility, and compare it via a likelihood ratio test to alternative, nonscalefree models, e.g., the lognormal. Across domains, we find that scalefree networks are rare, with only 4% exhibiting the strongestpossible evidence of scalefree structure and 52% exhibiting the weakestpossible evidence. Furthermore, evidence of scalefree structure is not uniformly distributed across sources: social networks are at best weakly scale free, while a handful of technological and biological networks can be called strongly scale free. These results undermine the universality of scalefree networks and reveal that realworld networks exhibit a rich structural diversity that will likely require new ideas and mechanisms to explain.
Networks are a powerful way to both represent and study the structure of complex systems of all kinds. Examples today are plentiful and include social interactions among individuals, both offline and online, protein or gene interactions in biological organisms, communication between digital computers, and various kinds of transportation networks. Across scientific domains and different types of networks, it is common to encounter the claim that most or all realworld networks are scale free. The precise details of this claim vary across the literature Albert et al. (1999); Pržulj (2007); LimaMendez and van Helden (2009); Mislove et al. (2007); Agler et al. (2016); Ichinose and Sayama (2017); Zhang et al. (2015), but generally agree that a network is scale free if the fraction of nodes with degree follows a powerlaw distribution , where . Some versions of this “scalefree hypothesis” make the requirements stronger, e.g., requiring that or that node degrees evolve by the preferential attachment mechanism Dorogovtsev and Mendes (2002); Barabási and Albert (1999). Other versions make them weaker, e.g., requiring that the power law holds only in the upper tail Willinger et al. (2009) or is merely more plausible than a thintailed distribution like an exponential or normal Albert et al. (2000).
Despite the frequency of these claims, there has been no broad evaluation of the empirical prevalence of scalefree patterns in realworld networks. If networks are in fact universally scale free, then many theoretical results about scalefree networks have broad scientific relevance. For instance, the several mechanisms known to build scalefree networks Price (1965); Simon (1955); Barabási and Albert (1999); Dorogovtsev and Mendes (2002); PastorSatorras et al. (2003); Berger et al. (2004); Leskovec et al. (2007); LimaMendez and van Helden (2009) would provide a common basis for understanding network assembly. And, such mechanisms could be used to create realistic synthetic networks for numerical simulations and experiments. Moreover, many studies have investigated how scalefree structure shapes the dynamics of processes running over the network. For example, scalefree random graphs are almost always susceptible to epidemics Newman (2002), a result with profound implications for social influence and public health applications. The universality of scalefree networks is thus a critically important question, and resolving its empirical status would inform efforts to apply results from network theory to many scientific questions.
If realworld networks are not universally or even typically scalefree, the status of a unifying theme in network science over nearly 20 years Albert et al. (1999); Barabási and Albert (1999); Carlson and Doyle (1999); Newman (2005); Mitzenmacher (2003) would be significantly diminished. Such an outcome would require a careful reassessment of the large literature that has grown up around the idea. Even if scalefree networks are not universal, their prevalence may be nonuniform across different domains of networks, e.g., it may be common for social networks but rare for biological networks to be scale free. Thus, a crucial followup question would be to assess this differential evidence by domain, and to characterize how realworld structures deviate from the scalefree pattern. For domains where there is little empirical support or where scalefree networks are relatively unusual, new models of structure and new mechanisms of assembly may be needed.
The validity and scope of the scalefree hypothesis is not uncontroversial. Some research has argued against its universality, on either statistical or theoretical grounds Tanaka (2005); Li et al. (2005); Pržulj (2007); LimaMendez and van Helden (2009); Willinger et al. (2009); Stumpf and Porter (2012); Golosovsky (2017). There have also been passionate and personal defenses of its status Barabasi (2016), along with many findings claiming to validate or support its universality Mislove et al. (2007); Agler et al. (2016); Goh et al. (2002); PastorSatorras et al. (2015); Gamermann et al. (2017). This controversy has persisted because studies have typically relied upon small, often domainspecific data sets, less rigorous statistical methods, and differing definitions “scale free” Redner (1998); Mislove et al. (2007); Agler et al. (2016); Goh et al. (2002); Ichinose and Sayama (2017); PastorSatorras et al. (2015); Zhang et al. (2015); Pachon et al. (2017). Additionally, relatively few studies have performed statistically rigorous comparisons of a fitted powerlaw distribution to alternative, nonscalefree distributions, e.g., the lognormal or the stretched exponential, which can imitate a powerlaw form in realistic sample sizes Clauset et al. (2009).
For example, Eikmeier and Gleick Eikmeier and Gleich (2017) recently investigated a number of realworld networks using rigorous methods for fitting and testing for the plausibility of power laws Clauset et al. (2009) in the singular values of the adjacency matrix, the eigenvalues of the Laplacian, and the degree distribution. Although they argue that they find broad evidence of scalefree structure in these networks, the evidence remains ambiguous in two crucial ways. First, the statistical plausibility of scalefree structure is strongest in the singular and eigenvalue analyses rather than in the degree distributions, which is a different kind of scalefree structure than the hypothesis typically posits. Second, their analyses did not include controls for spurious conclusions due to small sample sizes or comparisons against alternative distributions, both of which are necessary to reduce the likelihood of false findings of scalefree structure Clauset et al. (2009).
Beyond statistical concerns, the literature on scalefree networks also exhibits conflicting interpretations of the theoretical meaning of observing a powerlaw degree distribution. Many studies interpret this pattern as direct evidence that a network was assembled by the preferential attachment mechanism, or by one of its variations. As a result, the term “scalefree network” can confusingly refer to either a network with a powerlaw degree distribution or a network built by preferential attachment. This ambiguity has not helped resolve the controversy over the universality of scalefree networks, as a number of alternative mechanisms also produce powerlaw degree distributions without relying on preferential attachment Carlson and Doyle (1999); Newman (2005); Mitzenmacher (2003). This ambiguity highlights a simple fact: degree distributions reflect modest constraints on overall network structure Alderson and Li (2007) and are thus weak measures by which to identify generating mechanisms Mitzenmacher (2004).
Here, we carry out a broad test of the universality of scalefree networks by applying stateoftheart statistical methods to a large and diverse corpus of realworld networks. To account for the variability in how scalefree networks have been defined in the literature, we formalize a set of quantitative criteria that represent differing strengths and types of evidence for scalefree structure in a particular network. These sets of criteria allow us to extend the scalefree hypothesis beyond the domain of simple (unweighted, undirected, static) graphs and thereby conduct a more complete evaluation of its universality.
For each of 927 network data sets drawn from all domains of science, we estimate the bestfitting powerlaw model, test its statistical plausibility, and then compare it via a likelihood ratio test to alternative nonscalefree distributions. We analyze these results collectively, and then consider how the evidence for scalefree structure varies across domains. We conclude with a forwardlooking discussion of the empirical status of the scalefree hypothesis and offer suggestions for future research on the structure of networks.
I Network data sets
A key component of our evaluation of the scalefree hypothesis is the construction of a structurally and scientifically diverse corpus of realworld networks. We drew 927 network data sets from the Index of Complex Networks (ICON), a comprehensive online index of researchquality network data sets, spanning all fields of science Clauset et al. (2016). This corpus includes networks from biological, informational, social, technological, and transportation domains, and which range in size from hundreds to millions of nodes.
Because of natural class imbalance in the ICON index, the distribution of networks across domains within our corpus is not uniform (Table 1): roughly half are biological networks, a third are social or technological networks, and the remaining are informational or transportation networks. These networks span five orders of magnitude in size, are generally sparse, with a typical mean degree of (Fig. 1), and generally exhibit heavytailed degree distributions.
Domain  Num(Prop)  Multip.  Bip.  Multig.  Weigh.  Dir.  Simp. 
Bio.  500 (0.54)  277  41  384  29  37  39 
Inf.  15(0.02)  0  0  3  0  5  7 
Social  145(0.16)  4  0  5  3  0  136 
Tech.  200(0.21)  122  0  3  1  192  5 
Trans.  67 (0.070)  48  0  65  3  2  0 
Total  927(1.00)  451  41  460  36  236  187 
The scalefree property can be clearly defined for simple graphs, that is, for networks that are static, unipartite, unweighted, and undirected. For such networks, the degree distribution can be obtained unambiguously. However, many realworld networks are not simple, which complicates the extraction of a degree distribution.
Our ICON corpus contains many networks that have various combinations of directed, weighted, bipartite, multigraph, temporal, and multiplex properties. For each property, there can be multiple ways to extract a degree distribution and hence multiple ways to potentially test the scalefree hypothesis. We resolve these ambiguities by shifting perspectives and applying a consistent procedure. First, we define a network data set to be a particular network in our corpus, e.g., a directed, weighted, multiplex network. Then, for each network data set, we extract a set of simple graphs and their corresponding degree sequences, each of which can be tested for the scalefree property. Under this procedure, a directed network would produce three simple graphs and thus three degree sequences (Fig. 2), one for each of the indegree sequence, the outdegree sequence, and the undirected degree sequence. Appendix A gives complete details of the procedure for extracting a set of simple graphs from a network data set with arbitrary properties.
Applying this procedure to our corpus of 927 network data sets, we obtain 4477 simple graphs that serve as the input to our statistical evaluation. We exclude from this set any graph with either very small mean degree or very large mean degree . This filter removes 7376 extremely sparse and 12,146 extremely dense graphs, both of which are highly unlikely to be scale free. Some network data sets produce more simple graphs than others. For instance, an unweighted, directed temporal network with snapshots will produce simple graphs: 3 graphs for each of the snapshots, and 3 graphs for the timeaggregation of all the snapshots. Complicated network data sets can thus produce a combinatoric number of simple graphs. Hence, treating every graph independently in our evaluation of the scalefree hypothesis could lead to skewed results, e.g., if a few nonscalefree data sets account for a large fraction of the total extracted simple graphs. We control for this possibility by reporting our findings in terms of network data sets, rather than for individual graphs.
Ii Assessing Empirical Evidence
All versions of the scalefree hypothesis include some kind of statement about a network’s degree structure following a powerlaw distribution. Hence, a first methodological requirement for a broad evaluation of the empirical evidence for or against the universality of scalefree networks is the ability to test whether a particular degree sequence is consistent with a powerlaw distribution. For this question, we employ stateoftheart statistical methods Clauset et al. (2009), which allow us to choose the bestfitting power law, test its statistical plausibility, and compare the quality of its fit to alternative nonscalefree distributions.
A second methodological requirement stems from our expansion of the scalefree hypothesis to nonsimple networks. Because a given network data set can produce an entire set of degree sequences, we define five sets of criteria, which represent differing degrees of evidence that a particular network data set exhibits scalefree structure. Data sets that fail to meet any of these criteria are deemed Not Scale Free. Four sets of these criteria are nested, and represent the iterative addition of different characteristics, all found in the network science literature, about the properties of the powerlaw degree distributions. The fifth set represents the weakest possible version of the scalefree hypothesis, in which it is only required that the nonscalefree distributions are relatively worse models of the observed degrees than is the power law. Meeting this criterion does not provide direct evidence for the presence of scalefree structure, as networks that meet this criteria alone could still reject the power law as a direct model of their degree sequence.
ii.1 Fitting and comparing degree distribution models
For each degree sequence , we estimate the bestfitting powerlaw distribution model, where
(1) 
meaning that the powerlaw form holds above some minimum value . In this form, is the scaling exponent and is the normalization constant.
Degree distributions are integervalued and hence we let be discrete. In the literature, it is commonly assumed that the powerlaw form may not hold for the smallest degree vertices. To allow for this possibility, we jointly estimate the parameter for a powerlaw model of the upper tail of the degree distribution, and the minimum value for which that model fits Clauset et al. (2009). Technical details of this estimation procedure are given in Appendix B.
Fitting the powerlaw distribution will always return some values for and , but these give no indication of whether the fitted model is a statistically plausible explanation of the data. To assess this property of the fitted model, we use a goodnessoffit test (see Appendix B) to obtain a standard value. Following standard practice in this setting Clauset et al. (2009), if , then we deem the degree sequence to be plausibly scale free, while if , we reject the scalefree hypothesis.
Once the powerlaw model has been fitted to the degree sequence, and its plausibility evaluated, we consider whether it is a better fit to the data than several alternative, nonscalefree distributions. This step is crucial because the inability to reject the powerlaw hypothesis does not guarantee that the powerlaw is the best description of a degree sequence. To determine whether a scalefree distribution is better than alternative heavytailed distributions, we compared the fitted power law to the (i) exponential, (ii) lognormal, (iii) powerlaw with exponential cutoff, and (iv) stretched exponential or Weibull distributions (Table 2), all of which have been used previously as models of degree distributions Amaral et al. (2000); Buzsáki and Mizuseki (2014); Jeong et al. (2001); Malevergne et al. (2005); DuBois et al. (2012). Each alternative model is fitted via maximum likelihood to the empirical degrees , with given by the powerlaw fit, which ensures that the comparison with the powerlaw model is not unfair Clauset et al. (2009).
The powerlaw fit is compared to alternatives using a likelihood ratio test (see Appendix D), which uses the difference in loglikelihoods between the powerlaw and alternative models as a test statistic:
where is the loglikelihood of the powerlaw model and is the loglikelihood of a particular alternative model. The sign of thus indicates which of the two models is favored, power law () or alternative . An important property of the likelihood ratio test, however, is a third outcome: , which indicates that the data do not permit a distinction between the models, i.e., neither is favored over the other.
Because is derived from data, however, it is itself a random variable and thus subject to statistical variations Clauset et al. (2009); Vuong (1989). As a result, the sign of is only meaningful if we can determine that is far enough from to provide a clear conclusion. The standard solution for determining how to interpret is another hypothesis test, in which we calculate a value against a null model of . If , the sign of is not informative and the data cannot tell us which model, power law or alternative, is a better fit. If , then the data provide a clear conclusion in favor of one model or the other.
In order to report results at the level of an entire network data set, we apply the likelihood ratio tests to all the associated simple graphs and then aggregate the results. Specifically, for each alternative distribution, we count the number of simple graphs associated with a particular network data set in which the outcome favored the alternative, favored the power law, or had an inconclusive result. Normalizing these counts across outcome categories provides a continuous measure of the relative evidence that the data set falls into each of category.
ii.2 Definitions of a scalefree network
The methods described above can be applied to determine whether the degree sequence of a particular simple graph is convincingly scale free. However, because network data sets may produce many individual graphs as a result of simplification, the result for any one of several graphs is not sufficient evidence that the entire network data set is or is not convincingly scale free. Instead, the status of a network data set should depend on the results derived from all of its corresponding simple graphs. To capture the different levels and types of evidence of scalefree structure that our battery of tests may produce, we define five sets of criteria for a network data set (Fig. 3).
The weakestpossible definition of a scalefree network, which we call SuperWeak, is one in which the powerlaw form is merely a better description of the degree distribution than alternatives, regardless of whether the power law itself is a convincing model of the degrees. A network data set classified as SuperWeak is not necessarily scale free at all, and membership in this category simply indicates that the degree structure of the network data set is not obviously not scalefree. Formally, we define this category representing indirect evidence of scalefree structure as follows.

SuperWeak: For at least of graphs, none of the alternative distributions are favored over the power law.
The four remaining definitions are nested, and represent increasing levels of direct evidence that the degree structure of the network data set is scale free:

Weakest: For at least of graphs, the powerlaw hypothesis cannot be rejected ().

Weak: The requirements of the Weakest set, and there are at least nodes in the distribution’s tail ().

Strong: The requirements of the Weak set, and that both , and for at least of graphs none of the alternative distributions are favored over the powerlaw.

Strongest: The requirements of the Strong set for at least of graphs, rather than 50%, and for at least of graphs none of the alternative distributions are favored over the powerlaw.
The progression from Weakest to Strongest categories represents the addition of more specific properties of the powerlaw degree distribution, all found in the literature on scalefree networks or distributions. Broadly, these properties span requiring that a sufficient number of nodes be in the scaling regime, that the scaling parameter fall in a specific range, and that the powerlaw model be favored over alternatives. Requiring a minimum of 50 points in the tail of the distribution is a fairly weak requirement to rule out cases where we are obviously fitting to noise.
In the Strongest category, the threshold of of graphs is determined by our choice of the Type I error rate for the goodnessoffit hypothesis test. If all of the graphs for a network data set are indeed scale free, the goodnessoffit test will be expected to incorrectly reject the powerlaw model of the time. The choice of for the comparison with the alternative distributions follows a similar rationale.
ii.3 Method validation on synthetic networks
In order to evaluate the accuracy of these methods, we tested them on three classes of synthetic networks with mathematically wellunderstood degree distributions. Two mechanisms produce scalefree networks (a directed version of preferential attachment Easley and Kleinberg (2010) and a directed vertex copy model Newman (2010)), and one does not (simple ErdősRényi random graphs). Applied to synthetic networks generated by these mechanisms, our methodology correctly placed the synthetic data sets into scalefree categories suitable for their generating parameters, i.e., preferential attachment and vertexcopy data sets were categorized as scalefree networks, while ErdősRényi random graphs were not (see Appendix E).
Iii Results
We begin investigating the empirical evidence for the universality of scalefree networks by considering the distribution of estimated powerlaw scaling parameters , both for the corpus overall and for each group of data sets ranging from the SuperWeak to Strongest evidence categories. We then consider how the fitted powerlaw distributions fare relative to alternative distributions across data sets. Finally, we combine these results to form a quantitative assessment of the relative degrees of evidence for scalefree structure across all data sets in the corpus and by data sets drawn from different domains.
iii.1 Powerlaw distributions
A simple summary of each network data set is the median scaling parameter among its corresponding simple graphs. Across the corpus, the overall distribution of median parameters is concentrated around a value of , but also exhibits a long righttail, with of data sets having (Fig. 4). The range is sometimes identified as including the most emblematic of scalefree networks Barabási and Albert (1999), and we find that of network data sets have estimated parameters in this range. However, we also find that nearly of network data sets exhibit a median parameter in the range , which is a relatively unusual value in the scalefree network literature.
However, the powerlaw distributions represented by these small parameters are not necessarily statistically plausible. Hence, the shape or concentration of this overall distribution is not evidence for, or against, the universality of scalefree networks. A simple check of our results is whether there is a clear relationship between the size of a network and the median powerlaw parameter —a strong correlation, in either positive or negative directions, may be indicative of a systematic bias in our methodology. Instead, we find barely any correlation between and , with (), indicating little evidence of systematic bias (Fig. 5).
Across the five categories of evidence for scalefree structure, the distribution of median parameters varies considerably (Fig. 4, insets). For data sets that fall into the SuperWeak category, we find a distribution with breadth similar to that of the overall distribution, with a long righttail and many data sets with . Only a handful of data sets with exhibit even SuperWeak or Weakest evidence of scalefree structure. In our corpus, these particular data sets correspond to planar networks, representing mycelial fungal or slime mold growth patterns Lee et al. (2017). Even among the Weakest and Weak categories, the distribution of median parameters remains broad, with a a substantial fraction exhibiting . The Strong and Strongest categories require that , and we find that the few network data sets in these categories are slightly more prevalent near .
iii.2 Alternative Distributions
Independent of whether the powerlaw model is a statistically plausible explanation of a particular degree sequence, a model comparison between the fitted power law and alternative distributions can be instructive. In particular, the likelihood ratio test can reveal whether some other distribution is an equally good or even a better fit to the data. And, an outcome in which a power law is both a statistically plausible model of the data and a better model than alternatives provides strong evidence of scalefree structure.
Across our corpus of network data sets, likelihood ratio tests find only modest support for the powerlaw distribution over alternatives (Table 2). For instance, the exponential distribution, which exhibits a thin tail and relatively low variance, is favored over the power law (36%) nearly as often as the power law is favored over the exponential (37%). This outcome accords with the broad distribution of scaling parameters, as when (Fig. 4, 32% of data sets), the degree distribution must have a relatively thin tail.
The lognormal is a broad distribution that can exhibit very heavy tails, but which is nevertheless not scale free. Across data sets, we find that the lognormal is favored more than three times as often (45%) over the power law, than vice versa (12%), and the tests are inconclusive in the remainder of cases (43%). Thus, in 88% of cases, the lognormal was at least as goodfitting as the power law for degree distributions. This behavior reflects the wellknown fact that lognormals and power laws are generally very difficult to distinguish with finitesized samples Clauset et al. (2009).
The Weibull or stretched exponential distribution can produce both thin or heavy tails, and is a generalization of the exponential distribution. Like the lognormal, we find that the Weibull is more often a better statistical model of degree distributions (42%) than is the power law (33%). Finally, the powerlaw distribution with an exponential cutoff requires special consideration, as it contains the pure powerlaw model as a special case. As a result, the likelihood of the power law can never exceed that of the cutoff model, and the interesting outcome is the degree to which the test is inconclusive between the two. We find that just over half of the data sets favor the cutoff model over the pure powerlaw model, which suggests that, to the degree that scalefree networks are universal, finitesize effects in the extreme upper tail are quite common.


Test Outcome  
Inconclusive  
Exponential  37%  27%  36%  
Lognormal  12%  43%  45%  
Weibull  33%  25%  42%  


—  49%  51% 
iii.3 Assessing the Scalefree Hypothesis
Given the results of fitting and testing the powerlaw distribution across network data sets, and the results of comparing that model to alternative models, we can now classify each according to the degree of evidence it exhibits for scalefree structure. Accordingly, each network data set is assigned to one of the five categories described in Section II.2, or classified as Not Scale Free.
Across our corpus, fully 43% of network data sets fall into this latter category, admitting no evidence at all for scalefree structure (Fig. 6). Slightly more than half (52%) fall into the SuperWeak category, in which a scalefree pattern among the degrees is not itself statistically plausible, but remains marginally more plausible than alternative distributions. The Weakest and Weak categories represent network data sets in which the powerlaw distribution itself is a statistically plausible model of the networks’ degree distributions. In the Weak case, this powerlaw scaling covers at least 50 nodes, a relatively modest requirement. These two categories account for 33% and 24% of data sets, respectively, indicating that direct statistical evidence of a plausibly scalefree degree distribution is, in fact, relatively uncommon.
Finally, only 11% and 4% of network data sets can be classified into the Strong or Strongest categories, respectively, in which the powerlaw distribution is not only a statistically plausible model of the degree structure, but the exponent falls within the special range and the power law is a better model of the degrees than alternatives. Taken together, these results indicate that genuinely scalefree networks are remarkably rare, and scalefree structure is not a universal pattern.
The balance of evidence for or against scalefree structure in our corpus also varies by network domain (Fig. 7). The degree to which different domains exhibit evidence of scalefree structure should clarify the past and shape future efforts to develop new structural mechanisms. Here, we focus our analysis on network data sets from three domains: biological, social, and technological, which represent 91% of our corpus.
Among biological networks, a large majority lack any direct or indirect evidence of scalefree structure (61% Not Scale Free; Fig. 7a). The aforementioned fungal networks represent a large share of these Not Scale Free networks, but this group also includes some protein interaction networks and some food webs. Among the remaining networks, roughly one third exhibit only indirect evidence (35% SuperWeak), and a modest fraction exhibit the weakest form of direct evidence (22% Weakest). This latter group includes cat and rat brain connectomes. Compared to the corpus as a whole, biological networks are slightly more likely to exhibit the strongest level of direct evidence of scalefree structure (6% Strongest), and these are primarily metabolic networks.
In contrast, social networks present a different picture, in two specific ways. First, a minority of social networks lack any direct or indirect evidence of scalefree structure (19% Not Scale Free; Fig. 7b), and a large majority exhibit indirect evidence (71% SuperWeak). The former group includes the 2006 snapshot of collaborations in network science, and a number of both Facebook online social networks, and board of directors networks. The SuperWeak group includes a large number of both Facebook friendship networks and board of director networks.
And second, among the categories representing direct evidence of scalefree structure, not a single network data set falls into the Strong or Strongest categories. Hence social networks are at best only weakly scale free, with 70% and 55% exhibiting the weakest or weak direct evidence of scalefree structure, respectively. The social networks exhibiting weak evidence include about a third of the Facebook online social networks and half of the board of director networks.
Technological networks exhibit the smallest share of network data sets for which there is no evidence, direct or indirect, of scalefree structure (7% Not Scale Free; Fig. 7c), and the largest share exhibiting indirect evidence (92% SuperWeak). The former group includes some digital circuit networks and water distribution networks. Among the categories representing direct evidence, less than half exhibit the weakest form of direct evidence (43% Weakest). This group includes about half of the CAIDA autonomous systems networks, several peertopeer networks, and a few digital circuit networks. In contrast to biological or social networks, however, technological networks exhibit a modest fraction of networks with strong direct evidence of scalefree structure (28% Strong). Networks in this category include the other half of the CAIDA graphs. But, hardly any of the technological networks exhibit the strongest level of direct evidence (1% Strongest).
Extending the scalefree network hypothesis to cover networks that were not simple allowed us to draw on a much larger range of empirical network data sets. It is therefore possible that the numerous nonsimple network data sets we analyzed have distinct structural patterns from simple networks, and hence are less likely to exhibit a scalefree pattern. We test for this possibility by examining the classifications of the 187 simple networks within the corpus. Among these networks, a minority exhibit neither direct nor indirect evidence of scalefree structure (24% Not Scale Free), and a modest majority exhibit at least indirect evidence (66% SuperWeak; Fig. 8). Compared to the overall corpus, about twice as many simple networks exhibit weak direct evidence, and about the same proportion exhibit strong direct evidence. These differences can be partly explained by the distribution of simple graphs by domain, as 73% of simple graphs in the corpus are social, which exhibits similar proportions across the evidence categories. Hence, the general rarity of scalefree networks holds even when we restrict our analysis to simple graphs, and our inclusion of nonsimple graphs has not skewed our results.
To rule out potential bias of our stronger scalefree definitions against the hypothesis, we also examine the results when we remove the power law with exponential cutoff from our list of alternative distributions. As the power law is a special case of the power law with cutoff, our likelihoodratio test can only be inconclusive or result in favor of the power law with cutoff. In the case where the power law with cutoff is the best model, this case cannot be placed in the Strongest or Strong scalefree categories by definition. We found initially that 10.6% of data sets fall into the Strong category. When we include data sets for which the power law with exponential cutoff was favored over the power law, this increases negligibly to 11.4% of data sets. Additionally, if we also remove the restriction on the range of , the percentage of data sets in this Strong category increases to 23%. This is very close to the results for the Weak category (24%), which indicates that the majority of the decrease from the Weak to the Strong is due to the imposition of the bounds on rather than the requirement against favoring alternative distributions. There is a similarly negligible increase in the number of datasets in the Strongest category, from 3.8% to 4.4% when we allow data sets for which the power law with exponential cutoff is favored. This is all consistent with the fact that the construction of our likelihoodratio tests favors the powerlaw distribution since we use the that maximizes the likelihood of the powerlaw fit.
To summarize: across a large and diverse corpus, we find that it is remarkably rare for a network data set to exhibit the strongest form of direct evidence of scalefree structure, and this fact holds true across different domains. Recall that the Strongest level of evidence requires that the bestfitting powerlaw distribution (i) is itself statistically plausible and have an estimated parameter in the range , and (ii) is at least as good a model as any alternatives. Hence the rarity of networks that meet these criteria implies that in general a scalefree distribution is rarely the best model of a network’s degrees. Or, in other words, we find essentially no empirical evidence to support the special status that the power law has held in network science as a starting point for modeling and analyzing the structure of real networks. Instead, it is an empirical fact that realworld networks exhibit a rich variety of degree structures, relatively few of which are convincingly scale free.
Iv Conclusions
By evaluating nearly 1000 distinct realworld network data sets drawn from a wide range of scientific domains, we find that genuine scalefree networks are rare. In fact, less than 45 network data sets (4%) exhibited the strongest level of direct evidence for scalefree structure, and only 33% of network data sets exhibited the weakest form of direct evidence, in which a power law is at least a statistically plausible model of some portion of the upper tail of the degree distribution. Relaxing the criteria even further, to allow merely indirect evidence of scalefree structure in the degree distribution, admits only 52% of networks data sets, and nearly half (43%) of the data sets could be deemed Not Scale Free, meaning they showed no direct or indirect evidence of scalefree structure. These results clarify the empirical status of the common claim that scalefree networks are universal, and indicate that such claims are generally not empirically grounded.
Across different scientific domains, the evidence for scalefree structure is generally weak, but varies slightly in interesting ways. These differences provide some hints as to where genuine scalefree structure may exist. For instance, we find somewhat stronger evidence that scalefree structure occurs in some biological and technological networks. This pattern of evidence agrees with some theoretical work on domainspecific mechanisms for generating scalefree structure, e.g., in biological networks via the wellestablished duplicationmutation model for molecular networks PastorSatorras et al. (2003); Middendorf et al. (2005); LimaMendez and van Helden (2009) or in certain kinds of technological networks via highly optimized tolerance Carlson and Doyle (1999); Newman et al. (2002).
In contrast, we find that social networks are at best weakly scale free, indicating that if mechanisms like preferential attachment Albert et al. (1999); Price (1965); Simon (1955) or its many variants Leskovec et al. (2007); Berger et al. (2004) do operate in these contexts, they are strongly confounded by other mechanisms that cause statistically significant deviations from the expected powerlaw form. Class imbalance in the corpus prevents us from making broad conclusions about the prevalence of scalefree structure in informational networks or the theoretical relevance of scalefree mechanisms in that domain. However, the few informational networks we analyzed provided little indication that they would exhibit strongly different structural patterns that the domains with better representation in our corpus.
In general, the modest differences in the balance of evidence across the three domains for which we obtained sufficiently large samples—social, biological, and technological networks—seem to support a general conclusion that there is likely no single universal mechanism, or even a handful of mechanisms, that can explain the wide diversity of degree structures found in realworld networks. The failure to find universal patterns in the degree structure of networks indicates that a great deal remains unknown about how networks from different domains differ and what kind of structural patterns are common within them. We look forward to new investigations that detail these differences and commonalities.
The empirical rarity of scalefree networks presents both a puzzle and an opportunity. The strong focus on scalefree patterns over nearly 20 years has meant relatively less is known about alternative mechanisms that produce nonscalefree structural patterns. Hence, an important direction of future work in network theory will be development and validation of novel mechanisms for generating more realistic degree structure in networks. Similarly, theoretical results concerning the behavior of dynamical processes running on top of networks, including spreading processes like epidemiological models or influence models, may need to be reassessed in light of the genuine structural diversity of realworld networks.
The statistical methods and evidential categories developed and used in our investigation provide a quantitatively rigorous manner by which to assess whether some network exhibits scalefree structure or not. Their application to a novel network data set should enable future researchers to determine whether scalefree modeling assumptions are empirically justified.
Furthermore, large corpora of realworld networks, like the one used here, represent a powerful, datadriven resource by which to investigate the structural variability of realworld networks. Such corpora could be used to evaluate the empirical status of many other broad claims in the networks literature, including the tendency of social networks to exhibit high clustering coefficients and positive degree assortativity Newman and Park (2003), the prevalence of the smallworld phenomena Watts and Strogatz (1998), the prevalence of “rich clubs” in networks Colizza et al. (2006), the ubiquity of community Girvan and Newman (2002) or hierarchical structure Clauset et al. (2008), and the existence of “superfamilies” of networks Milo et al. (2004). We look forward to those investigations and the new insights they are sure to bring to our understanding of the structure and function of networks.
Acknowledgements.
The authors thank Eric Kightley and Johan Ugander for helpful conversations, and acknowledge the BioFrontiers Computing Core at the University of Colorado Boulder for providing High Performance Computing resources (NIH 1S10OD012300) supported by BioFrontiers IT. This work was supported in part by Grant No. IIS1452718 (A.C.) from the National Science Foundation. Code for graph simplification functions and powerlaw evaluations, as well as replication data, are available at https://github.com/adbroido/SFAnalysis.Author contributions
A.D.B. and A.C. conceived the research, designed the analyses, and wrote the manuscript. A.D.B. conducted the analyses.
References
 Albert et al. (1999) R. Albert, H. Jeong, and A. L. Barabási, Nature 401, 130 (1999).
 Pržulj (2007) N. Pržulj, Bioinformatics 23, 177 (2007).
 LimaMendez and van Helden (2009) G. LimaMendez and J. van Helden, Molecular BioSystems 5, 1482 (2009).
 Mislove et al. (2007) A. Mislove, M. Marcon, K. P. Gummadi, P. Druschel, and B. Bhattacharjee, in Proc. 7th ACM SIGCOMM Conference on Internet Measurement (IMC) (2007) pp. 29–42.
 Agler et al. (2016) M. T. Agler, J. Ruhe, S. Kroll, C. Morhenn, S. T. Kim, D. Weigel, and E. M. Kemen, PLoS Biology 14, 1 (2016).
 Ichinose and Sayama (2017) G. Ichinose and H. Sayama, Artificial Life 23, 25 (2017).
 Zhang et al. (2015) L. Zhang, M. Small, and K. Judd, Physica A 433, 182 (2015).
 Dorogovtsev and Mendes (2002) S. N. Dorogovtsev and J. F. F. Mendes, Advances in Physics 51, 1079 (2002).
 Barabási and Albert (1999) A. L. Barabási and R. Albert, Science 286, 509 (1999).
 Willinger et al. (2009) W. Willinger, D. Alderson, and J. C. Doyle, Notices of the AMS 56, 586 (2009).
 Albert et al. (2000) R. Albert, H. Jeong, and A.L. Barabási, Nature 406, 378 (2000).
 Price (1965) D. J. d. S. Price, Science 149, 510 (1965).
 Simon (1955) H. A. Simon, Biometrika 42, 425 (1955).
 PastorSatorras et al. (2003) R. PastorSatorras, E. Smith, and R. V. Solé, Journal of Theor. Biol. 222, 199 (2003).
 Berger et al. (2004) N. Berger, C. Borgs, J. T. Chayes, R. M. D’Souza, and R. D. Kleinberg, in Proc. 31st International Colloquium on Automata, Languages and Programming (ICALP) (2004) pp. 208–221.
 Leskovec et al. (2007) J. Leskovec, J. Kleinberg, and C. Faloutsos, ACM Trans. Knowledge Discovery from Data 1, 2 (2007).
 Newman (2002) M. E. J. Newman, Phys. Rev. E 66 (2002), 0205009 .
 Carlson and Doyle (1999) J. M. Carlson and J. Doyle, Phys. Rev. E 60, 1412 (1999).
 Newman (2005) M. E. J. Newman, Contemporary Physics 46, 323 (2005), 0412004 .
 Mitzenmacher (2003) M. Mitzenmacher, Internet Mathematics 1, 226 (2003).
 Tanaka (2005) R. Tanaka, Phys. Rev. Lett. 94, 1 (2005).
 Li et al. (2005) L. Li, D. Alderson, R. Tanaka, J. C. Doyle, and W. Willinger, Internet Math. 2, 431 (2005).
 Stumpf and Porter (2012) M. P. H. Stumpf and M. A. Porter, Science 335, 665 (2012).
 Golosovsky (2017) M. Golosovsky, Phys. Rev. E 032306, 1 (2017).
 Barabasi (2016) A. Barabasi, Network Science, Chapter 0 (Cambridge University Press, 2016).
 Goh et al. (2002) K.I. Goh, E. Oh, H. Jeong, B. Kahng, and D. Kim, Proc. Natl. Acad. Sci. USA 99, 12583 (2002).
 PastorSatorras et al. (2015) R. PastorSatorras, C. Castellano, P. Van Mieghem, and A. Vespignani, Reviews of Modern Physics 87, 925 (2015).
 Gamermann et al. (2017) D. Gamermann, J. Triana, and R. Jaime, “A comprehensive statistical study of metabolic and proteinprotein interaction network properties,” (2017), arXiv:1712.07683.
 Redner (1998) S. Redner, The European Physical Journal B 134, 131 (1998).
 Pachon et al. (2017) A. Pachon, L. Sacerdote, and S. Yang, “Scalefree behavior of networks with the copresence of preferential and uniform attachment rules,” (2017), arXiv:1704.08597.
 Clauset et al. (2009) A. Clauset, C. R. Shalizi, and M. E. J. Newman, SIAM Review 51, 661 (2009).
 Eikmeier and Gleich (2017) N. Eikmeier and D. F. Gleich, in Proc. 23rd ACM SIGKDD Internat. Conference on Knowledge Discovery and Data Mining (KDD) (2017) pp. 817–826.
 Alderson and Li (2007) D. L. Alderson and L. Li, Phys. Rev. E 75, 046102 (2007).
 Mitzenmacher (2004) M. Mitzenmacher, Internet Mathematics 2, 525 (2004).
 Clauset et al. (2016) A. Clauset, E. Tucker, and M. Sainz, (2016), The Colorado Index of Complex Networks, icon.colorado.edu.
 Amaral et al. (2000) L. A. N. Amaral, A. Scala, M. Barthelemy, and H. E. Stanley, Proc. Natl. Acad. Sci. USA 97, 11149 (2000).
 Buzsáki and Mizuseki (2014) G. Buzsáki and K. Mizuseki, Nature Reviews Neuroscience 15, 264 (2014).
 Jeong et al. (2001) H. Jeong, S. P. Mason, A. L. Barabási, and Z. N. Oltvai, Nature 411, 41 (2001).
 Malevergne et al. (2005) Y. Malevergne, V. F. Pisarenko, and D. Sornette, Quantitative Finance 5, 379 (2005).
 DuBois et al. (2012) T. DuBois, S. Eubank, and A. Srinivasans, Electronic Journal of Combinatorics 19, 1 (2012).
 Vuong (1989) Q. H. Vuong, Econometrica 57, 307 (1989).
 Easley and Kleinberg (2010) D. Easley and J. Kleinberg, Networks, Crowds, and Markets: Reasoning about a Highly Connected World (Cambridge University Press, 2010).
 Newman (2010) M. Newman, Networks: An Introduction (Oxford Univerity Press, 2010).
 Lee et al. (2017) S. H. Lee, M. D. Fricker, and M. A. Porter, Journal of Complex Networks 5, 145 (2017).
 Middendorf et al. (2005) M. Middendorf, E. Ziv, and C. H. Wiggins, Proc. Natl. Acad. Sci. USA 102, 3192 (2005).
 Newman et al. (2002) M. E. J. Newman, M. Girvan, and J. D. Farmer, Phys. Rev. Lett. 89, 028301 (2002).
 Newman and Park (2003) M. E. J. Newman and J. Park, Phys. Rev. E 68, 036122 (2003).
 Watts and Strogatz (1998) D. J. Watts and S. H. Strogatz, Nature 393, 440 (1998).
 Colizza et al. (2006) V. Colizza, A. Flammini, M. A. Serrano, and A. Vespignani, Nature Physics 2, 110 (2006).
 Girvan and Newman (2002) M. Girvan and M. E. J. Newman, Proc. Natl. Acad. Sci. USA 99, 7821 (2002).
 Clauset et al. (2008) A. Clauset, C. Moore, and M. E. J. Newman, Nature 453, 98 (2008).
 Milo et al. (2004) R. Milo, S. Itzkovitz, N. Kashtan, R. Levitt, S. ShenOrr, I. Ayzenshtat, M. Sheffer, and U. Alon, Science 303, 1538 (2004).
 Nakagawa and Osaki (1975) T. Nakagawa and S. Osaki, IEEE Trans. Reliability 24, 300 (1975).
Appendix A Simplifying Network Data Sets
Our corpus of realworld networks includes both simple graphs and networks with various combinations of directed, weighted, bipartite, multigraph, temporal, and multiplex properties. For each property, there can be multiple ways to extract a degree sequence, and in some cases, extracting a degree sequences requires making a choice. To resolve these ambiguities, and to provide a consistent set of rules by which to extract a set of degree sequences from a given network data set, we developed a set of graph simplification functions, which are applied in a sequence that depends on the graph properties of the input (Fig. 9). For completeness, we describe these specific pathways, and give counts of how many network data sets in our corpus followed each pathway.
At each stage in our processing we remove one property at a time, making the network simpler and never adding properties. Repeating this process for each property in succession converts a network data set into a set of simple graphs. Some networks are processed into a large number of simple graphs, due to combinatoric blowup of multiple properties. To minimize some of this, we treat weighted graphs differently depending on whether or not they have any multiplex, bipartite, or multigraph properties. Multiplex networks include temporal networks, so many of these have a large number of layers, which can generate many simple graphs, (see below). Thus to mitigate some of the potential blowup, if the graph has any of these properties we ignore the weights. If not, however, the data set is replaced with three unweighted graphs as follows. The goal of this transformation is to replace a potentially dense weighted graph with a set of unweighted graphs that are relatively sparse, but not so sparse as to be trivially disconnected so we identify and then apply three thresholds to the edge weights, so that the resulting unweighted graphs have . These threshold values are determined by the empirical edge weight distribution of the graph, and correspond to choosing the largestweight edges, respectively.
The lower value of or represents a very sparse graph, retaining primarily the largestweight edges, but not so sparse as to be likely strongly disconnected. The upper value represents a relatively dense network, retaining all but the smallestweight edges, but not so dense that the degree distribution is trivial. The middle value splits the difference between these. Our corpus contains only 3 weighted networks, 5 weighted directed networks, for 8 total weighted networks, so these networks represent a modest share of the resulting corpus.
Multiplex and temporal network data sets are composed of “layers,” each of which is a network itself. The multiplex network is replaced by a set of graphs, one for each layer and one for the union of edges and nodes across all layers. In this way, the multiplex or temporal property is removed, and the original data set replaced with a set of graphs. Each graph in this set is then further processed to remove any remaining nonsimple properties. A bipartite graph is replaced with three graphs: one each for the “Amode” projection, “Bmode” projection, and original bipartite graph. If present, multiedges are collapsed and weights discarded. Next, a graph is checked to determine whether it is connected. If it is not connected, it is replaced with two graphs: one which contains only the largest component and one that contains the entire, disconnected graph.
As a final step, directed graphs are replaced by three degree sequences: one for the indegrees, one for the outdegrees, and one for the total degrees; undirected graphs are replaced with the single degree sequence. The results of this sequential processing is a set of degree sequences that, as a group, represent the original network. Our corpus contains 2 pure multiplex networks, 320 multiplex multigraphs, and 130 multiplex directed networks; this yields 451 total multiplex networks.
Network data sets that are bipartite and not multiplex are first replaced with three graphs: one for the “Amode” project, one for the “Bmode” projection, and one for the original bipartite graph. Each of these graphs is then processed starting from just after the bipartite step described above in the multiplex or temporal network processing pathway. In our corpus, there are 16 purely bipartite networks, and 25 bipartite weighted networks; this yields 41 bipartite networks total.
Data sets that are multigraphs, but not multiplex/temporal or bipartite, are merely simplified by collapsing multiedges. Edge weights are then discarded, and the resulting graph is processed starting from the check for directedness as above. In our corpus, there are 137 multigraphs and 2 weighted multigraphs, 1 weighted directed multigraph; this yields 460 multigraphs total, including those that are multiplex.
Data sets that are only directed, with no other properties, are first checked for their connectedness (see above) and are then processed to produce three degree sequences: one each for the indegrees, outdegrees, and total degrees. In our corpus, there are 236 directed networks. In the case of a simple graph, the only check is for connectedness. Our corpus contains 187 simple networks.
Replication data, in the form of the corpus of degree sequences obtained by the above simplification steps, is available (see acknowledgements). The corpus of 927 original network data sets represents approximately 250GB of data, and is hence not easily shareable.
Appendix B Powerlaw analysis
b.1 Fitting the model
If follows a discrete powerlaw distribution starting at , then pdf of the power law has the form
where is the Hurwitz zeta function.
Estimating requires first choosing , which we estimate via the standard KolmogorovSmirnov (KS) minimization approach Clauset et al. (2009). This methods selects the that minimizes the maximum difference in absolute value between the (cumulative) empirical distribution on the observed degrees and the cdf of the best fitting power law on those same observations. This difference, called the KSstatistic, is defined as
We choose as the value that minimizes the . The estimate is chosen by maximum likelihood (the MLE), which we obtain by numerically optimizing the loglikelihood function Clauset et al. (2009).
b.2 Testing goodnessoffit
We assess the goodnessoffit of the fitted model using a standard value, numerically estimated via the standard semiparametric bootstrap approach Clauset et al. (2009). Given a degree sequence with elements, of which are and with MLE , a synthetic data set is generated as follows. For each of synthetic values, with probability we draw a random number from the fitted powerlaw model, with parameters and . Otherwise, we choose a value uniformly at random from the empirical set of degrees . Repeated times this produces a synthetic data set that closely follows the empirical distribution below and follows the fitted powerlaw model at and above .
Applying the previously defined fitting procedure to a large number of these synthetic data sets yields the null distribution of KSstatistics . Let denote the value of the KSstatistic for the best fitting powerlaw model for the empirical degree sequence. The value for this model is defined as the probability of observing, under the null distribution, a KSstatistic at least as extreme as . Hence, is the fraction of synthetic datasets with KS statistic larger than that of the empirical data set. Following standard practice, if , then we reject the power law as a plausible model of the degree sequence, and if , then we fail to reject the model Clauset et al. (2009). We note: failing to reject does not imply that the model is correct, only that it is a plausible data generating process.
Appendix C Alternative Distributions
c.1 Exponential
If follows a discrete exponential distribution starting at , then the pdf of the exponential has the form
As with the powerlaw distribution, we use standard numerical maximization routines to estimate the maximum likelihood choice of .
c.2 Lognormal
The lognormal distribution is typically defined on a continuous variable . To adapt this distribution to discrete values, we bin the continuous distribution and then adjust so that it begins at rather than at .
Let and be the density and distribution functions of a continuous log normal variable, where
and
We define and to be the density and distribution functions of a discrete lognormal variable, given by
and
We then generalize the distribution to start at some minimum value, i.e., rather than starting at 0, the distribution starts at , where is a positive integer. This pmf is obtained by renormalizing the tail of so that it sums to 1 on the interval to , yielding
Maximum likelihood estimation was carried out using standard numerical optimization routines. Additionally, we constrained the optimization in order to prevent numerical instabilities. Specifically, we required and . As a check on these constraints, we verified that in no cases did the likelihood improve significantly by allowing , and the constraint on prevents it from decreasing without bound (a behavior that can produce arbitrarily heavytailed distributions over a finite range in the upper tail). To initialize the numerical search, we set .
c.3 Powerlaw with exponential cutoff
If follows a discrete powerlaw distribution starting at , and with an exponential cutoff in the upper tail, then its pdf has the form
where is the Lerch Phi function.
We again estimate this distribution’s parameters and using standard numerical maximization routines.
c.4 Weibull (Stretched exponential)
A common approach to obtain a discrete version of the stretched exponential or Weibull distribution is to bin the continuous distribution Nakagawa and Osaki (1975). Let and be the density and distribution functions of a continuous Weibull variable, where
Define and to be the density and distribution functions of a discrete Weibull variable, given by:
and
As with the lognormal, we generalize the distribution to start at some minimum value, i.e., rather than starting at 0, the distribution starts at , where is a positive integer. This pmf is obtained by renormalizing the tail of so that it sums to 1 on the interval to , yielding
As with the other distributions, we estimate this distribution’s parameters using standard numerical maximization routines.
Appendix D Likelihoodratio tests
The powerlaw models were compared with the alternatives using a set of likelihoodratio tests. For each alternative distribution, we obtained the loglikelihood of the best fit. The difference between this value and the loglikelihood of the powerlaw fit to the same observations yields the likelihood ratio test (LRT) statistic .
When , the power law is a better fit to the data. Similarly, when , the alternative distribution is the better fitting model. Crucially, when , the test is inconclusive, meaning that the data cannot distinguish between the two models. The test statistic , however, is itself a random variable, and hence is subject to statistical fluctuations. Accounting for these fluctuations dramatically improves the accuracy of the test by reducing both types of incorrect decision rates Clauset et al. (2009). As a result, the sign of alone is not a reliable indicator of which model is a better fit. The now standard approach for controlling for this uncertainty is to calculate a value against the null model of . Only if that model can be rejected, is the sign of meaningful Vuong (1989). In this setting, if , then the absolute value of is sufficiently far from 0 that its sign is interpretable.
We obtain this value with the same method used in Ref. Clauset et al. (2009), originally proved valid in Ref. Vuong (1989). Note that
where is the loglikelihood of a single observed degree value under the powerlaw model, and is the number of empirical observations being used by a model (in our setting, this number is , but we omit that annotation to keep the mathematics more compact).
We have assumed that the degree values are independent, which means the pointwise loglikelihood ratios are independent as well. The central limit theorem states that the sum of independent random variables becomes approximately normally distributed as their number grows large, and that this normal distribution has mean and variance , where is the variance of a single term. This distribution can be used to obtain the value, but requires that we first estimate and . Note that we assume because the null hypothesis is . We then approximate as the sample variance in the observed
where
are sample means.
Under this null distribution, the probability of observing an absolute value of at least as large as the actual test statistic is given by the twotail probability
(2) 
Hence, following standard practice Clauset et al. (2009), if , then we reject the null hypothesis that , and proceed by interpreting the sign of as evidence in favor of one or the other model.
Appendix E Evaluating the method on synthetic data with ground truth
We tested the method on synthetic data sets to see if we recover ground truth structure. We generated 100 random 5000node networks by each of three methods. We expect two of the methods, preferential attachment Easley and Kleinberg (2010) and vertex copying Newman (2010), to generate scalefree networks, and ErdősRényi random graphs not to.
The first method, preferential attachment Easley and Kleinberg (2010), is one of the most commonly referenced scalefree generating mechanisms. The process is as follows. We begin with a 4node directed network, in which each node has 3 out edges, one to each of the other nodes. We then add one node at a time until we reach a total of 5000 nodes in the network. Each added node forms 3 out edges. For each edge, with probability the connection is formed preferentially. That is, the new node connects to an existing node with probability proportional to the indegree of that node. With probability , the connection is uniform, where each existing node has equal probability of receiving the new edge. We expect the indegree sequence of the final graph to follow a powerlaw distribution, while the out degree of every node is 3. In our processing, we extract the in, out, and totaldegree sequences. We’d expect that most of the time the totaldegree sequence will also look powerlaw, since it is the sum of a power law and a constant distribution.
We find 87% of the preferential attachment graphs fall into the SuperWeak category. Further, if we do not consider the power law with cutoff as an alternative model, this increases to 98%. This means 98% of the time, a powerlaw model is favored over alternatives for these graphs. When we consider the plausibility of the powerlaw fit, we see fewer networks. 62% of the preferential attachment graphs fall into the Weakest and Weak categories, 60% in the Strong category, and 0 in the Strongest category. This is not entirely unexpected, however. Each of these graphs is directed and splits into three degree sequences: one each for indegree, outdegree, and total degree. As predicted by the literature, the indegree sequences are usually plausibly powerlaw (80%), while the outdegree sequences never are, and the total degree sequences are again usually but not always powerlaw (74%). This accounts for the variance and lowerpercentage of networks showing direct powerlaw evidence. This also implies that the preferential attachment mechanism does better at generating networks for which a power law is a better fit than alternatives, than at generating genuinely powerlaw networks. The absence of graphs in the Strongest category is due to the fact that this category requires that 90% of associated simple graphs be plausibly powerlaw, and this generating mechanism gives at best 67%.
The vertexcopying method Newman (2010) is also expected to generate scale free structure. We again start with a 4node directed network, each node connected to the other 3. We add nodes one at a time until we reach 5000 total and for each follow the same procedure. Pick an existing node at random. For each outedge that has, copy it to the new node with probability . With probability , attach uniformly to one of the other nodes. Each node has outdegree 3. When we process this graph we expect again a powerlaw indegree distribution and usually a powerlaw totaldegree distribution.
The results are consistent with what we expect. 88% of the vertexcopying graphs are SuperWeakly scalefree, which jumps to 99% if we ignore powerlaw with cutoff. 74% fall into the Weakest and Weak categories, meaning the power law is plausible with at least 50 points in the tail of the degree sequence. 70% fall into the Strong category. Since outdegree sequences are never plausibly powerlaw we have none in the Strongest category.
ErdősRényi random graphs are simple and are generally known to have thinnertailed degree distributions. To generate these graphs, we add nodes and then for every possible edge, there is a probability of connection, where is the desired mean degree. We chose , so . We expect the majority of these to follow thinnertailed distributions than the power law.
The ErdősRényi random networks indeed have very different results. Only 16% are SuperWeak, though this increases to 31% without the power law with cutoff as an alternative. This indicates that these networks have thinnertailed degree distributions. 51% and 50% of these networks fall into the Weakest and Weak categories, respectively, but because the bestfit values are all large (the smallest is 7.55), none fall into the Strong or Strongest categories. This is again consistent with the thintailed nature of these networks. We may have expected to see a smaller fraction of these networks falling into the Weak categories, but this result indicates how inclusive these two categories are.
As our methods were able to fairly well recover the ground truth when tested on these synthetic networks, we feel confident in our results for the realworld networks.