The diverse and fair structures of growth with scale-free saturations
Alain Govaert, Emma Tegling, André Teixeira · 2025
Read The diverse and fair structures of growth with scale-free saturations by Alain Govaert, Emma Tegling, André Teixeira (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-15424.txt.
Introduction
Alain Govaert and co-authors study growth with scale-free saturations in a 2025 paper on how real growth processes level off. Growth regimes split into types driven by steady positive feedback, and a sublinear type driven by efficiency where growth slows as resources run short.
The team proposes a simple network growth process that mimics the sublinear regime. They characterize how its scale-free saturation zones shape the diversity and fairness of network structure. Scaling laws seen across complex systems emerge from the model. The work links resource limits to fairer, more varied network shapes. Systems scientists get a clean model of growth that stops itself.
Catalog record
| CREATOR | Alain Govaert, Emma Tegling, André Teixeira |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 68 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-15424
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The diverse and fair structures of growth with scale-free saturations Alain Govaert, Emma Tegling, André Teixeira Abstract Real-world growth processes and scalings have been broadly categorized into three growth regimes with distinctly different properties and driving forces. The first two are characterized by a positive and constant feedback between growth and growth rates which in the context of networks lead to scale-free or single-scale networks. The third, sublinear, regime is characteristic of biological scaling processes and those that that are driven by optimization and efficiency. These systems are characterized by a negative feedback in growth rates and as such naturally exhibit saturations – areas where growth ceases from a lack of resources . Motivated by this observation, we propose and analyze a simple network growth process that is analogous to this sublinear regime and characterize how its scale-free saturations impact the diversity and fairness of its structural properties and give rise to scaling relations observed throughout complex systems and science. Preprint: APS/123-QED I Introduction Many natural and man-made growth processes exhibit local saturations, that is, depletion of resources, under sustained growth. Oftentimes these saturations have detrimental effects on the system, as the competition for resources increases. The finite number of local resources effectively impose constraints on local growth and eventually forces the global system to expand or change in ways to sustain further growth. The rate at which new resources are, or have to be, added to a system with size N largely determines the “pace of life” which expresses itself by a power law scaling of the form N^{\beta-1} [ 1 ] . Bettencourt and colleagues identify three universal categories that can be briefly outlined as follows. The linear regime ( \beta=1 ) has a constant growth rate resulting in exponential growth associated to the individual level. The superlinear regime (\beta>1) exhibits increasing growth rates. This is observed in cities, whose continued growth requires a constant increase in the creation of wealth, resources, and information through innovations, signifying the fast pace of life in large cities. In contrast, the sublinear regime ( \beta<1 ) exhibits decreasing growth rates and thus eventually saturates. In cities this is found in the infrastructural economies of scale that are focused on efficiency and optimization rather than innovation. This is also the regime of biological systems whose pace of life decreases in growth, as exemplified by the allometric quarter power law scalings that predict metabolic rates decrease with body size [ 2 , 3 ] . We argue that these three general relations between growth, scaling, and resource availability have natural analogies to how growing networks organize themselves. Preferential attachment, or cumulative advantage, is analogous to superlinear scaling in which there is a positive feedback between growth and growth rates. This famoulsy leads to the rich-get-richer (or Matthew effect) and scale-free networks with power law degree distributions with superlinear exponents that mostly lie between two and three [ 4 ] . When preferential attachment is tempered, broad-scale degree distributions emerge that exhibit power-law behaviors with an exponential cut-off [ 5 , 6 , 7 ] . This exponential decay in “single scale” networks is analogous to the linear scaling regime of which uniform attachment is a prime example. This finally leaves the sublinear regime whose growth is hampered by saturations. In networks, saturations occur if sets of vertices cannot receive new connections. Depending on how fast the saturations occur this remains a local phenomenon or becomes prevalent throughout the network. In any case, local saturations tend to increase the distance between vertices, thus increasing the diameter and limiting the global connectivity of the network. This hurts the efficiency or optimality of information flow and resource exchange over the network, which is are the main driving forces of growth processes in the sublinear regime. Thus, it is natural for a network growth process in this regime to avoid local saturations by distributing its edges over the vertices not by how much they already have, but by how much they are missing . Indeed, this is analogous to the negative feedback between growth and growth rates that naturally arise from the usage of local resources. These growth processes thus represent the network analogy of how infrastructural economies of scale in cities [ 8 , 1 , 9 , 10 , 11 ] and biological systems [ 3 , 2 ] tend to grow. Yet, they received significantly less scientific attention than the other two scaling regimes and therefore relatively little is known about them. Statistical analyses however support the idea that real-world biological networks may grow in a different scaling regime and have different structural properties than, for example, social and technological networks [ 12 ] . Recent research on evolving networks have indeed recognized this, stressing the importance of physicality and the constraints that it imposes on network structure [ 13 ] . One of these constraints are saturations, typical to sublinear growth, that can result in networks to not only evolve but also grow in a constrained manner. The above empirical findings then suggest that scalings may lie hidden in the global structures induced by local sublinear growth in networks. As in scale-free networks, an idealized model can then be used to uncover these scalings from which the basic principles, and mechanisms that underlie the diverse structure of networks can be identified and built upon. Furthermore, by reversing the network analogy we can also learn about the possible effects that local sublinear growth and scalings may have on the global scale of complex systems across different research domains . To investigate this, we focus on two basic but important constraints that naturally induce saturations in growing networks: the minimum vertex degree d , and maximum vertex degree w . Both have important implications for networks and the processes and algorithms over them. For example, degree bounds commonly influence various spectral properties of networks relevant to dynamical systems that evolve over them [ 14 ] , Degree bounds are also crucial in the feasibility of property testing algorithms [ 15 ] and preference identification over networks [ 16 ] . In games on networks, the critical contagion threshold is known to be at least 1/w [ 17 , 18 , Corollary 3] , indicating that a bound on the maximum degree can aid in the effectiveness of mechanism design. Lower bounds on the vertex degree are desirable for the resilience of distributed control [ 19 ] , the speed of log linear learning [ 20 , Proposition 4] , and the structural (target) controllability of networks [ 21 , 22 ] determined by the size of its maximum matching, often studied in random graphs [ 23 ] . The above illustrates the importance of degree-related properties of networks and motivates their usage as a natural starting point to investigate how they may affect structural and scaling relations in sublinearly growing networks. Our contributions in this direction can be summarized as follows: • We propose an intuitive network growth process rooted in empirical sublinear scalings that incorporates a minimum of parameters, specific details, and is complementary to existing models. • Our analysis shows that local power law scalings in the saturation and volume of sets of