Complexity Thought: 2023 Influential Papers Collection
Unraveling complexity: building knowledge, one paper at a time
If you find Complexity Thoughts, click on the Like button, leave a comment, repost on Substack or share this post. It is the only feedback I can have for this free service.
The frequency and quality of this newsletter relies on social interactions. Thank you!
Dall-e 3 representation of this issue’s content
I didn’t expect to receive so many recommendations when I have submitted my invitation to suggest 2023 papers. Let me thank all the readers who have spent some time to post one or more suggestions.
In the first part you will find my overview of the suggestions, where I have filtered out the recommendations related to my own papers (although I am very thankful and honored for those recommendations).
If the same reader suggested more than one paper, I have applied some selection criteria, to favor diversity of topics and potential impact.
In the second part, I will list some of the works I have featured in 2023, limiting myself to at most 10 works (ordering is irrelevant).
From the community
Toward a cohesive understanding of ecological complexity
By Erica A. Newman, UC Berkeley
Ecologists use complexity as a descriptor of their field. This paper examines what aspects of complexity are most commonly invoked and studied in ecology, through bibliometric and text mining analyses. The paper finds that current research trends in ecological complexity are organized around basic theory, scaling, and macroecology, and suggests integration with the principles of complex systems science as way towards more progress in this field.
By Oliver López-Corona. IxM, IIMAS-UNAM
One of the main features of complex systems is its capacity to respond to perturbations by balancing self-organization and emergence. The paper relates this standard properties of complex systems, applied to Earth system, to antifragility using basic principles of out of equilibrium thermodynamic.
By Tim Waring, Associate Professor, University of Maine
Complexity science has yet to truly embrace the modern evolutionary science of human behavior and culture, but remains hooked on evolutionary analogies for social, cultural and institutional change instead. This paper uses the science of cultural evolution to address a major applied problem for the world today.
By Sabin Roman, University of Cambridge
The paper highlights several key topics of importance in complexity science: how chaos emerges and how network structure can help make a system more resilient and sustainable. The model is a set of differential equations which also allows for analytical insight into its dynamics (always a plus). Also, the area of application is of great importance: how societies can collapse and what can be done to save them. The conditions under which chaos emerges in the network is also novel and is again a network effect (not just diffusion between chaotic nodes). All in all, the paper presents key aspects of complexity science (chaos, networks, dynamics) in a topic of high priority (societal collapse, resilience, sustainability).
Complex systems of secrecy: the offshore networks of oligarchs
By Simon Dobson, University of St Andrews UK
Firstly, this is an application of complexity science to what are essentially the digital humanities (or at least to economics), which receives a lot less interest than (for example) connectomics, despite the fact that it's of immediate and widespread interest and impact in a time when tax evasion and illicit funding are of major global concern. Secondly, it derives some fairly complicated network information from a real dataset, and then uses this to suggest possible interventions: essentially developing algorithmic attacks against a network to disrupt it. It therefore provides an objective scientific basis for judicial and political action, which can be discussed on the basis of its likely effects, and I like the way the complexity science blends so well with sociological analysis.
Heterogeneity extends criticality
By Carlos Gershenson, SUNY Binghamton
Many models of complex systems are homogeneous for simplicity. Critical-like dynamics, even when desired, are found only near phase transitions. However, heterogeneity (common in nature) can yield "criticality for free". This prompts the exploration of the effects of heterogeneity in models of complex systems. Also in this work, a general analysis on when heterogeneity should be preferred is included. An open question remains: what would be an "optimal" heterogeneity?
By Anonymous
The use of network modelling and analysis takes into account the complexity of real world phenomena and can uncover actionable information about a system. This work does precisely that by identifying genes that were unknown to be relevant for male infertility across three species. Thus, the combination of information from network analysis, taking into account the structure of shortest-paths, with experimental evidence makes this work a hallmark of how complexity science can uncover actionable information about a system.
Opening up Echo Chambers via Optimal Content Recommendation
By Antoine Vendeville, Médialab Sciences Po, Paris, France
Echo chambers are a primary example of an emergent phenomenon in online social networks. They come with many problems for our societies, from the polarisation of opinions to the spread of fake news and conspiracy theories. Thus, it is essential to not only understand the phenomenon, but also propose ways to remediate it. This paper attacks both these angles, by (1) proposing a model of content propagation that reproduces the distribution of opinions observed in real-life data, and (2) introducing an optimisation method to find individual recommendation rates so as to minimise the echo chamber effect. The method is shown to be effective for a wide range of parameters on a real-life Twitter dataset related to the 2017 French presidential elections.
By Oliver López-Corona. IxM, IIMAS-UNAM
This papers sets the foundations of antifragile control theory which not only contributes to improve quantitative aspects of antifragility framework but also will be most in control of complex systems or control in complex environments.
Machine learning renormalization group for statistical physics
By Ruyi Tao
In this paper, a data-free neural network framework is proposed. In the absence of simulated data, only the symmetry of the model is given as the prior information for training, and finally the renormalization equation under the given symmetry can be found and different universal classes can be divided. This revelation for the study of complex systems lies in that it may be possible to classify complex systems first from the perspective of dividing universal classes, which has important guiding significance for finding universal theories applicable to complex systems.
Neural Network Pruning by Gradient Descent
By Zhang Zhang, School of Systems Science, Beijing Normal University
Biological complex systems can bring a lot of guiding knowledge to the design of AI: sparsity, modularity, energy saving, etc. Most of the guidance in previous papers was added to neural networks in an explicit manual design manner, which increases the complexity of the algorithm and limits the performance of the AI system. In this article, the author proposes a simple architecture, using only deep learning to cleverly introduce the prior knowledge of sparsity. Achieve SOTA performance with a simple framework and obtain an interpretable coefficient neural network. This provides insights into how to harness the power of deep learning to design smarter and energy-efficient systems.
From ComplexityThoughts 2023
I have written a few short essay in 2023, to integrate some of the research covered during this year. Below, you will find some of the papers I think have a great message or a huge potential (see the corresponding Issue of CT for details). Here, I’d just like to point to two essays that I have really enjoyed to write:
Are social media undermining democracy? From algorithms to complexity
What we (don't) know about the role of socio-technical systems in shaping human behavior → Read
What's going on with assembly theory? Claims, controversial claims and merits after 60 years of complexity science → Read
Source: CT
Papers
Models of Cell Processes are Far from the Edge of Chaos from CT#22
Our results suggest that—contrary to current theory—cell processes are ordered and far from the edge of chaos.
Hyperedge prediction and the statistical mechanisms of higher-order and lower-order interactions in complex networks from CT#21
Our results thus highlight the need for good reconstructions of low-order interactions in order to have reliable reconstructions of full hypergraphs
A rugged yet easily navigable fitness landscape from CT#20
Our work shows that ruggedness need not be an obstacle to Darwinian evolution but can reduce its predictability. If true in general, the complexity of optimization problems on realistic landscapes may require reappraisal.
Reactivity of complex communities can be more important than stability from CT#19
We show that reactivity can be a better predictor of extinction risk than stability, particularly when communities face frequent perturbations, as is increasingly common.
Unifying pairwise interactions in complex dynamics from CT#17
Our analysis highlights commonalities between disparate mathematical formulations of interactions, providing a unified picture of a rich interdisciplinary literature.
Indirect effects shape species fitness in coevolved mutualistic networks from CT#16
Our study shows how and why indirect effects can govern the adaptive landscape of species-rich mutualistic assemblages.
Replicable brain–phenotype associations require large-scale neuroimaging data from CT#13
Our findings demonstrate that large-scale neuroimaging data are required for replicable brain–phenotype associations, that this can be mitigated by preselection of individuals and that small-scale studies may have reported false positive findings.
Unveiling the transition from niche to dispersal assembly in ecology from CT#12
Our results suggest that tropical intertidal communities have low niche diversity and are typically in a dispersal-assembled regime where immigration is high enough to overfill the niches.
Emergent stability in complex network dynamics from CT#10
in this ensemble, two of the most ubiquitous characteristics of real-world networks—scale and heterogeneity—emerge as natural organizing principles to ensure fixed-point stability in the face of changing environmental conditions.