🧬 biology

Entropy partitioning reveals an organization–adaptability window in open flow networks

This paper introduces a Shannon entropy-partitioning framework for open flow networks to reveal an "organization–adaptability window," finding that while most natural ecosystems occupy this optimal state, urban systems typically fail to do so due to insufficient boundary exchange diversity relative to internal flow disorder.

Original authors: gérard merlin, julien Ramousse

Published 2026-07-31

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: gérard merlin, julien Ramousse

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Imagine the world as a giant, bustling city of invisible rivers. These aren't water rivers, but streams of energy and matter flowing through everything from a tiny pond to a massive metropolis. In science, we call these "open flow networks." Think of a forest: sunlight pours in, trees eat it, animals eat the trees, and waste flows out. Or think of a city: electricity and food arrive, people use them, and heat and trash leave. For these systems to survive, they have to walk a tightrope. If they are too chaotic, they fall apart; if they are too rigid, they can't handle changes. They need just the right amount of "organized mess" to stay alive. This balancing act is the heart of a field called non-equilibrium thermodynamics, which studies how things stay alive and structured while constantly burning energy. Scientists have long wondered: Is there a secret "sweet spot" where nature and human cities both find their perfect balance? And if so, are we humans doing a good job at hitting it?

Enter a new study by Gérard Merlin and Julien Ramousse, who decided to look at this balancing act through a fresh pair of glasses. Instead of just counting how much energy flows through a system, they used a mathematical tool called "entropy partitioning." In simple terms, entropy is a measure of disorder or confusion. The authors realized that not all disorder is the same. There is the "internal mess" (how chaotic the flow is inside the system) and the "external chatter" (how diverse the exchanges are with the outside world). They created a new way to measure these two things separately, treating them like ingredients in a recipe. They wanted to see if nature and human cities were mixing these ingredients in the same way to stay healthy.

The researchers tested their idea on 74 different real-world systems, ranging from pristine natural ecosystems to complex urban networks and industrial factories. They were looking for an "organization–adaptability window." You can think of this window as a Goldilocks zone for survival. To be in this window, a system needs to be organized enough to function efficiently (robustness) but open and diverse enough to adapt to changes (adaptability). If a system is too rigid, it breaks when the weather changes. If it's too messy, it collapses under its own weight. The authors calculated a score for each system to see if it landed inside this cozy, safe zone.

The results were striking and revealed a clear divide between the wild and the built world. When the authors looked at natural ecosystems, they found that a whopping 84.4% of them were sitting comfortably inside this organization–adaptability window. Nature, it seems, has figured out the recipe. These systems have a beautiful balance: their internal flows are structured, but they also have a rich, diverse exchange with their environment. However, when they turned their gaze to urban systems (cities), the picture changed dramatically. Only 38.1% of the cities managed to stay in the sweet spot. The rest were pushed out, mostly because they were "over-organized" on the inside but "under-connected" on the outside.

To make sure this wasn't just a fluke or a trick of the math, the authors ran a series of computer simulations. They created "null models"—fake networks that kept the same size and shape as the real ones but shuffled the internal flows randomly. They found that the natural ecosystems were still doing something special that the random models couldn't explain. The cities, however, were often just as chaotic as the random models, or worse. The study suggests that cities are failing the adaptability test not because their internal roads are too messy, but because they aren't diverse enough in how they trade with the outside world. They are like a fortress with a super-complex, efficient interior but only one or two gates for supplies, making them fragile when those gates get blocked.

Ultimately, this paper doesn't just give us a scorecard; it gives us a diagnostic tool. It suggests that for a city to be truly sustainable, it needs to stop focusing solely on internal efficiency and start diversifying its connections to the world around it. Nature has spent millions of years tuning this balance, and while our cities are impressive feats of engineering, they are currently missing a crucial ingredient: the right kind of external diversity to keep them flexible and alive. The "organization–adaptability window" is real, and while nature is living right in the middle of it, our cities are still trying to find the door.

Technical Summary: Entropy Partitioning Reveals an Organization–Adaptability Window in Open Flow Networks

Problem Statement
Complex open systems, ranging from natural ecosystems to urban metabolisms and industrial networks, must balance two opposing thermodynamic imperatives: maintaining internal structural organization to ensure functional coherence, and exchanging sufficient matter and energy with their surroundings to remain far from equilibrium. Traditional Ecological Network Analysis (ENA) metrics, such as ascendancy, often aggregate internal transfers, boundary exchanges, and dissipative losses into composite quantities. This aggregation obscures the specific mechanisms by which a system achieves robustness—whether through genuinely organized internal flows, diverse boundary exchanges, or simply high total throughput. The paper addresses the need for a diagnostic framework that can disentangle these components to identify a reproducible "organization–adaptability window" where systems are simultaneously robust and adaptable, distinguishing mature, sustainable configurations from those that are thermodynamically fragile.

Methodology
The authors introduce a normalized Max-Min entropy trade-off framework applied to 74 empirical open flow networks. The methodology relies on Shannon entropy partitioning rather than direct measurements of thermodynamic entropy production, using flow distributions as mesoscopic information-theoretic descriptors.

Entropy Partitioning: The framework decomposes system entropy into:
- Internal Entropy ( $H_{int}$ ): Measures disorder within the network of internal flows.
- External Entropy ( $H_{ext}$ ): Captures the diversity of exchanges with the environment (imports, exports, and dissipative losses).
- Dissipation Correction: A penalty coefficient ( $\lambda$ ) is applied to account for the functional organization of dissipative losses in far-from-equilibrium systems.
Derived Indices:
- Organization Index ( $\alpha$ ): A normalized metric combining internal flow entropy and effective boundary-balance entropy.
- Adaptability ( $\Psi$ ): Defined as a sigmoidal function of $\alpha$ , peaking at $\alpha = 0.5$ . This represents the system's capacity to reconfigure internal organization in response to changing boundary constraints.
- Robustness ( $R$ ): Defined as the uncertainty in the degree of organization ( $R = -\alpha \ln \alpha$ ), representing the trade-off between external openness and internal order.
Validation and Comparison:
- The study compares the new $\alpha$ metric against Ulanowicz's traditional organization index ( $A$ ).
- Constrained Null Models: To ensure observed patterns are not artifacts of topology or boundary forcing, the authors generated null ensembles preserving boundary conditions, active link topology, and total internal throughput while randomizing internal flow intensities.
- Pareto Front Analysis: Systems are plotted in a plane defined by normalized internal disorder and boundary exchange diversity to visualize the joint viability domain.

Key Results

The Organization–Adaptability Window: The analysis identifies a specific region in the entropy trade-off space where both robustness and adaptability are high ( $\Psi \geq 0.9$ and $R \geq 0.9$ ). This window corresponds to an organization index $\alpha$ near 0.5.
Natural vs. Urban Systems:
- Natural Ecosystems: 84.4% of natural ecosystems (using $\beta=2$ ) occupy this high-adaptability window. They cluster within a coherent diagonal band in the Pareto front, indicating a coordinated increase in boundary exchange diversity and internal flow organization.
- Urban Systems: Only 38.1% of urban systems fall within this window. Urban systems are predominantly displaced toward a region of high internal disorder relative to insufficient boundary exchange diversity. This suggests they are "over-organized" internally but lack the external diversity required to support their complexity, rendering them thermodynamically fragile.
Metric Performance: The new dissipation-corrected entropy trade-off metric ( $\alpha$ ) separates natural and urban systems significantly, whereas Ulanowicz's index ( $A$ ) does not.
Null Model Validation: Natural ecosystems show a strong, non-random departure from constrained null expectations for both organization and adaptability (median z-scores of 36.1 and 8.7, respectively). However, they do not show a comparable signal of proximity to the theoretical Shannon robustness optimum ( $1/e$ ). This indicates that the distinguishing feature of mature ecosystems is not a strict targeting of a theoretical optimum, but a specific, non-random compromise between organization and adaptability.
Pareto Analysis: The Pareto front reveals that urban systems often satisfy robustness criteria while failing adaptability constraints, a nuance lost in scalar indices.

Significance and Claims
The paper claims that entropy partitioning provides an operational diagnostic for assessing the thermodynamic organization of flow systems without reducing their complexity to total throughput alone. The authors posit that the identified "organization–adaptability window" serves as a macroscopic signature of thermodynamic maturity.

Diagnostic Utility: The framework allows for the differentiation of systems that are robust but inflexible (common in urban systems) from those that are both robust and adaptable (common in mature natural ecosystems).
Theoretical Grounding: By separating internal flow disorder from boundary exchange diversity, the method aligns with non-equilibrium thermodynamics principles, specifically the balance between minimizing internal dissipation and maximizing entropy export to sustain far-from-equilibrium states.
Sustainability Assessment: The displacement of urban systems from the viability window suggests that their thermodynamic fragility stems not from internal disorganization, but from a lack of diversified boundary exchanges. This offers a specific target for sustainability interventions: increasing the diversity of external exchanges relative to internal flow disorder.

The authors conclude that this approach offers a comparative tool for evaluating the sustainability and resilience of ecological, urban, and industrial networks, grounded in information-theoretic descriptors that bridge microscopic dynamics and macroscopic system behavior.

Technical Summary: Entropy Partitioning Reveals an Organization–Adaptability Window in Open Flow Networks

More like this