NEXIS Theory: Interaction Topology and Continuous Sustainability Inference
This paper proposes the NEXIS Theory, which redefines sustainability as a continuously inferred latent state driven by the co-evolution of subsystem conditions and time-varying interaction topology, demonstrating that topology-aware models outperform traditional state-only approaches in predicting system stability and outcomes.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
For decades, the way we measure whether a city, a building, or an ecosystem is "sustainable" has relied on a simple, static checklist. Scientists and policymakers have traditionally tallied up individual numbers: how much carbon a factory emits, how much water a neighborhood uses, or how efficient a specific piece of machinery is. These numbers are added together to create a single score, much like a report card. This method assumes that if you know the condition of every individual part, you know the health of the whole system. It treats the system as a collection of separate items, where the relationships between them are secondary details. However, a growing body of research suggests this view is incomplete. Real-world systems, from forests to power grids, are not just collections of parts; they are networks where the connections between things change constantly. The way components interact, the flow of information between them, and the strength of their bonds are just as critical as the parts themselves. If the connections break or rewire, the entire system can behave differently, even if the individual parts remain unchanged. Understanding these shifting connections is the key to predicting whether a system will thrive or collapse.
A new perspective proposed by researchers Ericson Lau and Daniel Chan at the Hong Kong Polytechnic University challenges the old checklist approach. They introduce a framework called NEXIS, which stands for Networked Exchange and Integration of Systems. Instead of looking at sustainability as a fixed score derived from isolated data points, NEXIS views it as a hidden, evolving state that emerges from the constant dance between five different domains: nature, engineered structures, digital networks, institutions, and human society. The core idea is that sustainability is not something you can measure directly with a single sensor; it is a latent condition that must be inferred by watching how these five domains interact over time. The researchers argue that the structure of these interactions—the topology, or the map of who is connected to whom and how strongly—is a primary driver of system behavior, not just a background detail.
To test this idea, the researchers did not build a new physical machine or run a complex simulation from scratch. Instead, they turned to four massive, publicly available datasets containing real-world data from buildings and energy systems. These datasets, collected from sources like the ASHRAE Great Energy Predictor III and the Building Data Genome Project 2, contain continuous streams of information about energy use, environmental conditions, and operational loads. The team used these records to reconstruct the "interaction topology" of these systems. In plain terms, they mapped out how different variables influenced one another over time, creating a dynamic picture of the system's internal connections. They then compared two ways of predicting how these systems would behave. The first method was the traditional approach, which looked only at the state of the individual components, such as temperature or energy demand. The second method, based on NEXIS, looked at both the components and the changing structure of their connections.
The results of this analysis suggest a significant shift in how we should understand system stability. The researchers found that when two systems had nearly identical conditions in their individual parts, they could still follow completely different paths if their internal connections were structured differently. In some cases, systems with similar energy demands and environmental conditions diverged sharply in their performance. The systems that maintained a dense, well-connected web of interactions between their parts tended to be more stable and recovered more smoothly from stress. In contrast, systems where the connections were sparse or fragmented showed more volatility and struggled to stabilize when conditions changed. This divergence happened even though the raw numbers for the individual parts looked the same, proving that the arrangement of the connections themselves carries vital information that the parts alone do not reveal.
Furthermore, the study observed that changes in the structure of these connections often happened before any visible change in the system's performance. For instance, the way different variables began to influence each other would shift during periods of stress, acting as an early signal that a system was about to transition into a new state. A model that paid attention to these shifting connections could detect these transitions earlier and more accurately than a model that only watched the individual numbers. This suggests that the "wiring" of a system is not static; it reconfigures itself in response to pressure, and this reconfiguration is a critical factor in whether the system remains sustainable or tips into failure.
The authors are careful to note that this work is a conceptual foundation rather than a final, definitive proof. They describe their findings as evidence of theoretical plausibility and empirical consistency, meaning the real-world data aligns with their theory, but more rigorous testing is needed to confirm the predictive power of the model across all types of systems. They acknowledge that their current analysis focused primarily on engineered and environmental data, and that fully integrating the social and institutional domains will require more complex data in the future. However, the consistency of the results across four different datasets provides a strong signal that the old way of thinking—treating sustainability as a simple sum of parts—is missing a crucial piece of the puzzle.
This research proposes a fundamental change in how we assess the health of complex systems. It suggests that to truly understand sustainability, we must stop treating the connections between parts as secondary and start viewing them as a primary, living component of the system itself. By shifting from a static report card to a continuous, structure-sensitive inference, we gain the ability to see the hidden dynamics that drive stability and collapse. The NEXIS framework offers a new lens, one that recognizes that the future of a system is not just determined by what its parts are, but by how they talk to each other. If this approach holds up under further scrutiny, it could transform how we manage everything from individual buildings to entire cities, moving us from a reactive stance of fixing broken parts to a proactive strategy of designing resilient connections.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.