Published in Applied Network Science, the study leverages a validated dataset of 336 topology snapshots spanning from January 2019 to July 2023 to examine how the network’s structure and operational stability have evolved over nearly five years. The authors, Danila Valko and Jorge Marx Gómez, aimed to determine whether the network’s growing size and complexity have compromised its ability to route payments efficiently, or if the underlying infrastructure remains robust despite visible topological changes.
The study addresses a significant gap in previous research, which often relied on limited temporal scopes or simplified graph representations. By utilizing a multi-stage pipeline to reconstruct historical topology states from over 35 million publicly available gossip messages, the researchers created a high-fidelity benchmark dataset. This approach allowed them to compute 47 distinct network-science metrics under various directed, undirected, unweighted, capacity-weighted, and routing-aware graph representations. The dataset was cross-validated against independent reference sources, including BitcoinVisuals and mempool.space, ensuring that the reconstructed topologies accurately reflect the publicly visible Lightning Network.
The central finding of the research is that the Lightning Network is undergoing gradual structural sparsification and increasing modularization, yet it maintains remarkable operational persistence. While metrics such as network density, clustering coefficients, and global efficiency have declined over time, indicating that the network is becoming less cohesive at the mesoscopic level, the distributional characteristics of the network remain stable. This suggests that the network is not disintegrating but rather reconfiguring into a structure that is more fragmented yet functionally consistent for payment routing.
One of the most striking observations is the persistent centralization of routing influence. The analysis reveals that the Gini index for betweenness centrality remains consistently above 0.95, indicating extreme concentration of routing paths through a small subset of highly connected nodes. This high level of centralization has been a subject of concern in previous studies, as it implies that a small number of hubs dominate the flow of payments. However, the authors note that this centralization does not necessarily translate into immediate systemic fragility, as these hub nodes tend to be long-lived and highly available, providing a stable backbone for the network’s operations.
Despite the structural changes, the technical retention of nodes and channels remains high. The average node retention rate between successive snapshots stays above 90%, meaning that the vast majority of nodes persist in the network over time. Similarly, the shared node capacity rate, which measures whether overlapping nodes maintain their relative share of total network capacity, remains close to 1.0. This indicates that the economic influence of the core nodes is preserved, even as the peripheral structure of the network evolves. The stability of these distributional metrics is further supported by Kolmogorov–Smirnov tests and Wasserstein distance calculations, which show minimal deviation between successive snapshots.
The operational stability of the network is further evidenced by the retention of routing-equivalent payment paths. When evaluated using realistic pathfinding strategies implemented in major Lightning Network software clients such as LND, c-Lightning, and Eclair, the channel retention rate decreases to approximately 70% compared to the unweighted representation. However, this reduction is not indicative of functional instability; rather, it reflects the dynamic nature of channel capacities and fees. The observed routing-equivalent paths remain largely preserved, suggesting that the network can adapt to topological changes without significantly impacting the ability to route payments.
The study also highlights the increasing number of bridges in the network, which are edges whose removal would disconnect parts of the graph. The rise in bridges after 2020 suggests that the network may be becoming more technically vulnerable to targeted node or channel failures. Nevertheless, the authors caution that an increase in bridges does not imply immediate routing failure, as many of these critical links correspond to high-capacity or long-lived channels that remain operational for extended periods. The growing number of detected communities also points to increasing modularization, although the authors note that community-detection algorithms can exhibit scale-dependent behavior, making it difficult to distinguish between genuine fragmentation and artifacts of network growth.
Limitations of the study include its reliance on publicly available gossip messages, which do not capture private channels, actual channel balances, or hidden liquidity constraints. Consequently, the operational stability measures reported are indicators of observable topology-based routing feasibility rather than direct evidence of end-to-end payment stability. Additionally, the snapshot density varies across the observation period due to historical archival gaps, which may introduce higher uncertainty in periods with sparse data. The authors acknowledge that the increase in detected communities may partially reflect algorithmic resolution limits inherent in modularity-based methods, and that the present study does not simulate adversarial attacks or liquidity shocks.
Despite these limitations, the study provides a valuable benchmark for future research on payment channel networks. The comprehensive characterization of the dataset and the proposed stability analysis framework offer a reproducible foundation for assessing the robustness and evolution of the Lightning Network. The findings suggest that while the network is structurally reconfiguring, it remains operationally persistent, with fragmentation at the mesoscopic level not currently translating into functional instability. This distinction between structural evolution and operational stability is crucial for developers and researchers aiming to optimize routing algorithms and enhance network resilience.
The implications of this research extend beyond the Lightning Network, offering insights into the design and evolution of other decentralized payment systems. The persistent hub dominance and centralization of routing influence highlight the need for strategies that promote economic decentralization and reduce correlated systemic risk. As the Lightning Network continues to grow, monitoring long-term fragmentation and ensuring that the network can withstand stress scenarios will be essential for maintaining its role as a scalable and efficient layer for decentralized digital payments. The study underscores the importance of rigorous, data-driven analysis in understanding the complex dynamics of blockchain-based infrastructure.
The study’s methodology relies on a multi-stage pipeline that transforms raw gossip messages into structured, geolocated network graphs. The researchers utilized the TimeMachine algorithm to replay, deduplicate, and chronologically order over 35 million messages, reconstructing coherent historical topology states at regular two-week intervals. This process generated 336 internally consistent snapshots, with overlapping or inconsistent reconstructions resolved by retaining the most connected graph per interval. To ensure analytical validity, the dataset was enriched with geographical metadata and cross-validated against independent reference sources, including BitcoinVisuals and mempool.space. Statistical tests, including the Kolmogorov–Smirnov test and Spearman’s correlation, indicated strong consistency between the reconstructed data and validation measurements, with a root mean square deviation of 0.57% for spatial coverage quality.
To assess the network’s structural properties, the authors computed 47 distinct network-science metrics, addressing structure, connectivity, resilience, and emergent patterns. The analysis distinguished between unweighted graph representations, which reveal technical backbone topology, and capacity-weighted representations, which reflect operational reality. Additionally, the study incorporated routing-aware weights derived from the cost functions of three major Lightning Network software clients: LND, c-Lightning, and Eclair. These cost functions account for variables such as payment amount, locktime, routing fees, and channel age. By applying these specific weighting schemes, the researchers could evaluate how realistic pathfinding strategies influence the perceived stability and efficiency of the network’s routing capabilities.
The results indicate that the Lightning Network exhibits persistent heavy-tailed degree distributions, consistent with hub dominance. The average estimated power-law exponent across snapshots was 2.077, with a bootstrapped coefficient of determination of 0.997. A strong positive correlation of 0.82 between node degree and shared node capacity suggests that structural prominence translates directly into economic influence. High-degree nodes not only maintain numerous connections but also control a disproportionately large share of routing liquidity. This coupling between topological centrality and economic capacity highlights the critical role of hub nodes in maintaining the network’s functional integrity, even as the overall density of the network declines.
Temporal stability was further quantified using novel metrics designed to capture structural persistence and transactional stability. The node retention rate, defined as the fraction of overlapping nodes between successive snapshots, remained above 0.9 throughout nearly the entire observation period. The shared node capacity rate, which evaluates whether overlapping nodes preserve their relative share of total network capacity, stayed particularly close to 1.0. These findings suggest that the vast majority of nodes persist over time and maintain a consistent impact on overall network capacity. The stability of these metrics is robust to the timing of measurements, although retention rates show a negative correlation with the time elapsed between snapshots, reflecting the natural churn of the network.
The study also examined the network’s resilience by analyzing the number of bridges, or cut-edges, whose removal would increase the number of connected components. The number of bridges rose noticeably after 2020, suggesting a potential increase in technical vulnerability to targeted failures. However, the authors note that this increase does not imply immediate routing failure, as many bridges correspond to high-capacity or long-lived channels. The size of the minimal edge cover also increased over time, indicating growing connectivity demands. This parallel increase suggests that new connectivity is expanding coverage with structurally critical links rather than reinforcing redundancy, a pattern that aligns with the observed decline in local clustering coefficients.
Limitations of the analysis include the reliance on publicly available gossip messages, which do not capture private channels or actual channel balances. Consequently, the operational stability measures represent indicators of observable topology-based routing feasibility rather than direct evidence of end-to-end payment stability. The snapshot density varies across the observation period due to historical archival gaps, with observations denser during 2019–2020 than during 2022–2023. To mitigate the effects of uneven sampling, temporal summaries were reported as quarterly averages with bootstrapped 95% confidence intervals. The authors emphasize that while the principal temporal trends remain consistent after aggregation, the magnitude of late-period estimates should be interpreted with appropriate caution due to sparser data coverage.
Subject of Research: Topological and temporal stability analysis of the lightning network
Article Title: Topological and temporal stability analysis of the lightning network
Article References: Valko, D., & Marx Gómez, J. (2026). Topological and temporal stability analysis of the lightning network. Applied Network Science, 11(1), Article 49. https://doi.org/10.1007/s41109-026-00820-4
Image Credits: AI Generated
DOI: 10.1007/s41109-026-00820-4
Keywords: Topological, temporal, stability, analysis, lightning, network, scientific research
News Source: Caitlin Barrett. (November 8, 2026). Lightning Network Shows Structural Shifts But Stable Routing. Scienmag.
