Abstract
Reaching consensus—a macroscopic state where the system constituents display the same microscopic state—is a necessity in multiple complex socio-technical and techno-economic systems: their correct functioning ultimately depends on it. In many distributed systems—of which blockchain-based applications are a paradigmatic example—the process of consensus formation is crucial not only for the emergence of a leading majority but for the very functioning of the system. We build a minimalistic network model of consensus formation on blockchain systems for quantifying how central nodes—with respect to their average distance to others—can leverage on their position to obtain competitive advantage in the consensus process. We show that in a wide range of network topologies, the probability of forming a majority can significantly increase depending on the centrality of nodes that initiate the spreading. Further, we study the role that network topology plays on the consensus process: we show that central nodes in scale-free networks can win consensus in the network even if they broadcast states significantly later than peripheral ones.
Introduction
In decentralized systems like Bitcoin, consensus is often viewed through the lens of computational competition, or "hash power." However, the physical topology of the underlying peer-to-peer network plays a critical, often overlooked role. When a miner discovers a new block, the speed at which that information propagates determines whether it becomes part of the main chain or is discarded as a "stale" block. This paper investigates how network heterogeneity—specifically the presence of central hubs versus peripheral nodes—influences the consensus race.
Modeling Consensus in Heterogeneous Networks
The study utilizes a discrete-time network model to simulate the interplay between block creation (mining) and block propagation. Mining is modeled as a stochastic process following an exponential distribution, while propagation occurs along the shortest paths of the network graph.
To understand the impact of structure, the researchers analyzed three distinct network topologies: Erdős-Rényi (ER) random networks, Stochastic Block Models (SBM) representing community structures, and Barabási-Albert (BA) scale-free networks.
As shown in Table 2, while the diameter of these networks remains relatively constant at 1000 nodes, the distribution of closeness centrality varies. This variation is the primary driver of the "centrality advantage," where nodes with shorter average distances to others can dominate the propagation process.
Empirical Calibration and Validation
To ensure the model's relevance to real-world systems, the authors calibrated their simulations using historical data from major blockchains. By incorporating parameters such as block creation times and median propagation delays, the model could accurately predict fork rates.
The model's validity is underscored by its prediction of a 1.45% fork rate for Bitcoin, which closely matches the historically observed value of approximately 1.69% (represented as $C(\Delta)$ in Table 1). This alignment suggests that the minimalistic model captures the essential dynamics of real-world blockchain propagation.
The Centrality Advantage
The core finding of the research is that closeness centrality acts as a temporal head start. In a competitive fork situation, the node that reaches the majority of the network first typically wins.
- The "Time Machine" Effect: A node with high centrality can effectively "erase" the lead of a peripheral node. Even if a peripheral node mines a block earlier, a central node broadcasting slightly later can still win the race because its block reaches the network majority faster.
- Scale-Free Resilience: In Barabási-Albert networks, this effect is most pronounced. Highly connected hubs can win consensus even when they are several time steps behind in the broadcast race.
- Community Persistence: In SBM networks, the presence of clusters creates "stickiness." Blocks propagate rapidly within a community, making it difficult for external competing blocks to displace them, which can lead to more persistent forks.
Conclusion
The physical structure of a peer-to-peer network is a silent kingmaker in blockchain consensus. This research demonstrates that network position is a form of power, allowing central nodes to exert influence disproportionate to their computational resources. These findings have significant implications for the design of future decentralized protocols, suggesting that security and fairness are inextricably linked to network topology.
Summary updated