A peer-reviewed study from Cornell University’s Initiative for Cryptocurrencies and Contracts (IC3) argues that Bitcoin and Ethereum are not as decentralized in practice as many in the industry have long assumed. Led by Emin Gün Sirer, the research examined network behavior over 2015 to 2017 using data gathered through the Falcon Relay Network (FRN), focusing on node distribution, interconnection, protocol requirements, and resilience against attacks.
Measuring decentralization in practice
Rather than relying on ideology or marketing narratives, the paper examined how decentralization actually works on the ground. FRN was designed to improve block propagation between miners and full nodes, reduce orphaned blocks, and make the network more efficient. According to the researchers, this architecture could help Bitcoin scale by reducing the tight link between block propagation speed and network size, potentially improving fairness among miners.
The study found that the bandwidth available to Bitcoin nodes had increased by 1.7 times since 2016. Based on that improvement, the authors argued that Bitcoin could support a larger block size without necessarily harming decentralization. In theory, a higher data-clearing rate per second could nearly double Bitcoin transaction throughput, a key issue in debates over fees and network capacity.
Ethereum appears more geographically distributed
One of the paper’s main conclusions is that Bitcoin nodes are more clustered than Ethereum nodes, both geographically and in terms of network latency. Ethereum, by comparison, has more nodes spread across more regions, suggesting that its full-node layer is more decentralized than Bitcoin’s.
The researchers also highlighted infrastructure concentration. More than half of Bitcoin nodes were found to be hosted in data centers, while Ethereum’s share was only slightly above one quarter. That matters because data centers are often controlled by corporations, creating possible centralization risks. The paper also warned that raw node counts can be misleading, especially in the presence of Sybil attacks that can artificially inflate perceived participation in peer-to-peer systems.
Mining concentration remains a core concern
The study found an even sharper centralization problem in mining. In Bitcoin, the top four miners controlled more than 50% of hash rate, while in Ethereum the top three miners crossed the same threshold. More broadly, the paper said that the entire blockchain history of both systems was effectively determined by fewer than 20 mining entities.
For smaller miners, Bitcoin also appeared to offer less predictable rewards than Ethereum. The authors linked that to Bitcoin’s slower block production, which creates more month-to-month statistical uncertainty for small participants. Ethereum’s faster block rate provides more opportunities for outcomes to converge toward expected averages.
Overall, the research does not dismiss the importance of Bitcoin or Ethereum, but it challenges simplified claims about decentralization. Its broader message is that decentralization should be evaluated across several dimensions, including node geography, infrastructure dependence, mining concentration, and network performance, rather than treated as a binary label.