The Importance of Ethereum Staking Diversity

Article by

Jake Greenstein

Date

June 21, 2023

Overview

Ethereum’s change from Proof-of-Work to Proof-of-Stake fundamentally altered Ethereum’s validator network topology and composition. Prior to the Merge, most Ethereum nodes were run by large mining organizations in cheap-energy locations that deployed ever-increasing amounts of computer hardware to produce blocks. Post-Merge, the cost to run a node has come way down, and anyone with 32 ETH can easily run a node through services like Blockdaemon & Quicknode or deploy the relevant software via AWS or Azure. You can also stake less than 32 ETH via liquid staking derivatives like Lido.

While this change lowered barriers for stakers and massively reduced the network's energy consumption (great!), the resulting network has tended towards centralization in several important dimensions. This increased level of centralization across software clients, organizations, and geographies presents a risk to the network’s health. This post will provide a quick overview of these dangers and why Hivemind believes it is important to stake a portion of our Ethereum in a “minority” setup.

Centralization Dangers

Above everything else, the Ethereum network is always supposed to run 24 hours a day, 365 days a year. Blockchain networks can achieve extremely high levels of uptime through decentralization. If parts of the network go down due to power outages, software issues, user errors, etc., other nodes in the network just keep running. Increased centralization across any major dimensions presents an increased risk of network failure and devaluation of ETH. Major dimensions include, but aren’t limited to, the following:

Validator Client

Ethereum nodes all need to run software that communicates with other nodes in the network. This software represents the largest point of centralization for most protocols. Many projects have only one client, and a bug in the client software can threaten the entire network. Issues can also harm investors, as slashing events can cause material loss.

Ethereum nodes require two clients: an Execution client and a Consensus client. The Execution client is discussed more frequently, as roughly half of Ethereum nodes use Geth, with another ~25%, ~13%, and ~12% using Nethermind, Besu, and Erigon respectively. One of the easiest ways to improve the decentralization of Ethereum is to use a minority client, and the trend away from Geth (which was used by over half the nodes in the network as recently as earlier this week) is positive for the network's diversification. There is more parity across Consensus clients, with ~37% of nodes using Lighthouse and another ~37% using Prysm. You can view Ethereum’s validator client diversity here.

The availability of multiple clients positively impacts blockchain diversity, and several recent developments excite Hivemind. Paradigm just launched Reth this week, and it would not surprise us to see this become widely adopted in the future. Jump has been developing an alternate client for Solana (Firedancer), and the team’s ultimate goal is to help increase network throughput and efficiency in addition to bolstering resiliency.

Cloud Provider

Cloud providers such as AWS and GCP help reduce the complexity of deploying nodes by allowing developers to focus on node software and not the underlying hardware management. However, as more deployments are hosted in the cloud vs. on bare metal servers, these providers represent another centralization risk. Outages such as AWS’s recent US-East-1 outage can take down many nodes simultaneously, which can threaten network stability. Furthermore, cloud providers can unilaterally decide to deplatform crypto companies. Hetzner’s decision to ban crypto nodes contributed to ~22% of all Solana stake going offline, causing temporary concern in the community.

Controlling Organization

Many stakers choose to stake via a node provider like Blockdaemon or a liquid staking project like Lido. These organizations present another risk to networks as they control significant stake and large number of nodes. Recently, Lido has come under scrutiny as nearly a third of all staked ETH is under the project’s control.

Decentralized validator technology can help alleviate some of these concerns, but as long as stakers leverage intermediaries to help bring stake online (whether it’s a technology operator, liquid staking protocol, or custodian), these operators will control a disproportionate amount of assets staked.

Geography

The geographic distribution of a network also presents a potential centralization vector. For example, ~46% of all Ethereum nodes are operated in the United States, likely due to heavy usage of cloud provider’s US regions. Another 12% of Ethereum nodes operate in Germany, where AWS locates their Europe region.

In the event of a large-scale disaster such as a hurricane, geographic concentration presents risk. Yes, it’s hard to imagine Germany or the Eastern Seaboard going offline, but these are the trade-offs blockchain network developers need to consider when trying to scale networks.

Hivemind’s Stake

Like many funds, Hivemind holds and stakes Ethereum. Financially, this helps generate yield on our holdings and keeps our capital productive. But – and perhaps more importantly – Hivemind’s staking helps secure the Ethereum network and protect our investments. After all, this is the logic that underpins Proof-of-Stake networks: large holders are incentivized to help ensure network operations.

Hivemind takes many factors into consideration when staking funds, but we must consider network diversity and the above factors, especially in light of the recent discourse around increasing centralization. To help play our part, Hivemind is moving a significant part of our ETH holdings into a new staking set-up involving a minority client, a minority technology partner (stay tuned for more exciting collaboration news here), and a minority geographic region. We encourage other large holders to do the same and help protect the Ethereum network.