Centralization, decentralizing
Last updated
Last updated
A public blockchain aims to be, among others, a decentralized and censorship-resilient structure. This translates into a simple goal: that no one, except the community of that blockchain, has the power to alter or bring down the network.
Decentralization undoubtedly helps achieve this goal. But...
If we look at their active validator data, it seems so. In the following link you can find a dashboard with a lot of practical information. +1M validators and a total of ~25% staked tokens!
However, if we review the components that make its operation possible, we see that a few (in some cases, a single actor) can cause very serious problems to the network. We separate it into layers:
The number of consensus and execution clients is very low. In fact, there are only four for each layer that are used at scale by node operators. In the following link you have the data for both clients... You will see that it is worrying!
Geth is the most used execution client by validators. In total, <75% of validators use this client. Saving the defense mechanisms that each validator may have, a crash of the Geth client would cause a highly serious problem in the network since up to ~75% of it would stop working.
Of course, node operators will have their firewall but... Everyone? And if it is all of them, in how long? With the last crash of Nethermind in January 2024, the second most used client with ~10% of validators under its belt when it had a critical failure, left 8% of the total network without signing attestations or creating blocks during a hour approximately. That is to say, few professional operators have automatic redundancy available for such a case. What if it had happened with Geth?
As for consensus clients, Prysm has around ~40% usage in validators. Therefore it also poses a risk to the Ethereum network.
At Ethernodes, thanks to the use of DVT and a private cluster that uses ALL execution and consensus clients, we mitigate this risk and ensure the continuity of our validators regardless of what happens with each client.
Another single point of failure is the existence of a widely used protocol that has a critical failure... such as Lido Finance. With a market share of total staking at <30%, a failure in their protocol would cause serious damage to the entire network. This is why it is always recommended to use the greatest possible diversity of suppliers.
At EthernodesV2, we spread staking across multiple protocols, in addition to using DVT, to ensure that no protocol is strong enough to impact the network.
