Mining Pools have become commonplace in every major Proof of Work (PoW) blockchain, as they allow miners to have a stable income. This has had the unintended side effect of greatly decreasing the number of entities validating the Ethereum blockchain.
This post analyses the degree of centralisation caused by Ethereum’s pool mining hubs. It also examines the upcoming transition to Proof of Stake (PoS) with Casper — and it's implications for Ethereum centralisation.
Let’s dive right in and take a look at what a mining pool is, and why someone would use one.
What exactly is a Mining Pool?
A mining pool is a pooling of computational resources by miners, who share their processing power over a network. If the pool mines a block, the reward is spread based on the contributions each miner provided. It is common for a pool to have a small fee for the organiser (0–2%). Miners opt for pools because they allow for consistent revenue, while mining alone may entail long periods of time before successfully solving a block.
Below are some of the biggest pools in the Ethereum network:
Is there a downside?
As miners gravitate to the most prominent pools, the number of entities validating blocks is reduced; thousands of miners may provide the hashing power, but ultimately the pool organiser controls what information is submitted in a block.
The diagram below uses squares to represent a block validator— with size relative to computational power.
As you can see — centralisation is a spectrum, which can range from one address controlling the network (fully centralised) to every address validating blocks. Under the PoW protocol (and also most PoS implementations) it’s preferable to have more parties competing for block validation, with no entity holding a large portion of hashing power. The current state of affairs is represented by the right hand image — an increasingly centralised ecosystem.
What’s the worst that could happen?
It’s well known that an organisation controlling the majority of a network’s power would be able to attempt an attack with a significant likelihood of success (the fabled 51% attack).
A successful attacker would be able to:
- Spend their Ether multiple times (double spending)
- Modify transactions within the past few blocks
- Create public mistrust of the blockchain’s authenticity
An attack is unlikely, as a pool’s profit outnumbers the potential gains of a short attack. That said, it is always important to analyse worst-case scenarios in potential cyber-security threats.
Let’s take a closer look at the distribution of miners that successfully obtained a block reward:
Only 14 different addresses mined the last 7 days of Ethereum blocks (39975 blocks)!
In the last 7 days (June 4th-11th 2018), Ethermine and Sparkpool alone accounted for 49.6% of the block rewards — and adding f2pool to the mix would reach a total of 64.1% of the network.
Mining Pools are not economically incentivized to perform such an attack, and it’s unlikely they will ever be. However, the Ethereum network should not have a failing point of 3 entities.
An ideal network is protected both from a technical and economicperspective. Ethereum developers are hard at work addressing changes that would allow for more distributed verification. Let’s further examine the plans currently being developed to improve centralisation — and analyse whether they resolve the problems created through the current PoW protocol.
What is Casper?
Casper is a Proof of Stake (PoS) consensus algorithm that, instead of using computational power to ensure decentralisation of block validation, relies on an economic stake in the network. In short, users lock Ether in the network, and gain voting power relative to their stake; they are in turn rewarded Ether over time.
How will this affect Centralisation?
In my previous blog post, I analysed Ether Balance distribution. One of the findings was that the top 10 addresses hold 11.4% of Ether.
While still a surprising distribution — this is much better than the previous scenario.While previously the top two pools controlled 49.6% of hashing power, the top address ‘only’ has 1.5% of total Ether.
From Mining Pools to Staking Pools
Taking part in Casper’s PoS system will have a large minimum requirement (at least 32 Ether)— preventing the average user from participating directly. Earlier in the article, we discussed how mining pools make it practical for small miners to participate in block validation — in this case, staking pools allow people with low ETH balances to take part in the network.
Won’t this lead to the same centralisation problem?
There is a crucial difference between Mining and Staking pools. In mining pools, participants are only risking the Ether they have not yet withdrawn from the pool. This means that miners don’t have an incentive to be responsible with whom they provide their computational power — in the short run, they make the same profit.
In Casper, participants have to deposit (and thus risk) their Ether: if there is any malicious activity from a pool, this will result in a penalty to all users. This forces users to be conscious of whom they trust. Furthermore, the percentage of Ether lost increases as the pool size increases:
Ether Penalty = 3 x Percentage of Users in the Network.
Thus, if a pool holding 25% of the network gets hacked (which is not unreasonable, seeing that Ethermine currently holds 25.6%), one would lose 75% of their deposit. On the other hand, someone in a network representing 1% of validators would only risk losing 3% of their holdings.
This relationship means that users will prefer to take part in small pools, where they are not be risking more Ether than they stand to gain. This in turn will keep nodes at reasonable levels — heavily disincentivising any group from growing to the degree current mining pools are at.
Given the effects examined above, Casper’s implementation will hopefully take us to a more decentralised validator network.
TokenAnalyst parses and classifies every on-chain transaction (currently from the Ethereum blockchain) with the goal of deriving fundamental insights to value crypto assets.