PoS blockchains losing their independence: how bad is that?
PoS blockchains losing their independence: how bad is that?
novembre 07 2022

Weekly stories are first featured in our Newsletter. Subscribe here to receive it directly in your mailbox every Monday

The main quality of a blockchain is its independence: its decentralized structure allows processing data without any central authority, ensuring uninterrupted operations and impartial approach to all the users.

At least, that’s the idea.

In practice, however, we start to notice weak spots that could compromise it: either in the blockchain’s design, or in the technical specificities that introduce gatekeepers able to imperil it.

Centralization in staking

Proof-of-Stake is rooted in an old financial principle of using money to make money: the bigger the stake, the bigger the chances to validate a block and get reward, which increases the stake… This process creates an oligarchy – a small group of entities having control over the blockchain.

In case of Ethereum, this is clearly illustrated by its current stakers: out of 14.8 million ETH in the deposit contract, 32% have been staked by only 5 centralized companies, and 35% – by 3 decentralized ones. As Santiment noted back in September, over 46% of all Ethereum blocks were added by just two entities – Lido and Coinbase.

Decentralized services like Lido do not pose an obvious threat (unless their code gets seriously compromised of course) as long as they keep the decision-taking decentralized. However, centralized stakers like Coinbase and Kraken not only are positioned to gain more influence over time, but they are also likely to censor transactions they process – for they are big companies incorporated in the US and obliged to comply with the US regulations, which now include banning addresses having used privacy tools like the Tornado Cash mixer.

What’s interesting is that Ethereum does seem to please the governments: Exaion, a subsidiary of the French state energy company EDF, is known for running 10 Ethereum nodes for itself and 140 ones – for its clients.

Centralization in block producing

Another danger to Ethereum’s censorship-resistance is less visible, but rapidly growing: it is block-building outsourcing software called MEV-boost. Validators are increasingly using MEV-boost relays to optimize their revenue, and it happens that there are only 7 of them.

In October, development company Labrys noted that 51% of Ethereum blocks were OFAC-compliant, because they were added by MEV-boost relays regulated under OFAC. This number has since grown to an alarming 73% (you can use this website to track this trend).

This means that transactions from/to wallets that are somehow related to Tornado Cash are much less likely to be picked up from the mempool, put into blocks, and processed.

Centralization in nodes hosting

Blockchains are stored and run on nodes, which can be hosted either on home computers or private servers, or on cloud services. The latter brings an element of uncertainty, as always with a trusted third party: it can suffer outages, it can decide to deny its services, it can censor… etc

In Ethereum’s case, almost 66% nodes of all the network are hosted on the cloud, and this alone is already weakening the blockchain’s independence. But there’s more: out of these nodes on cloud, over 60% is hosted by Amazon Web Services (AWS).

AWS outages already impacted their nodes operations in 2020 and 2021, and the servers hosting Ethereum nodes being physically located in the US, the risk of censorship is not null.

This problem is not limited to Ethereum, of course.

Solana nodes’ are big cloud users too, the biggest being Equinix (31% of total network stake), Hetzner (20%), AWS (14.7%) and OVH (8.7%). Last week we got another demonstration of why such trust could be misplaced.

Last Wednesday Hetzner’s cryptophobia reached a critical level and it decided to block all Solana activities and take 1000 of its validators offline. This move did not disturb the network (it would have taken 39% of the total stakers), but showed its weak link.

🎶 Fight the power 🎶… or not?

So how does one solve these issues?

Nodes hosting: it is possible for blockchain nodes to chose independent cloud service providers (Solana has even launched a Server Program initiative helping validators find and use different data centers) and decentralized cloud (Filecoin is one such example).

MEV-boost problem can be solved by opening more relays in non-OFAC-compliant jurisdictions and/or by decentralizing them. Also, one of the Ethereum’s planned upgrades (the one that Vitalik called “scourge” and revealed this Saturday) does intend to tackle this issue.

The most difficult problem, however, is the very principle of Proof-of-Stake. Whatever mechanism a blockchain might use to randomize block creators to try and protect itself from censorship (like Algorand does), it cannot change the accumulation of capital (and therefore power) by the biggest stakers.

The more ETH you stake, the more ETH you gain, the more you can influence the blockchain’s operations… all while applying your jurisdiction rules in the process.

How bad is that ? 🤔

Being OFAC-compliant will not become the end of the world for a popular blockchain, especially as the adoption grows. It’s unlikely that newcomers willing to play a crypto game will be reflecting on the principles of non-discrimination or other crypto ideals, from which Ethereum is slowly drifting away.

What’s more, economic interests of its oligarchy will motivate it to take the best possible decisions for Ethereum: they are interested in that their stakes grow, as well as the crypto industry, in which they operate.

However, there still will be people who hold in high regard crypto’s decentralized ideals and who will be trying to make them

What about Bitcoin?

Bitcoin is intrinsically more independent.

Proof-of-Work consensus is fairer that PoS: when a miner mines a block and gets their reward, it does not become easier to mine the next block, they still have to spend as much (or more) electricity as their competitors.

Bitcoin miners are less dependent on their country of residence (Chinese ban showed how fast they can pack their rigs and move), and they are economically incentivized to not reach the critical influence threshold (miner collectives that had reached it in the past had dissolved in order to not spook the users).

Of course, Bitcoin was not created as a Dapps-building platform: it is not suited for developing complex smart contracts… but some of its layer-2 solutions are, like Stacks for example.

This solution might please the Bitcoin Maxi movement, which is growing fast.

The possibility of someone coming up with a new type of consensus or blockchain structure altogether cannot be dismissed neither.

This will be a multi-blockchain world.