The ‘ETH Merge’ may be the most anticipated event in 2022 within the digital assets community. It refers to a radical change in the architecture behind Ethereum: moving from the consensus algorithm Proof-of-Work (PoW) to Proof-of-Stake (PoS). This is a deeper dive than our earlier short intro about PoS here. Changing the consensus algorithm is a big deal, as it determines whether blocks added to the blockchain are valid - in a way this defines the blockchain, as we will see. We’d like to share with you in some detail what this is about, as the consensus mechanism is a key technical part of the value proposition behind decentralized blockchains.

Image from Ethereum.org’s article about The Merge: it shows a spaceship that is under construction after it has already taken off - ‘fixing the plane as it flies’. This jokingly refers to the culture around Ethereum being different than e.g. around Bitcoin, where fundamental changes to the protocol are considered undesirable unless absolutely necessary.

The consensus algorithm is a core security consideration for decentralized blockchains, as it defines the ‘true’ state of the network at any given point in time: who has what asset balances and what is the state of the smart contracts deployed on the blockchain (if applicable). Bitcoin’s key innovation when it was launched was Proof-of-Work (PoW), where blocks are valid only if a significant number of computations has been made first. A block in the blockchain contains a list of signed transactions from sender addresses to receiver addresses and - if the blockchain has smart contract functionality - the state of the smart contracts at any given point in time. Nodes within the network only send blocks to other nodes if they are valid, so invalid blocks die out. A special kind of nodes are block proposers - we choose this term to make it general to chains outside of PoW systems. In the case of PoW systems, the block proposers are the miners. They receive incoming transactions and organize them together in a ‘block’. A block contains different metadata about the network, with the specifics depending on the protocol. The block header usually contains these metadata. The piece of information relevant here is the block hash - a unique number corresponding to the contents of that block. A hash is an example of a one-way function, a function for which it is easy to compute the output given the input, but not the other way around.

Example of blocks in the Bitcoin blockchain (source).

You usually read about PoW as miners solving ‘complex puzzles’, but this is somewhat misleading. What is really happening is that miners iteratively hash the contents of the block header using a different input number (‘nonce’) in the header. In the case of Bitcoin, one of the conditions for a block to be valid is that the hash of the block header starts with a specific number of zeros. There is no deterministic way to get the nonce that satisfies the zeros condition, so the only way is by trial and error. Of course, a miner could be ‘lucky’ and finds a number satisfying the condition right away, but this is very unlikely. The miner that first finds a hash satisfying the condition gets Bitcoin added to its wallet address. On top of that, the successful miner also receives the transaction fees from all senders in the block (but currently the fees per block are much lower than the mining reward for Bitcoin). This is how miners are incentivised and why mining can be lucrative. The probability of finding a block is such that on average it takes 10 minutes for Bitcoin miners to find a new hash. But note that this is all probabilistic, so it could happen that sometimes it takes longer or shorter, depending on when a valid hash is found. This mechanism thus ensures that a significant amount of computational ‘work’ must be done for a miner to produce a valid block.

So why is this important? The crucial thing is that the hash of the previous block header is added to the header of the next block. A hash is a technique from cryptography with the property that it changes completely even if you change only one detail about the information being hashed. For example, if we hashed this document, we would get some number as a result. If we changed only one letter, the hash would change completely. So by ‘chaining’ the hashes in the block headers, i.e. computing the hash of a block header including the previous hash in the block header, the latest hash is unique to the exact state of the blockchain at any given point in time. This is where the name ‘blockchain’ comes from. Suppose we changed one of the earlier blocks of transactions: then the hash in the header of that block would change. Since all blocks are linked together, all hashes of succeeding blocks would change as well. Nodes in the network check the hash in the block header and whether it satisfies the condition of leading zeros. Thus, in order for the new ‘truth’ to be propagated by nodes in the network, one would have to redo the search of valid hashes for all blocks following the one you modified. Moreover, someone changing a block has to compete with the rest of the miners to keep up, as nodes consider the longest blockchain the valid one. Therefore, changing the history of blocks is instractably costly, especially for ‘older’ blocks. This is why if you transfer assets on a blockchain, you wait for a certain amount of ‘confirmations’. The more blocks are added after the block with your transaction in it, the more secure it is, because it would take more work to change the history. Exchanges maintain a minimum amount of blocks that are needed per protocol to consider a transaction as ‘final’.

So that is how blocks are added to the blockchain under PoW. It provides a robust way to determine the ‘true’ state of the digital ledger - the chain of blocks of transactions - based on the fact that it is not possible to ‘guess’ a hash without performing significant effort - and using a significant amount of energy. This grounds PoW systems in the laws of physics (energy expenditure) and mathematics (no shortcuts to finding a valid hash), making it suitable as part of a global network where the rules are the same for everyone, independent of their background, beliefs and economic position.

The key question is now: are there any other ways to chain together blocks and determine the ‘true’ state of the ledger? This is where PoS comes in. Instead of energy being the scarce resource at stake, it is capital. In PoS, the block proposers are not miners, but so-called validators. These are nodes in the network that deposit 32 ETH as collateral. Within a fixed unit of time, a validator is randomly assigned to be the block proposer, organizing transactions into a block. Now, instead of a miner that finds a hash of the block header that satisfies a condition, other validator nodes check the block that has been proposed and sign the transaction if it follows the rules of the protocol. Validators have a public-private keypair just like any other Ethereum address and can sign transactions with their private key. So how does PoS ensure that validators follow the rules of the protocol? This is where the collateral is used: if a validator violates the rules of the protocol, part of - or the whole sum - of the collateral can be ‘slashed’, i.e. removed. Other validators can receive a bounty for reporting misbehaving validators, so are incentivized to do so. This incentivizes the network to cooperate and follow the same set of rules. This is how capital is the scarce resource securing the protocol instead of energy expense. Validators can be identified because they have to sign blocks and they are incentivized to stay online because there is a penalty for not remaining online a sufficient amount of time. A validator’s stake is partially slashed for every day of inactivity. This is different than in PoW, where miners can join and leave the network at will, as they have to spend energy anyway in order to participate. In PoS, multiple validators are needed to be present to sign blocks, while security depends on validators rotating frequently and at random, so there is a need for mandatory attendance.

Simplified overview of the PoS-based chain: a block proposer and committee of validators are selected randomly from the active validators. A committee votes on a block created by the block proposer each slot, containing transactions and state transitions for smart contracts. For more information, see this source. Note: using sharding, the process shown is run in parallel over up to 64 shards, greatly improving the amount of blocks that can be processed per epoch.

In more detail, what happens is roughly as follows: for each time slot of about 12 seconds, one block proposer and a subgroup of at least 128 validators are chosen from the active set of validators. After the block proposer creates a block, members of the committee of validators check if it follows the rules of the protocol (also e.g. checking for double spends) and sign (‘attest’) the block if they consider it valid, after which it is added to the blockchain. This happens for epochs of each 32 slots (~6.4 minutes). Just like in PoW, blocks are chained together using iterative hashing of the block header, making the history immutable without valid signatures from committees of validators.

So what is ‘The Merge’? The Merge is the event where Ethereum actually moves to the PoS consensus algorithm by a hard-fork. Concretely, this means that the nodes of the Ethereum network update their software to a new version that implements the PoS algorithm. It is a hard-fork, because the ‘old’ way of proposing blocks by miners is no longer valid and leads to blocks not propagating on the new chain. Note that the decision to upgrade node software is completely opt-in and voluntary and not forced by anyone. It is coordinated and facilitated by the Ethereum community, including the Ethereum Foundation. A number of nodes may keep running the ‘old’ PoW-based software, resulting in a separate chain that is incompatible with the rules of the upgraded Ethereum chain. Hard-forks are a recurring phenomenon in blockchains, because of their decentralized mechanism: as no single entity is in control, the state of the network emerges from the community of nodes, each making independent choices about which software to run. Notable hard-forks in the past have been Bitcoin / Bitcoin Cash or Ethereum / Ethereum Classic, which we will not treat here. It is expected that a sufficient number of nodes will accept the Ethereum upgrade and that the PoW chain will die out naturally, given how costly it is to keep running it while no one is likely to build on top of it.

The PoS-based chain of Ethereum is called the ‘Beacon chain’ and has already been running on its own since December 2020. By ‘on its own’ we mean that it has not been connected to the Ethereum main blockchain, but simply consisted of a group of validators forming and signing blocks among each other. After the Merge, the Beacon chain takes over as the consensus mechanism and nodes that have upgraded their software will look for signatures from validators as described above. The upgrade is supposed to go live halfway through September. Some key features of post-Merge Ethereum:

• Energy use of the network is reduced by about 99.95%, as PoW is replaced by PoS.

• The blockchain history remains unchanged, as the Ethereum main blockchain will continue adding blocks to itself, only under different rules for which new blocks are valid. Therefore, holders of Ethereum, ERC-20 tokens or NFTs don’t have to do anything.

• ETH issuance will drop by 90%, as validators are incentivized differently than miners, leading to lower issuance. This makes ETH scarcer as inflation drops. After a previous upgrade of Ethereum, called ‘EIP1559’, a significant part of the transaction fees in each block are ‘burned’ - i.e. removed from circulation. This means that at some point, Ethereum can even become a deflationary currency, with more coins being removed from circulation than added over a period of time. This would make ETH more scarce and thus potentially more valuable.

• Validators are expected to earn approximately 7% APR, creating a large interest from investors to stake capital in ETH validators. The down payment of 32 ETH which is at risk of being slashed if you don’t know what you’re doing opens the door for companies that provide ‘validators-as-a-service’. These companies provide a safe and convenient way for anyone to earn yield on ETH. Examples are exchanges like Coinbase and even open-source projects like Lido.

It is fascinating to see where PoS is going. It offers a way to build blockchains without large energy requirements. As PoW hashrate grows, it becomes more costly to become a miner, as it rewards economies of scale. It is therefore argued that PoS leads to greater decentralization of network governance, as anyone can boot up a node if they can put down 32 ETH as collateral, without the need to acquire specialized mining hardware. On the other hand, since capital deployed is the core metric for influence, how decentralized the system is depends on the distribution of ETH, which tends to be concentrated. Additionally, ‘validator-as-a-service’ type businesses would further concentrate the voting power into a few large entities. The trade-off between security, decentralization and scalability in blockchain architectures is known as the blockchain trilemma - a great subject for future articles. The PoS architecture enables new scaling solutions that allow for higher transaction throughput on the Ethereum blockchain. This is important, because of the notoriously high transaction fees that currently arise when the Ethereum blockchain is heavily used. A key example of a scaling solution is sharding - a technique that allows for groups of validators to work in parallel on new blocks independently. This feature is on the roadmap as the next upgrade to the protocol, made possible by the new PoS architecture. Sharding will allow up to 64 parallel instances of validators proposing and signing blocks, leading to a massive speed-up compared to the current sequential architecture.

However, since energy and capital are fundamentally different concepts, PoS makes different technical trade-offs in the context of the blockchain trilemma. Capital is more concentrated and more ‘sticky’: you only need to acquire it once to have a continuous influence on network governance, whereas PoW energy generation is fleeting and continuous, so an attack requires persistent effort in a race against other miners. Additionally, capital is not always scarce, as governments could theoretically create it at will, while energy is limited to real-world constraints. Furthermore, it is easier to deploy capital in a short period of time than to acquire specialized mining and energy generation equipment. Finally, the concentration of capital into a small number of entities poses concerns for the Ethereum’s promise of being a decentralized network. PoS is also not as battle-tested as PoW, which has secured Bitcoin for more than 12 years. This is the blockchain trilemma in action. A side-by-side comparison of these tradeoffs and their consequences on security, decentralization and scalability will be the focus of one of our future articles.

As we mentioned in our earlier article, not all applications may need the security and robustness that Bitcoin offers. The lack of smart contract functionality and sturdy but unchanging architecture make Bitcoin unsuitable to be e.g. the decentralized cloud computer of the world. This is where Ethereum makes different tradeoffs, allowing it to be more flexible and iterate faster, but at the cost of lower levels of security and decentralization. The large amount of developer activity on Ethereum is testimony to its strengths and the sprawling landscape of early applications like NFTs and DeFi show what its innovative culture can achieve. Different applications have different needs, allowing for a range of different architectures, each with different technical trade-offs that give rise to unique strengths and weaknesses. If PoS provides sufficient security to host decentralized applications all over the world, protocols that have adopted it will become extremely valuable alongside Bitcoin.

At Agga, we are thrilled about the breadth and pace of digital asset protocols and platforms, each making different tradeoffs and enabling a range of impactful applications. These developments affirm our conviction in both diversified investing and digital asset protocols. We see a world where Bitcoin and PoS-based blockchains coexist, each playing a different role within the ecosystem. We also see that there is much confusion about the terminology and the technical tradeoffs that are made, as they contain new concepts such as miners, PoW, hashes, etc. Usually, when a truly new concept is invented, it is poorly understood. It takes time, effort and education for humans to collectively understand revolutionary new concepts. This is where the investment opportunity lies, as tangible adoption will take time. There are two major pieces of uncertainty: what protocols will be successful and when mass adoption will occur. As these two are very hard if not impossible to predict in practice, we prefer a systematic investing approach in the digital asset space. At the same time, it is important to understand the basic technical details underpinning digital assets to fully appreciate the value they add as an asset class. In any case, the fact that adoption takes time provides room to build, experiment and add value to the space for crypto veterans, newcomers and institutions from traditional finance alike, especially in bear markets.

P.S. Shoutout to students and recent graduates: at Agga, we’re always looking for talented people to join us, who share our interest in the future of digital assets. Check out our website for the latest internships and vacancies: https://agga.capital/careers.

• Ethereum.org articles about The Merge: https://ethereum.org/en/upgrades/merge/

• Ethereum docs: https://ethereum.org/en/developers/docs/consensus-mechanisms/pos/

• Overview by Consensys: https://consensys.net/blog/blockchain-explained/the-ethereum-2-0-beacon-chain-is-here-now-what/

• Technical specification with Python code snippets: https://eth2book.info/altair/part3