After The Merge
What's next for Ethereum
Hey there,
Substack finally allows referral campaigns. We have been gifting paid subscriptions to other good writers for readers sharing our content for a while. You can now share our content and receive paid access to Unchained or Kerman’s newsletters. If you enjoy reading our content, consider sharing the link below for free goodies. We may or may not have something in store for the top referrer.
We have been tracking games, Web3 social and payments infra on this newsletter for quite some time. Part of what enables these applications are improvements at the protocol layer. We covered a few of them in the past through our pieces on account abstraction and ordinals. But occasionally, you have enough progress on a protocol to warrant writing on the improvements that may come soon.
If you are trying to trade volatility - two impending factors may affect the market. One is the constant repricing of the probability of a Bitcoin ETF. The other is the rapid decline in activity in private markets. We will go into both of them at another point, but for now, let’s look at the technological advancements being made on Ethereum as a protocol.
Why even bother with any of this? As of late June, the market cap of all crypto assets is around $1.2 trillion. Out of which, ~$590 billion is Bitcoin, and ~$130 billion is stablecoins, where innovation is slower. Ethereum is 47% of the industry's market capitalisation once you exclude stablecoins and Bitcoin. Among the smart contract platforms (total ~$340 Bn market capitalisation), 2/3rd is Ethereum. You cannot build a sound understanding of the industry while ignoring Ethereum.
Ordinals and the ETF conversation have driven an incredible amount of interest in Bitcoin, but the fact remains that Ethereum has the most momentum. Between RWAs and CBDCs, we see the protocol evolve from being that smart contract protocol used only by DeFi hackers and NFT bros.
We have gone from ‘rollups will be launched’ to ‘we need to reduce the cost of data that rollups post on Ethereum’. This piece takes the reader through the most important recent Ethereum upgrades and shares what’s in store. Ethereum’s roadmap has changed over the years and will most likely keep changing as the technology matures, and we are presented with new information.
Small editor’s note before we start. Most pieces look at the narrative and broader implications of things. I wanted to do that with much of this piece too. But roadmaps for technical products are not up for creative interpretation. So as much as possible, we have stuck to a very factual explanation of what may occur in the future. We have broken down concepts as much as possible, but you may notice this one being a bit more dense than what you are used to.
What’s New Today?
Over the past month, you may have seen quite a bit about the “merge” on your timelines. It is a transition Ethereum made from Proof of Work (PoW) to Proof of Stake (PoS). In the case of PoW chains, external machines engage in a race to determine who gets to create a block depending on their ability to find certain numbers that satisfy predefined conditions. In the past, the energy consumption required by PoW Ethereum has been a point of contention for artists and the general public to embrace Ethereum as a protocol on which they can build.
It was a parallel chain that ran on Proof of Stake. Most of Ethereum’s functions continued to exist on the PoW chain, at which point functions like transactions, smart contract instructions and balances were also moved to the PoS chain. The merge bought together the execution and consensus functions of Ethereum as a chain to a PoS-based system earlier this year. Instead of rewards to miners, they go to individuals staked on the network.
As of April 2023, the network began allowing users to withdraw their staked ETH on the network. Earlier, if you had staked ETH, there was no way you could get those tokens out for an emergency. Now you could, after a brief wait time. There were a few more changes that were integrated with these transitions. A few notable ones are mentioned below.
EIP 3651 reduced gas costs related to MEV payments while accessing coinbase addresses (addresses associated with receiving block rewards)
EIP 6049 warned everyone about using the SELFDESTRUCT opcode since developers intended to change how it was used. These changes are mentioned in the later section.
EIP 3855 introduced a new instruction to reduce gas costs
EIP 3860 brought changes to the Initcode (a code executed when a contract is created) in a way that limits it and introduces a gas charge to avoid its abuse.
Although Ethereum’s current roadmap will take several years, its Cancun-Deneb upgrade is slated for later this year. The long-term roadmap is broken down into four parts – scalability, security, user experience, and future-proofing. We break down what each of those looks like further below.
Scaling 200 Times
Scaling is the Achilles hill of all decentralised public blockchains. The original idea of Ethereum was to achieve it via sharding (splitting the network into multiple networks to allow parallel execution) the Ethereum blockchain. But as the concept of rollups emerged, the idea of sharding took a backseat, and Ethereum’s roadmap became rollup-centric.
When it comes to Ethereum scaling, you’ll hear two new terms - Danksharding and Proto-Danksharding. We’ll gradually unpack these terms but know that Danksharding is the final form and Proto-Danksharding is the intermediate step to achieve it. The idea of a rollup-centric Ethereum is that it should mould itself in a manner that it acts as a host for several rollups.
Before moving on to the dreadful terms mentioned above, here’s a quick refresher about rollups. Ethereum is slow and often gets expensive when the demand increases. Demand here refers to the need for Ethereum’s blockspace and its VM (virtual machine) increasing. So, with rollups, we alleviate the load on the EVM by taking computations elsewhere.
Different flavours of rollups (optimistic, validity, zero-knowledge) get created based on how they post transaction data (whether they post it on Ethereum or elsewhere) and prove that the calculations they performed are correct and in accordance with Ethereum’s rules. Vitalik’s breakdown of Rollups from 2021 is useful if you want to learn more about them.
Rollups input batches of transaction data (compressed) onto Ethereum. These inputs are referred to as calldata. There are two predominant shortcomings with calldata
There’s a limit (937,000 bytes) to how much calldata can be included in a block. Why does this limit exist, you ask? For the same reason that Ethereum and Bitcoin have limitations to the size of each block – to ensure that the system requirements to run nodes and validators remain in check to protect decentralisation.
And for the same reason, we can’t just increase the calldata limit to whatever number we want. This limit also restricts how scalable rollups as a solution are or can be.
Calldata is expensive since every node on the Ethereum network processes it and lives on it forever, but rollups don’t need this data forever.
Proto-Danksharding, or EIP-4844, is a way to enhance the calldata limit. Instead of calldata, this EIP proposes introducing a BLOB (Binary Large OBject) space that gets attached to blocks. By getting attached to a block, we mean that this space is not visible to the EVM or the execution clients (which means that the EVM doesn’t have to store or process it. This blob space is available for consensus clients. When rollups post data into blobs, consensus clients attest that they have seen the data.
This data is available for sufficient time (1-3 months) for everyone to see and verify whether rollups have indeed posted correct data. After this time, consensus clients delete the data, and only their attestations remain. This way, consensus clients ensure they don’t suffer from bloating. On top of this, the EIP also proposes ways to easily verify this data so that the verifier doesn’t need complex machinery to challenge rollups.
Rollups help make transactions cost 12 -33% of the Ethereum L1. The changes brought by EIP 4844 are expected to reduce costs associated with rollups’ L1 dealings to 5% of the current expenses. The idea is to scale Ethereum to a point where transactions cost less than $0.001.
Improving Validator Management
Two solutions are in the research stage that should soon contribute to security: proposer boost and view merge. There is a discussion on replacing the idea of proposer boost with view-merge. A simple explanation of both terms is provided here, but if you want to dive deep, you can start here and here. Proposer boost means giving extra temporary weight to the block producer, making re-organising the chain and replacing the block more difficult.
In the case of view merge, validators are to merge their views of the state of the chain (the state of the chain is nothing but the entire view of the balances of addresses and contracts, etc.) before they attest to a new block. This helps prevent attacks that rely on validators not having a temporary consensus regarding the chain's state.
One of the ways to attack Ethereum is to take control of the large validators. The key that controls validators is spread across many devices to avoid such threats, making it difficult to obtain its power. The distributed validator technology (DVT) ensures no single points of failure in the validator stack.
Think of this as multi-sig but for nodes. Validators can split their private key across multiple computers. This achieves two critical goals. First, an attacker will now have to break into multiple computers to get access to the validator’s private key. Had Ronin been using DVT, the attacker would probably have had to work harder to access the private key controlling the bridge’s funds worth $600 million+. Second, hardware failure of one of the systems will not affect the validator’s node. This avoids slashing penalties due to downtime.
Apart from getting control of the keys, the attacker can identify the next validator and try to spam them with denial of service attacks (DoS). Secret leader election (SSLE) allows only the selected validator to know they have been chosen to produce a block. This ensures that attackers have no clue who will produce the next block.
In addition to the block rewards, MEV or maximum extractable value, has been one of the significant sources of revenue for the validators or block producers in Ethereum. They typically do so by re-arranging transactions (and introducing transactions of their own) in a manner they can profit from.
Ethereum proposes to share these more fairly across validators by introducing proposer-builder separation (PBS). Currently, the same validator creates and broadcasts the block. The block producer (or builder) and broadcaster (or proposer) are separated with PBS. The latter gets to choose a block from a market of block builders.
Easier Transactions & Lighter Nodes
Companies obsess over the number of clicks while designing user experiences for their customers. This is partly for two reasons –
The lesser the number of clicks, the lesser the cognitive load on the user.
More clicks equals more time. When fighting for users’ time and attention, you better ensure they are hooked onto your application. If a task takes too long to execute and someone reduces the time before you do, you will likely lose users.
While this is the case with almost every web2 company, crypto is far from achieving UX optimisation. We have a long way to go if we want users to use our applications because they are good irrespective of the decentralisation ethos.
The thing is, we already have innovators using our products. Selling the fundamentals of blockchain tech will not help us achieve ‘mass adoption. DApps must stand shoulder to shoulder with traditional applications to onboard the masses. This means that protocols like Ethereum need to change several things regarding UX. This article discusses transactions and accounts in the form of account abstraction and improving node infra so that the average user can see and feel what it is to run a node and not rely on any centralised party to verify if they have received their money.
Account Abstraction
The most important upcoming upgrade to improve user experience is account abstraction. We have already laid the case for it in this article. For a quick recap, EIP 4337, or Account Abstraction Using Alt Mempool, deployed to the mainnet in March 2023, was the first step toward account abstraction.
It brings elements of account abstraction without changing the existing protocol. Think of 4337 as a new layer through which existing wallets can act as smart contracts instead of just addresses. It introduced a special transaction type called user operation used to interact with smart contracts.
These user operations are held in a separate memory pool. Bundlers or block builders package these transactions and submit them to the network. But what is this achieve, or what will it achieve? Biconomy does a good job of exploring the possibilities here in short gasless transactions, social logins, batched transactions, paying gas in tokens other than ETH, and more.
We covered how it works in the past, so we’ll use this as a refresher on AA adoption. As of June, over 20k wallets interacted through AA-enabled wallets. It is quite likely that this standard will entirely power the next wave of applications in gaming, media and Web3 social. We recommend downloading Vybe wallet if you want to test if for yourself.
Better Node Infrastructure
Do you run a node for Ethereum? Likely not. It requires significant devices with significantly high memory, storage, CPU, and technical knowledge. To help decentralise nodes, the goal should be to reach a point where hardware requirements do not come in the way of people interested in running nodes. How do we get there? Ethereum defines something called statelessness. It does not eliminate the need for nodes, but changes how they handle data.
To understand that concept, we need to understand what weak and strong statelessness are in the first place. Weak is where the responsibility of storing the full state lies only with the block proposers, whereas the rest of the nodes verify blocks without storing any data. Strong statelessness is where state data is not stored in blocks. But don’t we need state data to confirm the block is valid? It turns out there is something called a witness.
Before understanding how witnesses work, here’s a quick refresher on verifying a block. Using the state data in a block, nodes compute the state root (by hashing transactions together until a single hash is found). A block is verified when this computed root is the same as the one provided with the block.
A witness contains fragments of the state data required to compute the root hash. With this witness, nodes can verify whether the block producer has executed transactions and updated the state correctly.
Fine, now we understand that we need witness data to replace the state data in a block so that nodes can figure out whether the proposed blocks are valid. But this also means that the witness data needs to propagate quickly to all the nodes. If the size of this data is on the higher side, only the nodes with high bandwidth connections can verify blocks. This is where the centralisation vector lies.
Ethereum currently uses Merkle trees to find the state root of the block. The issue is that the witness data obtained from Merkle trees are bulky and cannot support stateless clients. As a step towards strong statelessness, we have something called Verkle Trees. Compared to Merkle trees, they are 1/23rd in size and can provide witness data that can quickly propagate through the network.
Verkle trees are currently running on the test net. In the next phase, when clients are updated to support them, they will run on private and public clients. Between faster propagation and lower amounts of storage needed for witness data, we may see a generation of low-cost node providers coming to market. (Do let us know if you are building it).
All of these changes are occurring in parallel. Some may evolve to be something else when they come to production. We do not have a definite view on how the roadmaps evolve. But for now, it is safe to say that the prices likely do not reflect the intellectual density of the talent working on Ethereum today. Surely, there can be other L1s (or L2s), but replicating a network of contributors shipping at the pace these folks do, may be difficult.
We may see technical changes by the end of the year with the Cancun-Deneb upgrade. We don’t go deep into each of them, but for the technically inclined, here’s a quick breakdown of what it would include. Fair warning - there’s a ton of jargon involved.
EIP 1153 – introduces new ways (via new operational codes or opcodes) to interact with transient storage. Transient storage is nothing but temporary storage that is created and destroyed in a single transaction. Because this storage doesn’t require disk access, it has much lower gas consumption.
EIP 4788 – while this is still in the works, it aims to give smart contracts access to beacon root from within the EVM. Okay, I get that it doesn’t make any sense. Smart contracts currently access the beacon chain's state via different oracles. With this EIP, smart contracts can read the beacon chain’s state without relying on oracles, making smart contracts more trustless.
EIP 5656 – introduces a new instruction called MCOPY that allows smart contracts to copy data from one place to another. To do this, two opcodes MLOAD and MSTORE are currently used, which are much more gas-intensive.
EIP 6780 – limits the use of the SELFDESTRUCT opcode.
Its current use is to destroy a smart contract and transfer its ETH to a designated address. It was created as a response to the DAO hack in 2016. But smart contracts seldom use it. So to make contracts more predictable, 6780 limits the opcode’s use to only two cases – applications that use it for retrieving funds continue to work, and applications that use SELFDESTRUCT in the same transaction as they created a contract will continue to work.EIPs 4788 and 4844 mentioned above apply to the consensus layer as well.
EIP 7044 – signed voluntary validator exits (validators unlocking their ETH from staking contracts) are valid only for two network upgrades. This EIP proposes to make signed voluntary exits perpetual. This will simplify the staking experience and improve the overall staking UX.
EIP 7045 – increases the maximum slot that an attestation can be included to the end of the next epoch. Let’s break it down. An attestation is a message by a validator that they have seen a block. An epoch is a period of 32 slots, each 12 seconds long. Ethereum shuffles validator committees every epoch. The attestation mentioned earlier is valid only for certain slots. This EIP increases the number of slots for which the attestation is valid, thereby improving the's security of Ethereum.
What Next
We have been debating internally about protocol economics. There’s the age-old argument that SMTP, HTTPS and RSS have all changed the world without requiring people to pay heavily for it. Is a variation of the same possible for Ethereum? We are thinking of the answer for the next article, but here’s what’s evident for now. Ethereum is an intellectual black hole, attracting some of the smartest people of our time and enabling use cases we did not think of a few years back.
These applications are in their infancy, like Netflix in a pre-broadband era. Changes we see to the protocol in the form of EIPs require massive coordination to develop code and its deployment and ensure nothing breaks. We look at the price charts and often presume there isn’t much happening. But if this piece has anything to convey, it should likely be that an army of distributed developers is doing a ton of work that is often taken for granted.
Getting on Linux’s IRC servers and asking for help downloading different operating systems was fun a decade back. Developers in 1990s, contributed to those systems. We no longer discuss changes to Linux distros with the same excitement because many are stable and used across devices. There may come a time when the same is relevant for Ethereum too. Users may take for granted that the protocol is ever-evolving and that they don’t have to worry about anything. But that is the end state. We are not there yet.
Chains in our industry go down often. Some have chain re-org as frequently as a young working professional has anxiety attacks. There is work to be done.
Until we reach there, it is time to buidl.
Saurabh Deshpande
Update: An original version of this article referred to danksharding as “danke-sharding”. The editor we use likely thought it is a German term. Gratitude to the kind souls at LobsterDAO for pointing it out.
Telegram and Pitch Decks
Join in with close to 5000 researchers, investors, founders & overall great human beings. We don’t exactly talk much, but it would help you stay close to what we are focusing on & connect with others building cool things.
Fill out the form below if you are a founder building cool things and in the process of raising money or looking for feedback on what you are pursuing. We like the builders.