In November 2017 at BeyondBlock Taipei, Vitalik Buterin laid out his vision for Ethereum 2.0: sharding for scalability and Casper as the consensus algorithm for the transition to Proof-of-Stake (PoS).
Sharding was laid out as “creating a blockchain where you have a hundred different universes, and each universe is a different account space, and you can have an account in some universe, or a contract in some universe, and you can send a transaction in some universe, and it can only affect things in that universe.” He continued to explain that all of the shards would be interconnected and would ‘share consensus.’
Eventually, the lines between these shards could possibly be blurred and could even have execution across shards.
Vitalik further mentioned how, as opposed to those chains which had settled on using some form of ‘super-node,’ he thought it was important that Ethereum could be run by a network of ‘consumer laptops.’ Vitalik did admit at the time that “there is no precise conception of what goes into Ethereum 2.0.”
What is Ethereum 2.0 now?
Ethereum 2.0 may include sharding and other changes in the future. For now, when ‘The Merge’ happens, it’ll simply just the change from the Proof-of-Work (PoW) chain as the primary chain, to the PoS chain.
It won’t even be possible for people to withdraw their stake until a planned hard fork several months after the Merge.
What is the Merge?
For months, people have began staking and participating in PoS on a ‘Beacon Chain.’ The Merge is when that consensus layer is combined with the existing state and actual transaction execution level.
Practically, it’s when PoS takes over security and consensus roles for PoW in Ethereum. It’s currently expected to happen at a ‘Total Terminal Difficulty’ of 5.875×10^22, around September 14th or 15th. Once it occurs, Ethereum will be entirely PoS.
What about the EthPOW fork I’ve heard about?
There have been discussions about a potential PoW fork of the Ethereum state at the time of the Merge. Justin Sun is involved, as he often is when there’s an opportunity to profit from the efforts of other protocols’ developers. His exchange Poloniex listed tokens ahead of the Merge to allow speculation on it.
The fundamental problem with non-majority forks for smart contract chains is that a variety of tokens are likely to only have value on a single chain. This is especially true for stablecoins where the issuer will often explicitly decide which chain will continue to have value.
Tether, Circle, and Paxos have all announced their intent to follow the PoS chain, which means their tokens on the other fork will be effectively valueless. This also means that any other protocols on the minority fork will have major issues — and lose value — because the assets won’t be worth $1 as expected.
This is compounded by the fact that there are other stable-assets like Dai that depend on the stable-value of these coins. These would also be expected to have serious issues, as would protocols that implicitly or explicitly expect Dai to be worth close to $1.
However, there may be opportunities for well-connected actors to profit in the first couple blocks as the value tries to transition between $1 and $0. Many of the people promoting this fork likely hope to do so.
Essentially, this means stablecoins decide forks for smart contract chains — if they’re allowed to become important enough in the ‘ecosystem.’
What is staking?
Staking is the process of participating in consensus for PoS chains. Ethereum 2.0 will use the Gasper consensus algorithm, which is a combination of Casper developed by Vitalik Buterin and Virgil Griffith, and the algorithm for LMD GHOST fork choice developed by Vlad Zamfir.
Users with at least 32 ether who can run a validating node are able to earn staking rewards. Staking rewards are provided for proposing a block or attesting to a block. Currently, proposers receive rewards if someone is slashed.
Once the validator is active it’s eligible to be chosen as either a ‘proposer’ who suggests the next block for the network, or as an ‘attestor’ who signs off on the most recently proposed block.
If you have more than 32 ether and want more staking rewards, then you’ll need to run more nodes as each is treated as having an ‘effective balance’ of 32 ether.
If you have less than 32 ether then you can’t individually participate in staking. You can choose to ‘pool’ your ether with other depositors through a service like Lido. This allows you to still receive some of the rewards from staking and doesn’t require you to manage your own hardware or the full amount of ether. It also had a token that allowed you to get liquidity before staking withdrawals are available, and was in part responsible for destroying Three Arrows Capital.
Currently no one who is staking can withdraw, and that is intended to change in a future hard fork.
What is slashing?
Slashing is the punishment for bad behaving validators. If a set of keys votes more than once on a proposal, proposes two blocks for the same slot, or tries to modify history they can get slashed. Slashed validators lose a portion of their stake and are forced to exit from the consensus process.
What is inactivity leak?
Inactivity leak is a penalty for a node not being online. The more nodes that are offline, the more aggressive this drain is. When the network is still working, this drain is relatively small, but if a very large number of validators go offline, this becomes a much more severe leak.
The intention of this design choice is to drop inactive validators below the minimum 16 ether required to keep staking, so that eventually online validators will again be able to run the chain.
Is staking secure compared to PoW?
That’s a complicated question. Probably. Maybe. Yes. No. Depending on how you want to structure your question, who you want to ask, and how you think about certain assumptions will strongly affect your conclusion to this question.
PoW has a tendency to centralize into mining ‘pools,’ whereas it’s more economical for a PoS staker to work independently.
However, because PoS delegates awards proportional to those who already have the largest portion rather than those who are spending the most, it can tend to have a ‘rich get richer’ effect.
Only, slashing provides an opportunity to impose a very strong cost on attackers. It can effectively destroy the value they hold, and eventually leave them unable to continue the attack as their nodes are removed from consensus.
Andreas Antonopoulus has discussed how in PoW if you are under attack in this way then one of the most appealing solutions is to change the PoW algorithm and effectively make all existing ASICs worthless (if there are other chains they are not truly worthless, but if Bitcoin were to change then their value would plummet).
However, this is a very broad recourse, and does not specifically target the attacker, where as PoS can specifically target only the ‘bad’ actor.
PoW has advantages for token ‘distribution,’ by forcing the burning of hard assets (energy) in order to acquire more coins, but it’s less clear that it has advantages for long-term security, especially when distribution of new assets slows or stops.
A full discussion of the relative security tradeoffs between the two setups is beyond the scope of this piece.
What happened to sharding?
Sharding was still a part of Vitalik’s vision of Ethereum in 2021. But since then, roll-ups have become better and a new concept called ‘danksharding’ has arisen as the future scaling option for Ethereum.
The original vision of sharding involved separate universes which would share security at the consensus level. These shards provided separate ‘execution’ environments where the actual transactions and interactions would occur, and would require additional structures to allow communication between them. The complexity of this vision has made it less appealing over time.
What is a roll-up?
Roll-ups are somewhat analogous to Vitalik’s ‘ghost chains‘ that he described in 2014. They are effectively separate blockchains that attempt to share security with the base layer.
Roll-ups perform their execution off the main-chain and then post data to the main chain. Optimistic roll-ups assume transactions are valid unless a ‘fraud proof’ is submitted. Zero-knowledge roll-ups involve the validators for the roll-up submitting a proof that shows the transactions they performed are valid.
They are also now the vision for Ethereum’s future.
What is danksharding?
Danksharding is a new vision for Ethereum scalability that is meant to deal with the fact that roll-ups create and post so much data to the base layer. It’s meant to provide a way for nodes to verify that data is available, without having to verify that data itself.
Ethereum intends to rely on a scheme called ‘KZG commitments,’ which rely on creating a polynomial that allows other nodes to verify the data was encoded correctly when combined with validators sampling to make sure that data is available.
For this plan to work appropriately it does require that there be enough honest nodes storing the data to piece it together, and that nodes need to be able to communicate with each other to recreate the full block.
Danksharding relies on proposer/builder separation in order to function. However, it’s not part of the Merge and is instead on the roadmap of Ethereum.
What is proposer/builder separation?
Proposer/builder separation (PBS) is a way to separate out the building of blocks and the actual proposal of the blocks to the rest of the network. The idea is that Miner Extractable Value (MEV) has a centralizing effect in PoS, because it allows certain nodes to receive rewards disproportionate from their stake. Separating out those roles is meant to democratize access to MEV and to blunt that centralizing incentive.
The implementation of PBS is not final, but the general concept is that block builders would submit bids and block headers. Proposers would select a header and bid, likely favoring the largest bid. A committee attests that the block header was the winning header. The block builder shares the block body, and then a separate committee attests to that and it’s added to the chain.
Proposer/builder separation is not a part of the Merge but is on the roadmap for Ethereum.
However, smaller validators who worry about the process of needing to build blocks can instead rely on tools like Flashbots MEV-boost. This is a tool that allows validators to configure their nodes to take blocks sent to them by ‘relayers’ who are sending the blocks full of MEV built by builders.
Why is PBS important for danksharding?
In short, because danksharding shifts more responsibilities to the block builders. They are expected to download all of the different data, a role that increasingly would be beyond individual validators. This is exacerbated with roll-ups as they post a large quantity of data. Validators will instead only check that a small portion of the data in a block are present.
This effectively ‘shards’ the responsibility of keeping and checking that all necessary data is available, allowing for greater scaling.
What other things are needed or useful for danksharding?
Ethereum Improvement Proposal 4844 is meant to introduce ‘Proto-Danksharding’. It’s not danksharding, as all validators still need to validate all the data, rather than responsibility being ‘sharded.’
What it does do is change how the data from roll-ups is stored. Instead of being around indefinitely, they will instead be pruned after a month, which helps reduce the amount of state that validators will need to continue to carry.
It also changes how the Ethereum fee market works, breaking storage of this data out from the existing Gas calculation.
EIP-4844 is not a part of the Merge but is on the roadmap for Ethereum.
Ethereum Improvement Proposal 4444 is a change in how clients serve data to other nodes. It demands that nodes no longer broadcast historical data older than a year to other nodes. This effectively ends ‘full sync,’ where people would sync from the first block. Instead they would start from a checkpoint and sync from there.
EIP-4444 is also a part of the roadmap for Ethereum and is not part of the Merge.
What happens to the data?
The short answer is that it’s stored by other people. Who are those people? Well that’s a little complicated.
Individual applications will likely need to find ways to store and share that information, so Optimism and Arbitrum, for example, will be incentivized to ensure that the data needed for their applications is still available.
Larger infrastructure providers like Infura will likely end up having an important role in maintaining this data as well. EIP-4444 specifically cites efforts by The Portal Network and The Graph to provide markets and availability for the data.
What will change for users?
Very little! Especially for those who are not actively participating in consensus. For those who are, they will need to change from their likely GPU based setups for PoW to staking setups for PoS.
Will this make my transactions cheaper?
No, at least not meaningfully. However, it might in the future. If Ethereum begins to implement the steps towards statelessness, and no longer requires validators to download all the data, then it will likely become possible to increase the gas limit.
This is unlikely to occur until after PBS, after EIP-4444, after EIP-4844, and so should not be expected immediately. Many of these same changes will also allows roll-ups to increase their throughput, and the amount of transactions possible will go up. However, the Merge and the shift to POS will itself not make your transactions cheaper.
Will this make my transactions faster?
No, at least not meaningfully.
Will this make number go up?
It’s impossible to say but do let us know on Twitter.