A Simple Understanding of Ethereum Layer 2 Solution Arbitrum

This article was published on Blue Fox Notes.

Recently, both Uniswap and Sushiswap are planning to use Arbitrum's L2 solution, making Arbitrum a focal point in the Ethereum L2 landscape. So, how can we simply understand Arbitrum?

Arbitrum is also an L2 solution in the rollups series

There are several Ethereum Layer 2 solutions, among which the Rollups series is the most notable. Previously, Blue Fox Notes introduced Optimistic Rollups and Zk Rollups, which can be referenced in the article "Understanding ZK Rollup and Optimistic Rollup: Important Scaling Directions for Ethereum."

Arbitrum is also a solution in the Rollups series. These solutions essentially place security on the Ethereum chain while executing computation and storage off-chain. In simple terms, participants submit their transaction data to the Ethereum chain, where users can view the transactions, but the computation and storage of the transactions occur off-chain. Arbitrum periodically sends a complete state hash of what has occurred on Arbitrum to Ethereum, and this hash is placed on-chain as a result, achieving immutability and finality.

Relatively speaking, Arbitrum is similar to Optimistic Rollups in terms of framework thinking; both belong to the category of fraud proofs for result verification, while ZK Rollups fall under the category of cryptographic validity proofs.

In terms of assumptions, fraud-proof Rollups assume that the results submitted on-chain are valid, and anyone can extract and execute Layer 1 transaction data to check whether the contract execution is correct by comparing states. Cryptographic proof Rollups use mathematical methods to achieve validity through zero-knowledge proofs, which are essentially impossible to falsify. Thus, cryptographic validity proofs are proactive proofs, while fraud proofs are passive proofs, leading to disputes when people disagree on the results, requiring arbitration.

Overall, ZK Rollups have higher security and achieve finality faster, which means their withdrawal times are quicker, making them more suitable for scenarios like transfers. However, both Optimistic Rollup and Arbitrum have challenge periods that need to end before final confirmation and withdrawal can occur, which is a drawback. On the other hand, generating cryptographic validity proofs for ZK Rollups incurs high computational costs and is expensive, requiring more time for maturity and implementation.

In the long run, ZK Rollups may be the ultimate solution, but their maturity and implementation will take longer, while Ethereum's scaling needs are more urgent, as evidenced by the recent rise of smart chains. The slow rollout of ZK Rollups gives opportunities to Optimistic Rollups and Arbitrum.

Therefore, from a medium to short-term implementation perspective, fraud-proof Rollups (Optimistic Rollups and Arbitrum Rollups) have faster deployment opportunities. Fraud proofs are inherently "optimistic," presuming that all submitters are good actors unless proven guilty. As long as there is one validator, a fraud proof can be submitted, ensuring safety. Thus, the assumption of having one honest validator becomes crucial. Only one validator is needed to submit an online fraud proof, with a fraud proof time window lasting up to a week (or more), which also leads to slow final confirmation and prolonged withdrawal times.

In summary, fraud proofs and cryptographic validity proofs each have their pros and cons. ZK Rollups are non-interactive, solving problems through mathematical methods, while Optimistic Rollups and Arbitrum Rollups are interactive. In this sense, Optimistic Rollups and Arbitrum Rollups are similar, although they differ in specific paths.

Arbitrum is a multi-round interactive Rollups solution

From the above, Arbitrum and Optimistic are both fraud-proof Rollups solutions, and the core issue for both is: how to verify the correctness of the results? Could someone send malicious proofs? Therefore, both have challenge mechanisms. Validators can send assertions to the chain and raise disputes. If the assertion is false, they will lose their staked collateral.

The main difference between Arbitrum and Optimistic lies in how they resolve disputes. When a validator submits a rollup block to L1 and someone believes it is incorrect, what should be done?

Arbitrum uses a multi-round interactive protocol to resolve disputes, breaking down large-scale disputes into smaller ones until the most critical step is identified, and then determining its correctness through Ethereum contracts. By splitting disputes, Arbitrum aims to achieve a more efficient resolution.

Participants in Arbitrum can stake collateral, asserting that a certain state will eventually be confirmed. If the assertion is wrong, the user's staked collateral will be forfeited. Two users staking on different blocks means only one of them is true. To prevent attacks, the disputing assertor must stake collateral assets to deter malfeasance.

In the initial design, the Arbitrum protocol processed one dispute assertion at a time. A participant would raise a dispute assertion, which would have a challenge period during which anyone could challenge it. If no one challenges, the dispute assertion is considered valid.

However, this linear progression model has bottlenecks. Activating one dispute assertion at a time is inefficient, and the VM process is limited. At the same time, malicious actors can deliberately delay the VM by challenging dispute assertions. Even though they need to stake funds, they can still cause delays.

Therefore, in the new design, Arbitrum can handle multiple dispute assertions simultaneously, rather than linearly. The difficulty for malicious actors to slow down the process is significantly increased. Currently, Arbitrum employs a method of processing multiple dispute assertions simultaneously, with one staker handling one assertion at a time, while different stakers can handle multiple assertions concurrently.

As assertions are confirmed as valid, the state of the VM continues to progress. Generally, people will choose honest behavior, as this ensures their staked collateral will not be forfeited. Unless there is a reason for malicious attack, there is no need for people to stake their assets on the wrong branch.

To achieve a trustless characteristic, in Arbitrum's design, any honest participant can advance the correctness and progress of the VM. If a participant consistently stakes on the correct branch, they will win all dispute assertions. If others disagree, they will only lose their staked collateral. Only when honest participants come together can they avoid punishment.

* (Alice and Bob stake on different branches, from Arbitrum's Ed Felten)*

If Alice in the image above asserts that her proof is correct, then Bob's staked collateral will be forfeited, while Alice will benefit.

In single-round interactive Rollups, assertions include the results of each call, and challengers point to specific calls within the assertion that have erroneous results. On-chain contracts simulate the challenged calls and check for errors. If there is an error, the entire assertion is canceled, and its assertor loses their staked collateral. If no successful challenges occur after the challenge window, the assertion is accepted and has finality.

In multi-round interactive Rollups, there is also a challenge time window, and there are multiple interactions between the assertor and the challenger, with on-chain contracts acting as referees to determine who is wrong. The main consideration for the multi-round interaction design is to minimize on-chain work required to resolve disputes. By maximizing the number of interactions between challengers and assertors, the on-chain workload is minimized. Single-round and multi-round interactions essentially balance on-chain costs and dispute resolution times.

Both single-round and multi-round interactive Rollups require writing all calls to contracts and their data on-chain, but they differ in what they need to submit as part of the assertion. The single-round interaction model requires simulating the complete call on-chain, which is more costly. In contrast, multi-round interactions can reduce the scope of disputes and write relatively less data on-chain, resulting in lower execution costs. However, it is not without drawbacks, as it increases rounds and time.

Additionally, the multi-round interaction model of Arbitrum may face delay attacks. Since the system is trustless, malicious attackers can slow down the process, but they will incur costs as their staked collateral will be forfeited. It is also important to note that malicious actors cannot prevent honest participants from continuously building honest branches or stop honest participants from obtaining final confirmation on honest branches. What they can do is stake collateral on the wrong branch and delay on-chain confirmation of the honest branch. They can create multiple dispute assertions, and if there are enough honest participants, they can simultaneously address these dispute assertions and capture the attackers' staked collateral. In other words, the more honest participants there are, the more collateral assets the attacker needs to conduct delay attacks, making the attack more difficult.

Moreover, in multi-round interactive Rollups, effective "assertions" can become "trustless finality," allowing anyone to be confident that their final confirmation is inevitable. To ensure their final confirmation, users can participate in the protocol to defend the assertion, and even act alone, they can promote the occurrence of their final confirmation.

If there are no challenges, the confirmation speed of multi-round and single-round interactions is similar; if someone challenges, the time consumed in multi-round disputes will be slightly longer. In this regard, single-round interactions have an advantage. Single-round interactions do not allow confirmation delay attacks, while the benefit of multi-round interactions lies in smaller on-chain data, allowing them to handle contracts that exceed Ethereum's gas limit (for computation and storage).

Fraud-proof Rollups have a certain time cycle for final confirmation. Therefore, transfers through Arbitrum cannot achieve rapid finality like ZK Rollups. A possible solution is through third-party intervention. Third parties can immediately transfer funds to users, but users will need to pay an early withdrawal fee. Generally, third parties believe that uncertain transfers will eventually be realized, and they will stake collateral on honest outcomes. Since anyone can transfer funds to users needing quick withdrawals, a market will emerge, leading to balanced fees that are neither too high nor too low.

Arbitrum's EVM Compatibility

Arbitrum aims for complete compatibility with Ethereum, supporting EVM without the need to rewrite programs. This complete compatibility is a significant move that we have seen the power of on smart chains. It is the most important measure for smoothly embedding into Ethereum's current ecosystem.

dApp developers write their smart contracts using the Solidity language, which are compiled into Arbitrum VM and can run on Arbitrum Rollups.

Progress of Arbitrum

The Arbitrum technology has currently launched Arbitrum One. Arbitrum contracts are deployed on the Ethereum mainnet and have begun accepting projects, currently in the mainnet testing phase. According to Arbitrum, over 250 teams have applied to enter, and it is open to all developers. In the coming weeks, we can expect to see more DeFi projects entering the Arbitrum ecosystem.

Currently, some important participants in the Ethereum ecosystem, such as Etherscan, are also supporting the Arbitrum One chain.

Ethereum and L2 Implementation

Interestingly, Arbitrum does not have a native token and uses ETH as fuel, with all fees paid in ETH, which is also beneficial for ETH, not to mention the potential future demand for it.

For now, Ethereum's scalability requires a combination of short-term and long-term strategies to meet its rapidly growing needs, especially the explosive growth demands of DeFi and NFTs. In the long run, L2 may see ZK Rollups prevail, but due to their high costs and delayed implementation, it may take a long time to digest. Meanwhile, Arbitrum and Optimistic Rollups may land earlier; although their withdrawal times are long, they are not without solutions. Additionally, growth in practice may also inspire new ideas.

In addition to the gradual implementation of L2, with the possible launch of EIP-1559 in July this year and the potential PoS at the end of the year, 2021 is destined to be an unusual year for Ethereum.