Layer 2 (L2) plays a pivotal role in scaling not just Ethereum but the entire blockchain space. Many high throughput Layer 1 (L1) blockchains witnessed a significant drop in user adoption during the bear market, in contrast L2 have demonstrated resilience and even attracted more users during the same period. This article delves into the evolving landscape of L2 protocols, their architectures, fraud systems, zkEVM, L2 business models and features. It also explores the distinctions between L2 appchains and L1 appchains, briefly touches on modular rollups, and discusses an upcoming Ethereum upgrade set to reshape the L2 economic landscape.

 

Layer 2 inherits Ethereum's security while providing faster and more cost-effective transactions to users through EVM-compatible wallets. L2 serves as the foundational infrastructure for launching protocols on L2 or deploying their own L2 to have custom functions tailored to specific business models, a crucial factor for institutional adoption.

The L2 landscape has evolved significantly. First-generation L2 solutions like Polygon, Optimism, and Arbitrum focused on scalability while inheriting security of Ethereum, aiming to compete in metrics like transactions per second, fees, and on-chain applications. Optimism and Arbitrum gained widespread adoption in 2022 and 2023 during their airdrop events, marking the peak of Optimistic rollups. Second-generation L2 protocols revolve around zero-knowledge (ZK) rollups, such as zkSync Era, Polygon zkEVM (powered by zkSNARK), and Starknet (powered by zkSTARK). Capital poured into zkSync Era due to expectations of lucrative future airdrops, while Polygon zkEVM and Starknet played catch-up. Current ZK rollup fees are less competitive than optimistic rollups, but they are expected to decrease with increased transaction volumes. During ETHCC 2023 in Paris, Linea (a ZK rollup by ConsenSys) and Mantle (a EigenDA powered modular Optimistic rollup by BitDAO) were introduced, marking the onset of the third generation of L2 protocols.

There are many other great L2s not mentioned above. For example, Metis, Scroll, Boba and application specific rollups such as dYdX, Immutable X, Loopring and so on.

 

Polygon sidechain were the first notable L2 on Ethereum, had a peaked token valuation of $20billion in year 2021. Sidechain is blockchain that allow asset from Ethereum to be bridged over and run in parallel with Ethereum, but does not derive the security or data availability from Ethereum. In contrast, rollups bundle thousand of transactions into a single transaction on Ethereum, make reverting the rollup transactions impossible without reverting Ethereum, hence inherit the security of Ethereum. There are two approaches to rollups, Optimistic and ZK. Optimistic rollups “optimistically” assume transactions are valid, until it is challenged and proved wrong with a fault proof mechanism. A good fault proof would allow one honest actor to win against any number of adversaries. Currently there are no rollup with fault proof in productive, however Arbitrum had released an interesting fault proof mechanism, BOLD which worth a read. ZK rollups do not need fault proof mechanism, instead ZK rollups securing the transactions with validity proofs where transactions are cryptographically computed off-chain, and then compressed data is supplied to Ethereum. Therefore Optimistic needs a significantly higher window period for withdrawal compared to ZK. ZK rollups gas cost higher than Optimistic rollups currently, due to higher off-chain computation costs. However, ZK rollups will cost less with increasing transactions executed in the rollups until it reaches an optimal amount for Ethereum batch.

Other types of Layer 2 including, plasma chain, it works just like sidechain but it posts data back to the main chain, hence inherit the security of Ethereum. Enshrined rollup, it works like a classic rollup but it shares the consensus integration at the Layer 1. Enshrined rollup performs the computational work off-chain and then posting the result back to Ethereum. This will increase Ethereum transaction throughput, as Ethereum does not need to perform computations but only check the validity of the submitted proof, which is a much less resource-intensive task. Lastly, Validium, it works like ZK rollup but it does not post the data availability to Ethereum.

 

There are two usual types of ZK proving systems, ZK-SNARKs and ZK-STARKs. ZK-SNARKs are based on elliptic curves for their cryptographic function for improved security and confidentiality, while ZK-STARKs are based on hash functions. ZK-STARKs are more scalable (but has larger proof size), more transparent in initial setup phase and are quantum attacks resistant compared to ZK-SNARKs.

 

A zkEVM is an attempt to build an Ethereum Virtual Machine (EVM)-compatible for rollups with zk-proof while preserving the EVM codes and the knowledge derived from it. zkEVM can be categorised it into type 1, type 2, type 3 and type 4, which is the measure of compatibility and performance of the rollup. Type 1 has the highest compatibility, it allows execution clients to be shared for rollup blocks, but it also takes the longest time to generate proofs. One example working toward this direction is Taiko, currently in testnet. Type 2 has minor changes that do not touch the application layer, compatible to almost all Ethereum dapps, but still requires long proof generation time. Scroll, Polygon zkEVM and Linea are notably working towards this direction. Type 2.5 only modify the gas costs, makes faster proof generation by compromising a few incompatibility. Kakarot zkEVM is known working toward type 2.5 zkEVM, Kakarot is not a blockchain itself, it is a set of CairoVM that support EVM opcodes and expected to be deployed as an enshrined EVM in Starknet L2 and L3, and eventually Type 1 zkEVM in future roadmap. Type 3 is the category where majority of rollups today be, for example Scroll, Polygon, Linea and so on. They are able to be compatible to most of the Ethereum dapps with minor changes that are involved the application layer. Type 3 zkEVM is able to generate fast proof generation. Type 4 creates a new VM in other higher level language that makes ZK development easy with minimum cost and fastest proof generation time. The new VM reduces EVM compatibility and higher difficulty implementing future Ethereum upgrades. ZkSync and Starknet are the leaders of Type 4 zkEVM.

Polygon has been the pioneer of Ethereum Layer 2 since 2017, although part of the communities argue Polygon Proof of Stake (PoS) is technically not a rollup and the bridge between Ethereum post centralisation risk. However, they continue to innovate and develop from multiple angles including Polygon PoS, Polygon zkEVM, Polygon Miden (STARK-proofs non-EVM) and Supernets.

Polygon announced Polygon 2.0, which intends to convert Polygon PoS to Polygon Validium, which is similar to Polygon zkEVM, but instead of posting the data availability to Ethereum, Validium only posts ZK proof to Ethereum and data availability is secured by PoS own set of validators. This design reduces the gas cost and retains self-sovereignty.

Polygon 2.0 also includes migrating MATIC token to POL token with 1:1 ratio, enables multi chain validation, similar to shared security model by Cosmos. It gives significant potential accrued value to the token holders from staking demand from Polygon Validium, gas fees, shared security reward to governance voting power.

Polygon CDK launched in August 2023, it could be the close competitor to SuperChain which also helps developers to launch a ZK Layer 2 on Ethereum while with the capability to access to Polygon Supernets ecosystem. Polygon CDK is designed to be highly customisable, including cost, throughput, compliance, customVM, even Layer 1 or Validium is able to adapt to CDK and access to Polygon aggregate layer. Immutable X, Gnosis Pay, Wirex, Astar, Canto and more are adopted to Polygon Supernets.

ZK generally is viewed to be superior to Optimistic scaling, although Optimistic, in fact, provides higher throughput in scaling with lower cost. However, ZK could outperform Optimistic with improving space-efficiency. Even Arbitrum co-founder Edward Felten stated Arbitrum will switch to ZK if the time is right in Singapore's IOSG panel discussion 2023.

 

There are several catalyst events Arbitrum token holders should pay attention to, first, Stylus, enabling other programming language in writing smart contract with much efficient computation and memory. Second, Orbit also known as Arbitrum “Layer 3”, launching of custom chain that settles to Arbitrum One or Nova. Third, development of permissionless fraud proof BOLD. Fourth, the Short Term Incentive Program (STIP) from September to November to incentivise Arbitrum dapps users. Fifth, active ARB Staking Proposal discussion in the governance forum. Sixth, collaboration with Espresso for the development of Timeboost ordering and shared sequencing.

50M ARB intended for STIP, forecast with matching protocol reward assumption, ARB TVL should increase at least 4.7%, and based on the last OP airdrop event pattern ratio, transaction is expected to increase on average 50% during the airdrop event, 4% to 5% of TVL growth, combined with historical retention rate of 23%, $920K could potentially be sticky TVL, 0.55% growth in the long run.

It is worth mentioning, Arbitrum is one of the rare rollups with backup solutions in the events of sequencer or proposer failure. Users can force transactions to be included in Arbitrum by sending them to Ethereum, or replace a current whitelisted proposer after 6day 8hour 43minute 36second of inactive status.

 

The SuperChain by Optimism seeks to integrate otherwise siloed L2s into a single interoperable and composable system, launching an L2 is as straightforward as deploying a smart contract to Ethereum. Instead of deploying on Cosmos and Polkadot, SuperChain enjoys features like complete EVM compatible environment, easy deployment using OP Stack, high scalability optimistic rollup, inherit Ethereum security, shared bridging and sequencing among other SuperChains. OP Stack is designed to be highly versatile with Bedrock architecture, separating the Consensus and Execution client by using Ethereum’s Engine API. This allows Bedrock to seamlessly implement gas optimisations like EIP-4844 proto-danksharding, Cannon Optimistic fraud proof, proof agnostic design that able to switch the consensus client to other proof such as ZK or multiple client implementation and more. OP and ARB tokens are both governance tokens and lack of direct value accrued to their token holders, however, this does not mean they are not profitable in any sense. Arbitrum is notably leading the Layer 2 in 2023 in term of asset bridged in, Arbitrum once disclosed in May 2023, had accumulated 3,352ETH in revenue and 5,954ETH in refund from sequencer, demonstrated the protocol profitability. Today, Arbitrum has an average daily profit of 24.32ETH after deducting cost for posting and a profit margin of 28% compares to Optimism 13.50ETH average daily profit and 23% profit margin. After the Bedrock upgrade, Optimism increased its average daily profit to 15.75ETH since July and 30% profit margin since July, while Arbitrum had an average daily profit of 19.17ETH and profit margin of 32%.

Besides OP Mainnet, Optimism has another revenue source from The Optimism Collective, where SuperChains will contribute to the Collective which will provide retroactive incentives for public goods which benefit the Collective economy. Base has committed either 2.5% of Base's total revenue from sequencing or, if it's more, 15% of the profit Base makes from L2 transactions after subtracting the costs of submitting data to L1 to the Collective. In return, it earns up to 2.75% of the OP token to become an active member in the Law of Chain onchain governance. Along with the adoption of SuperChain, other participants might make a similar contribution and reviving FAT thesis. Only this time is Optimism instead of any L1.

In regards to token supply, circulating supply ratios of OP and ARB tokens are approximately 19% and 13% according to CoinGecko, while future unlock inflation of 42% and 44% relatively. The majority of these unlock are team and investors allocations.

Messari did a great retention analysis for Optimism and Arbitrum in July, showing that Optimism had a higher retention rate from their airdrop events.

Crypto protocols face the similar challenge across the industry, which users are not sticky and always flow into protocols that offer higher incentivised activities or new innovation and capital efficient mechanism. Protocols need to constantly innovate and provide a sustainable incentive (earning > emission distributed) to keep the users. Optimism showed a relatively stabilised retention between 10% to 20%, while Arbitrum had a falling users retention and expected to drop further until any major catalyst materialised. Arbitrum has 56% transactions related to LayerZero integrated protocols, which potentially apprehends the current level of user retention.

Being the top 2 rollup in Ethereum, Arbitrum performed tremendously in the first half of the year with its airdrop event, while Optimism is catching up in the second half of the year with its SuperChains development and adoption.

However Arbitrum appears to be relatively “cheaper” in term of market capitalisation to the multiple metrics including total value locked (TVL), protocol fees, active users, transactions and treasury fund. These multiples are treated as a reference as metrics like TVL could be double counted or boosted by native token valuation instead of actual inflow of stablecoin, daily active users could be fabricated by single individual, transactions could be washed, treasury funds from Optimism and Arbitrum consist more than 99% of their own tokens.

 

Base was launched by Coinbase with OP stack in August 2023 and immediately hit the market by storm. TVL grew from $61million on the day launched to $201million at the end of August. Average daily profit of 36ETH after deducting L1 posting, which is significantly higher than Arbitrum and Optimism, because of its insanely high average profit margin of 57%.

Majority activities on Base started off with meme tokens such as BALD (surged 30,000x before crashing), Friendtech hype (has an average 19% fee contribution to Base Network) and new dapps such as Aerodrome (accounts for 35.5% of Base TVL) incentivise users to farm on their platform.

Base will be facing challenges from Linea, Mantle and other new Layer 2s that provide incentivised mining schemes or potential airdrop. Besides to retain the existing users, Base needs to continuously innovate, potentially materialise the 10 key focus that Coinbase founder Brian Armstrong mentioned and provides real value to the chain.

 

Appchain first generalised by Polkadot and Cosmos, featuring high customisable infrastructure, enhanced scalability and interoperability, giving the team higher control over the protocol. This idea is proved to be correct as protocols with higher budget often target to deploy their own chain as the end goal, these protocols including MakerDAO, dYdX and Frax Finance. Layer 2 protocols are ready to capture the business as well, starting with SuperChain by Optimism, SuperNet by Polygon, Orbit by Arbitrum, Hyperchain from zkSync Era, Starknet L3s and so on. Among all the competition, SuperChain has been a successful start by on-boarding BASE, Worldcoin, Zora, Aevo, opBNB and other exciting upcoming Layer 2s such as Celo, Manta, Farcaster, UniDex, Mode, Lyra and so on. The advantages that SuperChain over other Layer 2s include being one of the very first layer 2 appchain stacks launched, maintaining good institutional relationships, continuous improvement of technological development and, mostly importantly, OP stack allows the built chain deployed as an Optimistic Layer 2 instead of a fractal scaling Layer 3 which settles on the native Layer 2 chain, allowing less dependency on the native Layer 2 chain. It would be interesting to see if the Layer 3 model will prove to be superior to Layer 2 or protocols will switch the model.

Despite the clear advantages such as EVM compatible chain, highly customisable and inherit security of Ethereum, some protocols would still insist choosing Layer 1 appchain in Polkadot Parachain, Cosmos or Avalanche’s Subnet. These Layer 1 appchains offer a higher degree of customisation, sovereignty, forkability in unexpected catastrophic events and maximum level of Maximal Extractable Value (MEV) capture. Layer 2 is not forkable without forking the underlying Layer 1, therefore Alex, co-founder of zkSync proposed “L1 Fork as the Court of Final Appeal” to attempt to address the issue and Rune, co-founder of MakerDAO proposed a Solana codebase NewChain for the backend of MakerDAO.

 

The emergence of modular blockchains represents a significant shift in blockchain architecture compared to traditional monolithic blockchains. In monolithic blockchains, functions such as execution, settlement, consensus, and data availability are tightly integrated into a single blockchain. In contrast, modular blockchains separate these functions across different network participants, allowing for greater flexibility and scalability.

Here are some notable protocols in the categories of modular blockchains:

Data Availability: Eigen DA by Eigen Layer, Celestia, Avail, zkPorter, StarkEx

Execution: Eclipse, Dymension, Saga, LayerN, Fuel

Shared Sequencing: Espresso, Astria.

Shared sequencing solutions, like Espresso, have the potential to introduce new revenue sources or tokenomics models for blockchain protocols.

Proving: =nil;, Risc Zero

Other Resources-as-a-Service (RaaS): AltLayer, Gelato, Caldera