Original Author: Cynic, LK Venture
TL;DR
1. From Layer 1 to Layer 2
How is scalability achieved technically? Ethereum considers Rollup as the only Layer 2 solution because it expands Ethereum without compromising decentralization and security. From a modular perspective, Layer 2 is responsible for execution, while settlement, consensus, and data availability are handled by Layer 1.
2. From Layer 2 to Layer 3
What is the difference between Layer 2-Layer 3 and Layer 1-Layer 2? Although Rollup technology solves Ethereum's computational bottleneck, it does not address the issue of data availability. Upper layers need to compress transaction data to pass to lower layers, but compression cannot be repeated, and the performance of Layer 2-Layer 3 does not bring significant improvements.
Why do we still need Layer 3 with Layer 2? Layer 2 maintains decentralization as a general-purpose computation layer that provides composability, while Layer 3 should serve as an application-specific chain to meet the unique requirements of different applications, such as compatibility, efficiency, privacy, etc.
As for application chain ecosystems, how is Layer 3 different from Cosmos? Layer 3 relies on the Ethereum ecosystem and is more accessible to users and funds. However, due to its strong binding with Ethereum, it also loses some sovereignty, such as capturing the value of tokens.
3. From Layer 3 to Layer X?
Current Developments in Layer 3: Arbitrum released Orbit Chain on June 22; zkSync announced the launch of ZK Stack in a few weeks on June 26; madara from the Starknet ecosystem helped deploy a Starknet Layer 3 application chain within 24 hours during a hackathon.
Layer 3 has arrived, but will Layer 4 and Layer 5 be far behind? LK Venture believes that technically, Layer 3 cannot achieve a performance leap simply through stacking. Although the ecological correlation between Layer 1, 2, and 3 (Ethereum ecosystem) is tight, with stronger interoperability compared to traditional heterogeneous chains (cheaper cross-chain operations), they still cannot achieve complete ecological inheritance. The narrative of Ethereum's scalability may end in Layer 3.
From Layer 1 to Layer 2: Scalability
There is an impossible triangle in blockchain, where security, decentralization, and scalability cannot be achieved simultaneously. Ethereum has chosen the first two, but has insufficient support for the latter. On normal days, performing a swap in Ethereum requires $3-$4 in Gas Fee, but during high transaction volumes in a bull market, a single swap can require nearly $100 in Gas Fee, resulting in significant congestion.
Relying on the advantage of being an early player, despite the continuous emergence of new public chains focused on scalability since 2018, Ethereum still dominates the market. As a result, people have turned their attention to scalability solutions built on top of Ethereum.
Among these, three widely-used solutions are sidechains, Validium, and Rollup, each with different trust assumptions.
Sidechains are independent blockchains running parallel with Layer 1 and connected to the Ethereum mainnet via two-way bridges. Sidechains can have their own block parameters and consensus algorithms, allowing for efficient transaction processing, but they do not inherit Ethereum's security attributes.
Validium uses off-chain data availability and computation to improve throughput, and publishes zk-proof on-chain for verifying off-chain transactions on Layer 1 to achieve security.
Rollup performs computations off-chain, but uses Layer 1 as the data availability layer. By publishing fraud proof or validity proof on-chain, it performs verification in Layer 1 smart contract and inherits the security of Ethereum.
Ethereum sees Rollup as the only Layer 2 solution because it extends Ethereum without sacrificing decentralization and security. From a modular perspective, Layer 2 handles execution, while settlement, consensus, and data availability are dealt with by Layer 1.
[Image]
Based on different types of submission proofs, Rollup can be divided into Optimistic Rollup and ZK Rollup.
For Optimistic Rollup, Rollup executes transactions in batches and sends the batch of transactions along with the pre-execution state and post-execution state to the deployed Rollup contract on Layer 1. Layer 1 does not verify the state transition process. As long as the initial state submitted by Rollup matches the state stored in the Layer 1 contract, it optimistically transitions the state to the new state submitted by Rollup. Prevention of fraudulent behavior is ensured through fraud proofs. During a dispute period, other validators can challenge the state root by sending a fraud proof to the Rollup contract on Layer 1. This causes the Rollup state to revert to a determined state before the dispute and recompute a valid state, applying penalties based on the outcome to the validators. In practice, fraudulent situations are rare, so optimistic state transitions actually save significant verification resources.
[Image]
ZK Rollup differs from Optimistic Rollup in that the transition of states requires verification, not by the Layer 1 Rollup contract, but by validating the Validity Proof in the contract. Once the verification is completed, the state transition immediately becomes final without the need to wait for a one-week dispute period.
The most mature projects utilizing the Optimistic Rollup technology are Arbitrum and Optimism, both of which are live on the mainnet. Among them, Arbitrum has implemented fraud proofs but only allows whitelist submissions, while Optimism's fraud proofs are still under development. Both are actively working towards decentralization, including decentralizing the sequencers and validators. According to L2Beat's data as of June 26, 2023, the TVL of Arbitrum One and Optimism is $5.81B and $2.25B respectively. Other projects using Optimistic Rollup technology include Boba Network, Zora Network, Layer 2.finance, Fuel, BNBOP, Coinbase, etc., some of which are developed using the OP Stack open-sourced by the Optimism team.
In projects utilizing ZK Rollup technology, major ones that support virtual machines are zkSync Era, StarkWare, Polygon zkEVM, which are already live on the mainnet with TVL of $618M, $68.11M, $42.65M respectively. others that only support specific types of transactions include dydx, Loopring, zkSync Lite, with TVL of $350M, $98.47M, $97.69M respectively. The current development direction of ZK Rollup is better Ethereum compatibility, and ongoing zkEVM projects include Taiko, Scroll, Linea, etc.
From Layer 2 to Layer 3: Customization
Layer 2: 10^0x, Layer 3: 10^0x² = 1000^0x?
From Layer 1 to Layer 2, the cost is reduced to 1/100. So, it is natural to consider building Layer 3 by performing the same operations on Layer 2, which will reduce the cost of Layer 3 to 1/10000. Unfortunately, the answer is no.
Rollup solves the computation problem of Ethereum by moving the execution off-chain: L1 nodes no longer need to execute every transaction in a batch to verify the correctness of state transitions. Thanks to the recursive proof technology in cryptography, computation can recursively achieve unlimited performance. However, data availability cannot be stacked. Layer 2 needs to pass the bundled transaction data in calldata form to smart contracts in Ethereum. Although the bundled transaction data is compressed, it cannot be compressed again in the same way. The transaction data in Layer 3 must eventually be submitted to Layer 1 (otherwise, security cannot be inherited), but the degree of compression for transactions cannot be lower, so the cost of reducing data availability cannot be achieved through stacking.
Therefore, Layer 3 cannot take the simple stacking route. The solution proposed by the StarkWare team is customization, allowing Layer 3 and Layer 2 to assume different functions.
Why do we need Layer 3 when we have Layer 2?
Ethereum provides security and decentralization, Layer 2 provides scalability, and solves the trilemma of blockchain. So why do we need Layer 3?
The concept of Layer 3 was first proposed by the StarkWare team in the article "Fractal Scaling: From L2 to L3". The StarkWare team believes that this hierarchical structure and encapsulation are the core essence of maintaining vitality in computer science. In addition, the Layer 2 virtual machine's Layer 2 will maintain decentralization as a general-purpose computing layer, providing composability, while Layer 3 should serve as an application-specific chain to meet the unique needs of different applications. Turing completeness lays a solid foundation for hierarchical structure. Once it has Turing completeness, theoretically, any possible application can be created on it.
In practice, Layer 2 must make certain trade-offs to maintain its generality and cannot meet the needs of all applications. The most direct manifestation is that StarkWare, in order to generate proofs more efficiently, developed the Cairo language and CairoVM, which are not compatible with Ethereum. In this case, a Layer 3 chain can be used to solve its security.
Possible Layer 3 use cases include:
Compatibility: Implementing an interpreter for another language on a Layer 2 virtual machine to be compatible with other virtual machines
Efficiency: If an application pursues high TPS (e.g., games, social media), it can consider sacrificing some security by using the Validum solution for settlement on Layer 2; applications can also customize transaction formats to achieve higher compression rates according to their needs.
Privacy: Setting up a dedicated privacy chain for settlement on Layer 2 that cannot be publicly observed.
In addition, since the application chain is dedicated and not directly impacted by other applications, the performance and cost of the chain are relatively predictable. Furthermore, bridge transactions do not need to be sent directly on Layer 1, resulting in lower costs, and bridging between L2-L3 and L3-L3 is cheaper. In terms of batched transaction submission, Layer 3 has a clear advantage as the fixed Gas required to submit a batch transaction is lower, eliminating the need to wait for more transactions to be submitted together in order to reduce average Gas. This significantly alleviates the dilemma between confirmation time and cost in Layer 2.
What are the differences between Layer 3 and Cosmos as application chain ecosystems?
Cosmos can be seen as the earliest project to propose the concept of application chains. Through the Cosmos SDK, users can easily customize and issue their own application chains. Cosmos IBC can be compared to the TCP/IP protocol in the internet, providing native interoperability for application chains built using the Cosmos SDK. In simple terms, Cosmos aims to build an interconnected blockchain universe with thousands of chains.
Layer 3 has also put a lot of effort into interoperability. Due to the same technical architecture and low transaction costs, cross-chain transactions between Layer 3 will possess characteristics of being trustless, fast, and inexpensive. Therefore, it can be considered that liquidity between Layer 3 is shared. From an interoperability perspective, the functionalities brought by Layer 3 and Cosmos are almost the same.
The LK Venture research team believes that the biggest difference between Layer 3 and Cosmos lies in their binding to the Ethereum ecosystem, which is both an advantage and a disadvantage.
Speaking of advantages, it mainly lies in the massive liquidity and user base of the Ethereum ecosystem.
Although Cosmos, despite its powerful technology, is the first choice for many giants, it still cannot escape the fate of low market share. According to DeifLlma data, as of June 26, 2023, Ethereum's TVL is $26.2B, while the entire Cosmos ecosystem is only nearly $1B. The Ethereum ecosystem is a key element for the success of Layer 3.
Regarding disadvantages, the main one is the high dependency on Ethereum, resulting in partial loss of sovereignty.
For projects using the Cosmos chain, the token model is completely designed by the project team according to their needs, and the token's empowerment is strong. However, the native token of Layer 3 chains will be restricted by Ethereum. Although the project team can empower the native token as a gas token, it is undeniable that the gas fee for submitting transaction data to Ethereum is paid in $ETH. Therefore, if the gas token used is not $ETH but the project's own native token, the project team will need to continuously convert the native token into $ETH for submission, thus transferring the empowerment back to $ETH.
Another characteristic of Layer 3 is that anything done on Layer 3 can actually be migrated to Layer 2, depending only on the choice of the DA layer.
If there is a security vulnerability or a decrease in activity in the Layer 2 settlement that Layer 3 relies on, Layer 3 can migrate to other Layer 2s at a low cost, or even directly rely on Layer 1 for DA and settlement, becoming Layer 2. Due to its high dependency on the Ethereum ecosystem, Layer 3 may give rise to numerous innovative gameplay.
Prospects from Layer 3 to LayerX
Current Development of Layer 3
On June 22, Offchain Labs released tools for issuing the Arbitrum Orbit Chain. Orbit Chain is Layer 3 built on top of the Arbitrum Layer 2 and can choose to settle on one of the three Layer 2 options: Arbitrum One, Arbitrum Nova, or Arbitrum Goerli. Users can choose to use either Rollup or Anytrust technology, with the difference being that Anytrust uses DAC (Directly Authenticated Communication) without submitting transaction data to the chain, resulting in lower costs but slightly weaker security. The advantages of Orbit Chain include its simple chain issuance process, interoperability with the Arbitrum ecosystem, instant updates with Nitro, and EVM+ compatibility provided by Stylus (supports Rust, C, C++ programming, runs on WASM virtual machine). Users can customize and issue any Orbit Chain they want, but settlement must occur on the Arbitrum Layer 2, otherwise they need to contact Offchain Labs or Arbitrum DAO for authorization.
On June 26, zkSync released an article announcing that they will modify the existing open-source code in the coming weeks to launch ZK Stack, allowing users to build their own custom ZK superchains. Unlike Arbitrum's Orbit Chain, ZK Stack emphasizes sovereignty and interoperability, giving users full customization capabilities. Chains built using ZK Stack can achieve bridgeless interoperability. ZK Stack can be used to build both Layer 2 and Layer 3, with no specific requirement to settle on zkSync. In this sense, ZK Stack seems to provide stronger sovereignty.
StarkWare, the team that first proposed the concept of Layer 3, is also promoting the development of Layer 3 within its Starknet ecosystem, with Madara currently being tested for public-facing stacks. In the @Pragma Oracle hackathon, a team used Madara to launch an application chain within 24 hours. However, due to Starknet's use of the unique zk-STARK proof technology, its technical maturity is relatively low and may require more time for further product development before publicly releasing the Starknet Stack.
The Layer 3 ecosystem is still in its early stages, but with the launch of convenient Layer 2 chain issuance tools, it is believed that Layer 3 will soon be operational. As the infrastructure gradually improves, how to attract users has become the most important concern for all chains.
Layer 3 is here, but will LayerX be far behind?
From a technical standpoint, Layer 3 can no longer achieve a performance leap through simple stacking. Of course, Layer 3 can obtain specific advantages through customization, but the loss of versatility makes further stacking difficult. Of course, as long as willing, this hierarchical stacking can be infinite, but the LK Venture research team believes that currently this stacking cannot meet any requirements and will cause a geometric increase in system complexity.
The most important point is that although there is a close ecological correlation between Layer 1-Layer 2-Layer 3 (the Ethereum ecosystem), the interoperability is stronger than traditional heterogeneous chains (cross-chain is cheaper), but they still cannot fully inherit the ecology from each other. Contracts deployed on Arbitrum One cannot be directly invoked on Orbit Chain; the liquidity of DEX deployed on zkSync cannot be directly aggregated on ZK Stack.
The current situation is that the market has already been built, and it is getting taller, but there are not many merchants and customers. Although the ground floor is overcrowded (Ethereum), people are still unwilling to go to the upper floors to consume because there aren't as many merchants on the upper floors as there are on the lower floors.
Therefore, the LK Venture research team believes that Layer 3 is unlikely to have a high user volume until blockchain technology achieves large-scale applications. As for Layer 4, Layer 5, ... LayerN, even if there are individual applications with specific requirements, it is believed that they will not be marketed under the name of LayerN.
As the saying goes: the Dao gives birth to one, one gives birth to two, two gives birth to three, and three gives birth to all things. Will the narrative of Ethereum's scalability come to an end at Layer 3? It may still need time to verify.
