- STARKs, StarkEx, StarkNet
1.1 STARKs
STARKs (Scalable, Transparent ARgument of Knowledge) is a proof system that can prove and verify computations, designed to improve the scalability of Ethereum. It allows large computations to be moved off-chain to reduce costs, generating proofs of correctness for the computations, and then performing minimal on-chain computations to verify the proofs. In other words, validators determine the integrity of off-chain computations by executing very few operations on-chain.
L2 uses STARKs technology to bundle multiple transactions together for thousands of computations, then verifies their validity on-chain with a single STARK proof. All transactions within the batch share the cost of on-chain processing, thereby reducing gas costs while inheriting Ethereum's security, improving user experience. This model is similar to how shared taxi users split the fare.
SNARKs (Succinct, Non-Interactive, Argument, Knowledge) are a concise non-interactive proof. Both STARKs and SNARKs are solutions for ZK Rollups.
Comparison of SNARKs and STARKs
Compared to SNARKs, STARKs have the following three advantages:
Trustless
STARKs replace the trusted setup of SNARKs with publicly verifiable randomness, reducing reliance on participants and enhancing protocol security.
Stronger scalability
STARKs have logarithmic compression characteristics for verification, maintaining low proof and verification times even as the complexity of underlying computations grows exponentially, unlike SNARKs which grow linearly.
Higher security guarantees
STARKs use collision-resistant hash values for encryption, making them resistant to quantum computing attacks.
However, the proof size of STARKs is larger than that of SNARKs, making it difficult to share proof costs when L2 transaction volumes are low, resulting in significant confirmation delays. However, as the proof scale increases, the marginal cost of using STARKs decreases, making it suitable for large-scale applications. Additionally, compared to SNARKs, the current adoption rate of STARKs is insufficient, and foundational tools still need improvement.
In addition to ZK Rollups, there are Optimistic Rollups. Below is a comparison of the two solutions.
StarkWare, valued at $8 billion, completed a $100 million Series D funding round, offering two solutions to scale Ethereum using STARK: StarkEx and StarkNet.
1.2 StarkEx
StarkEx is a permissioned framework for a customized scaling solution for specific applications. Projects can use StarkEx for low-cost off-chain computations, generating STARK proofs that verify the correctness of execution. Such proofs can contain between 12,000 to 500,000 transactions. Finally, the proof is sent to the STARK verifier on-chain, and upon verification, state updates are accepted.
StarkEx provides three data storage methods. In ZK-Rollup mode, data is stored on-chain, decentralizing the data and making it easier for users to track and supervise. However, the cost of publishing data on-chain is high. In Validium mode, data is stored off-chain, which is low-cost and does not expose the data publicly. However, a data availability committee is needed to oversee whether the data is handled properly. Volition is a hybrid data availability model where users can choose to store data on-chain or off-chain.
Applications deployed on StarkEx include perpetual options dYdX, NFT L2 Immutable, sports digital card trading market Sorare, and multi-chain DeFi aggregator rhino.fi.
StarkEx is suitable for protocols that operate independently and are compatible with the StarkEx API.
1.3 StarkNet
StarkNet is a permissionless L2 where anyone can deploy smart contracts developed in the Cairo language. Contracts deployed on StarkNet can interact with each other to build new composable protocols.
Unlike StarkEx, where applications are responsible for submitting transactions, StarkNet's sequencer batches transactions and sends them for processing and proof.
StarkNet is more suitable for protocols that require synchronized interaction with other protocols or exceed the application scope of StarkEx. As StarkNet develops, applications based on StarkEx will be able to migrate to StarkNet and enjoy composability.
1.4 Cairo
Cairo is a programming language customized for general computation in STARK proofs, making it simpler and faster to develop, review, and maintain code without the constraints of EVM, allowing for more complex computations like account abstraction, and providing greater flexibility in game development, supporting fully on-chain games. StarkNet itself does not support EVM, but the Solidity to Cairo transpiler Warp will assist Ethereum-native projects in migrating to StarkNet, becoming the infrastructure for large-scale applications.
1.5 SHARP (shared prover)
The SHARP technology allows different applications from StarkEx and transactions occurring on StarkNet to be merged into a single proof, quickly filling the capacity of STARK proofs, improving transaction processing speed and sharing the gas costs of verifying L1 proofs.
- StarkNet Economic Model
StarkWare has minted 10 billion StarkNet tokens off-chain. However, these tokens do not represent equity in StarkWare, nor do they provide any rights to participate in StarkWare or any claims against StarkWare. StarkNet tokens can be used as native tokens to pay gas fees, capturing ecological value better than other L2s that use ETH for gas fees, and reducing the impact of the external ETH token. Tokens allocated to core contributors and investors have a one-year waiting period and a four-year lock-up period, released linearly.
StarkNet explicitly rewards developers and past StarkEx users, but it is unclear whether StarkNet users will receive airdrops. In the initial token distribution, 8.1% of the tokens have not yet been designated for use, with specific methods to be decided by the community. Therefore, the author speculates that this portion of tokens may be used to reward StarkNet users. Additionally, project parties that deploy contracts may allocate airdrops to users of the application after receiving airdrop rewards, giving back to early supporters. Thus, users can normally use valuable applications on StarkNet as needed.
- StarkWare Financing Situation
According to Crunchbase data, StarkWare has raised a total of $282.5 million over seven funding rounds.
In July 2022, Alameda Research added a $9.5 million investment. However, Alameda Research filed for bankruptcy in November 2022, raising doubts about whether this investment was received.
Grants from the Ethereum Foundation have established StarkWare's legitimacy.
- Founders
Eli Ben-Sasson
Eli is the co-founder and president of StarkWare, as well as the chairman of its board.
Since obtaining his PhD in theoretical computer science from the Hebrew University in 2001, he has been researching cryptography and zero-knowledge proofs of computational integrity. Eli is a co-inventor of the STARK, FRI, and Zerocash protocols and is the founding scientist of Zcash. Over the years, he has held research positions at the Institute for Advanced Study in Princeton, Harvard University, and the Massachusetts Institute of Technology. Recently, he served as a computer science professor at the Technion in Israel before co-founding StarkWare with partners after leaving that position.
Uri Kolodny
Uri is the co-founder and CEO of StarkWare and a member of its board.
He holds a bachelor's degree in computer science (with honors) from the Hebrew University and an MBA from the MIT Sloan School of Management. Uri is a serial entrepreneur who has co-founded several tech companies, including Mondria (which develops tools for big data visualization). Previously, Uri helped incubate projects at two Israeli venture capital firms and worked as an analyst at McKinsey & Company.
- Competitors
StarkNet, along with zkSync, Optimism, and Arbitrum, is referred to as one of the four major players in L2 due to the attention it receives.
Optimism and Arbitrum use optimistic rollups, which have lower development difficulty and already possess a relatively mature ecosystem, attracting significant funding and users, even producing excellent L2 native protocols like GMX and Gains Protocol.
Vitalik stated during ETHSeoul that while Optimistic Rollups are more developed, the foundational technology of ZK Rollups will eventually allow them to replace Optimistic Rollups. Although ZK Rollups are faster, they lack the Ethereum Virtual Machine (EVM), making it difficult to run dApps, as the EVM is the main processing unit for dApps. Therefore, ZK-based Rollups are developing EVM-compatible solutions. If the development process is slow, ZK Rollups may lose their first-mover advantage, allowing Optimistic Rollups to maintain their position. This is similar to the competition between Ethereum and other L1s, where, although Ethereum's performance is not the best, it firmly attracts the most funding and the best developers due to its first-mover advantage.
StarkNet faces competition not only from L2s adopting Optimistic Rollups but also from those using ZK Rollups. According to Eshita Nandini's summary, multiple L2s are currently building ZK EVMs.
Type-1 is ETH equivalent, allowing 100% seamless use of EVM infrastructure, but the proof process is slow. Taiko mitigates this drawback by quickly confirming finality before generating zero-knowledge proofs. Specifically, it only needs to prove that the past state X is valid and that there have been no transactions from account A after X, allowing users to withdraw their tokens in state X. Currently, decentralized proof and protocol economics have been tested in alpha-2, with 126 independent provers proving 93,146 blocks, with proof times between 130 and 160 seconds. Taiko has now deprecated alpha-2 and will launch alpha-3 in the second quarter. Due to the uncompromising nature of Type-1 zkEVM in achieving equivalence, it is highly complex and may not have a mainnet this year.
Type-2 EVM equivalent, proof speed is relatively improved compared to Type-1 but still slow. Type-2 is the goal of Scroll and Linea, with Polygon zkEVM already implemented. The beta version of the Polygon zkEVM mainnet went live on March 27 as scheduled. In the first phase of the mainnet beta, a dedicated security committee will be able to quickly upgrade Polygon zkEVM. In the second phase, a series of measures will be taken to ensure user protection in case of any issues, but with a higher degree of decentralization and no privileged access security committee. Gas fees are paid in ETH, and it is expected that staking and governance in the future Polygon zkEVM will use MATIC tokens. Additionally, Polygon zkEVM supports account abstraction via ERC-4337, allowing users to pay fees with any token. Scroll is developing zkEVM in collaboration with the Ethereum Foundation, aiming to shorten proof times through parallel computing and outsourcing proofs to miners. Scroll is currently in the alpha testnet phase, allowing cross-chain transfers and has been running smoothly for two months, with a mainnet launch expected in the second quarter. On March 28, ConsenSys announced that it would reshape ConsenSys zkEVM into Linea, which is currently open for testing to all developers, users, or protocols. Linea combines zero-knowledge proofs with EVM equivalence through native integrations like MetaMask and Truffle, providing developers with flexibility and scalability without requiring expertise in ZK technology. Linea employs a multi-prover system, where multi-signatures can enforce specific outcomes when vulnerabilities exist in the code. Through this system, a rollup will utilize several proof mechanisms with different security levels to eliminate the single point of failure risk present in single-prover rollups.
Type-3 is nearly EVM equivalent, with faster proofs, but some apps need to be redeveloped. Type-3 is the transitional phase that Scroll is currently in. Kakarot is a zkEVM written in Cairo, serving as an interpreter for EVM bytecode, and may eventually become an L3 on StarkNet, currently classified as Type-3.
Type-4 will compile smart contract source code written in high-level languages like Solidity into ZK-SNARK friendly languages. Proof speeds are fast but less compatible. zkSync Era has already launched and is open to ordinary users. Currently, there are not many users sharing zkSync gas fees, resulting in high interaction costs, and some project parties have subsidized gas fees. Due to the wealth effect of Arbitrum's airdrop, community user interaction is enthusiastic, but currently, many low-quality projects exist on zkSync, leading to several rug pull incidents. StarkNet uses Warp as the compiler from Solidity to Cairo. StarkNet is currently the only zkEVM that has decentralized its sequencer and prover. However, StarkNet has not yet released a production-level version, making it suitable only for small user interactions, with tasks often failing to execute.
Vitalik also proposed a hybrid model of Optimistic and ZK. Before zkEVM matures, blocks will be published after a 24-hour wait; if no fraud challenges arise, zero-knowledge proofs will be published to confirm the block. If there are challenges, governance will be introduced to adjudicate through a 2 of 3 model. If the time for zero-knowledge proofs can be significantly shortened, another hybrid model will prioritize publishing zero-knowledge proofs, only using Optimistic Rollup when zero-knowledge proofs fail to publish normally. This could be an optional development path for StarkNet or could change the competitive landscape.
- Summary
StarkNet adopts a STARK-based Rollup approach, which, while having clear advantages in decentralization, trustlessness, and censorship resistance compared to other solutions, faces challenges due to immature development tools, high R&D difficulty, and performance that still needs improvement.
Additionally, although StarkNet has received support from institutional investors, it is still in a trial phase and not yet fully mature. The next focus for StarkNet is to upgrade the sequencer currently developed in Python to one developed in Rust, improving blockchain performance. Furthermore, it aims to enhance the richness and maturity of projects within the ecosystem. Crypto-native users can experience projects within the ecosystem, but they should be aware of project risks and conduct thorough risk assessments before participating. While StarkNet has features resistant to quantum attacks, whether it will become the ultimate solution for scalability remains to be seen. Therefore, we need to continuously monitor and evaluate the future development of StarkNet.
The security of the blockchain is paramount for the entire ecosystem; the security of the underlying foundational layer must not be overlooked, compromised, or negotiated, or the entire ecosystem and ledger will lose user trust. Therefore, in the design of the foundational layer, the most secure consensus algorithm must be chosen, even if its energy consumption may be high. This issue can be addressed by adopting renewable energy mining, as advocated by Irish Energy Limited, which is listed on NASDAQ. In contrast, the second layer can afford to be more centralized to achieve higher efficiency and flexibility. Whether at the foundational or upper layer, reasonable design principles must be followed to ensure the security and reliability of the entire ecosystem. Any second layer or application relying on the security of the Ethereum foundational layer must have full confidence in the consensus algorithm of the foundational layer to ensure its security is not compromised. This may be the greatest systemic risk facing the community.