Analysis | Comparing Ethereum Layer 2 Solutions zkSync and Starkware from 6 Perspectives

Source: DeFi Dao

Recently, @dareal_sisyphe published a comparative article on Ethereum Layer 2 solutions zkSync and Starkware, analyzing them from six perspectives: team, technology, data availability, funding and supporters, current products and roadmap, and personal impressions. Below is a compilation of the entire content.

Saying that ZK rollups will be huge is an understatement.

Two beasts have been researching this technology in the background for several years. Their moment to shine is approaching. So, what opportunities exist in this field? What are the differences between these two projects? When will they issue tokens?

Here is a comparison of zkSync and StarkWare.

A disclaimer: this is a vast topic, and I am doing my best to grasp it. I may take some shortcuts or make some mistakes in my explanations. Feel free to correct me or provide more precise suggestions in the comments.

There are already plenty of resources detailing what ZK Rollup is and how it differs from Optimistic Rollup.

Here, I will focus only on the basics behind zk rollup and the differences between the two main players in the zk game—zkSync and StarkWare.

Basics: Ethereum is what we call a "monolithic" chain. This is because consensus, execution, and data availability all happen on the ETH blockchain. This is not sustainable and is the reason it cannot scale. This means all nodes must perform these three functions. To scale, Ethereum must become "modular." This means using the blockchain only for what it does best: consensus, and shifting the burden of execution and data availability (DA) off-chain. But how do we move execution and DA in a way that fully inherits L1 security?

zk rollups: With zk rollup, thousands of transactions are batched into a single, mathematically verifiable zk proof, and only this proof is stored on-chain.

Next, let’s compare zkSync and @StarkWare. We will strive to keep it clear and follow the perspectives below for comparison.

• Team and Origins
• Technology
• Data Availability
• Funding and Supporters
• Current Products and Roadmap
• Personal Impressions and Opportunities

Team and Origins

StarkWare: Founded in May 2018. The team consists of world-class cryptographers and scientists. The core members are former chief scientists of Zcash, who have been pioneers in the zk field for years. They are turning academic theories into real-world products through Starknet. They are undoubtedly the most renowned team in this game.

zkSync: Founded in December 2019. In fact, there is a "second best" in the zk field, which is the Matter Labs team building zkSync. The team was co-founded by @gluk64, and I couldn't find more information about the smart minds behind it, but the technological breakthroughs brought by zkSync 2.0 speak for themselves. They have an outsider vibe and move quickly.

Technology

Both projects have a similar architecture, with a rollup smart contract inserted into the Eth blockchain to store zk proofs of L2 state transitions. Additionally, they both have two types of off-chain actors responsible for powering the network.

• Provers: A small number of nodes responsible for heavy lifting. They compute all the transactions and aggregate them into concise zk proofs. They run on dedicated hardware and can be black boxes. What we know is that they mathematically cannot forge false ZK proofs.
• Verifiers: A large number of nodes responsible for censorship resistance. They verify the validity of the proofs submitted by the provers. No specific hardware is required. Anyone can run them.

Moreover, both projects must compete to overcome a major technical challenge: creating a universal ZK proof system.

This is a holy grail because until now, all ZK lines require specific applications. This means each application has only one ZK rollup, with no EVM compatibility.

Guess what, both zkSync and StarkWare have achieved this, but technically in different ways.

StarkWare: The cryptographic technology they use is based on STARKs. They are pioneers in this technology and have two main advantages over SNARKs (used by zkSync):

• "T" stands for "transparent," meaning the system works without a trusted setup. Their production speed is 10 times faster than SNARKs. However, the problem with STARKs is that the technology is not as mature as SNARKs, and if it allows Turing completeness, it becomes difficult to make it EVM compatible.
• Additionally, Starkware created Cairo, a specific programming language that allows running discretionary programs powered by STARKs. However, since few people are willing to learn a new programming language for smart contracts, they are currently collaborating with @nethermindeth to create Warp, a transpiler that can seamlessly convert Solidity smart contracts into Cairo to make @StarkWare EVM compatible.

‌ zkSync: The zk system used is based on SNARKs (especially PLONK). Therefore, the security of the entire system relies on a trusted ceremony launched in the fall of 2019. It involves many well-known cryptocurrency players, including Vitalik Buterin. If at least one participant is honest, then this system can prove to be secure, so I feel comfortable and do not think the term "trusted setup" can be used against zkSync.

Although the zkSync team has made many optimizations, SNARKs are still slower than STARKs. However, their system allows for EVM compatibility in a more natural way than STARKs. In fact, smart contracts can (almost) be converted opcode by opcode by the zkSync compiler, making Solidity a "first-class citizen" of zkSync without the need for an intermediary language or transpiler.

Data Availability (DA)

zk rollups shift the burden of transaction computation off L1, allowing ETH to scale to 2-3K TPS. This is good, but not enough. With a "rollup only" approach, the DA burden still lies on L1, while transaction data is written to L1 via calldata.

DA is super important. It is the key that allows you to see what happened with transactions on Etherscan. Without this, the execution of transactions becomes a black box. If you prefer cheap transactions over directly tracking on L1, that might be acceptable. Therefore, zk rollup must offer users a choice: either your transaction data is written on L1, but at a higher cost, or you prioritize ultra-cheap transactions and trust some entity on L2 to ensure the availability of your transaction data. This way, L1 can be freed from the DA burden, and ZK rollup can scale to 20-30k TPS.

StarkWare: Addresses the DA challenge with the Volition system. Volition allows end-users to choose between rollup settlement on-chain (L1 DA) and effective settlement at the individual transaction level (off-chain DA). In the case of effective settlement, off-chain DA is guaranteed by a centralized "Data Availability Committee" (DAC) composed of well-known cryptocurrency entities.

This sounds centralized, but it is a trade-off for end-users who voluntarily choose ultra-low-cost transactions.

zkSync: The solution to the DA challenge is zkPorter, a sharding infrastructure that runs seamlessly with zkSync's zk rollup. Imagine zkSync rollup as shard0 that guarantees on-chain DA. Other shards can then choose different DA policies and store transaction data off-chain.

The main difference between StarkWare's Volition and zkSync's zkPorter is that for Volition, the choice of settlement is transaction-based (users choose DA for each transaction), while for zkPorter, it is account-based (a zkPorter account can only produce transactions with off-chain DA).

Additionally, for zkPorter, the off-chain DA system is more decentralized, as DA is guaranteed by a "guardian" network incentivized by zkSync's native token rather than a centralized "DAC."

Funding and Supporters

StarkWare:

• Completed a $6 million seed round in May 2018 (Pantera/Naval/Vitalik).
• Completed a $30 million Series A round in October 2018 (Paradigm/Sequoia/Cb Ventures).
• Completed a $75 million Series B round last March (Paradigm/3AC/Alameda).
• Completed a $50 million Series C round last November (Paradigm/3AC/Alameda).

Currently, StarkWare is valued at $2 billion. This is world-class funding with notable investors. Big names and the Ethereum Foundation are involved. Vitalik himself has commented on most of the articles produced by Starkware. Shu and Sam are also on the same boat. How could such a project fail?

zkSync:

• Completed a $6 million Series A round last March (Binance/Cb Ventures/AAVE/Balancer/Curve).
• Completed a $50 million Series B round last November (Horowitz/Placeholder/Crypto.com, etc.).

There are fewer big names on the board, which looks like a large DeFi/CEX cryptocurrency family fund. This sounds amusing, but it is actually important because the success of ZK rollup will largely depend on the integration of DeFi protocols and direct CEXs. So I am very optimistic about the rapid integration of the zkSync ecosystem.

Current Products and Roadmap

StarkWare has undergone incredible evolution as they persistently turn their top cryptographic papers into real-world innovations. Their roadmap is summarized as follows:

They first launched StarkEx, which corresponds to the "planetary" stage and allows the creation of permissioned, application-specific zk rollups powered by Cairo and STARKs. If you are unclear, think of @dydxprotocol, @Immutable, or @deversifi. These are three major applications powered by the production version of StarkEx.

So far, StarkEx has processed over 5 million transactions and $250 billion in value through these applications. The efficiency of StarkEx has been proven, and StarkWare will soon enter the "constellation" stage of its roadmap. On November 29, they launched the Alpha version of StarkNet on the mainnet. StarkNet is the permissionless, multi-application, general-purpose zk rollup we have been looking forward to.

Initially, StarkNet will be driven by a centralized validator, and applications will be deployed in order through a whitelist, similar to Optimism. Thus, their plan is to grow the ecosystem and gradually decentralize StarkNet in the "universe" stage of the roadmap.

zkSync's roadmap is summarized in four stages:

The first stage corresponds to the launch of zkSync 1.0 in June 2020, roughly corresponding to zk rollup but without smart contract integration, allowing for sending and receiving tokens. Despite the lack of composability, promising projects have already been deployed on it, and all metrics show exponential interest in zkSync 1.0.

The second stage will kick off with the mainnet deployment of zkSync 2.0, which corresponds to everything we are looking forward to: full EVM compatibility and composability of smart contracts on zk rollup.

zkSync 2.0 was initially planned for mainnet deployment in August, but technical issues delayed the release. These difficulties have now been resolved on the testnet, and last October, zkSync announced it was nearing completion of the technical steps and deployed an AMM-like testnet to demonstrate EVM compatibility.

The delay measures taken by Matter Labs to ensure LLVM/Solidity compatibility may have been frustrating at first, but it will help zkSync gain a lot of time as it allows all ETH tools and dependencies to integrate natively with zkSync 2.0.

Personal Impressions and Opportunities

StarkWare is truly impressive. This is the most promising infrastructure project I have seen in a while. An all-star team, world-class innovation, top-tier funding. Notably, they are not just building a ZK Rollup. They are thinking of everything. We talked about Cairo and Warp. But they are also working on bringing true randomness to Eth with Veedo, facilitating L1 to L2 communication through conditional transactions, Batch-Long flash loans, and other genius ideas. Their approach to solving the liquidity fragmentation between L1 and L2 with "distributed AMMs" is indeed encouraging. They are looking globally.

The only technical issue I see with StarkWare right now is the Solidity integration via the Warp transpiler. StarkNet is primarily optimized for Cairo, while the Solidity forwarding is an addition on top of it. I am not a strong technical person, but I worry that "Solidity translation" may not always be straightforward in every case, and it could cause compatibility issues for some smart contracts.

Additionally, another issue I see is: where are the opportunities for us commoners? StarkWare is already valued at $2 billion, but there are currently no plans for a fair token release.

On the other hand, I also appreciate zkSync's community-first approach and the emphasis on "robustness as a first-class citizen." Their entire technical effort relies on a true EVM experience without any trade-offs, which could pay off in terms of adoption. zkSync is also deeply integrated with the current DeFi/CEX landscape, which bodes well for future protocol adoption and integration.

In terms of opportunities, zkSync has announced the existence of a native token, and they will conduct a fair token release and potential airdrop at some point in the future. This is super good for the community, but I feel that the native tokens of ZK rollups will be overhyped, as everyone and their teams are waiting for it.

I also hope StarkWare launches a native token in the "universe" stage. Because they will have to decentralize their network at some point and incentivize the provers.

In summary, my view is that in terms of zk rollup, the most impressive execution comes from StarkWare, but the grassroots community/open-source atmosphere makes zkSync more appealing to me.

Here are my final personal notes:

In terms of opportunities, perhaps we should shift our focus from the native tokens of rollups to emerging projects that will find user bases on these rollups. ETH is a paradigm shift, and projects that tried to replicate orders on-chain were quickly replaced by AMM models that were better suited for this new environment. Similarly, ZK rollup is a paradigm shift for Eth, and perhaps (just perhaps), replicating AMMs on rollups is a foolish idea; maybe TVL will not be an important metric on rollups; maybe CLOBs are worth reconsidering.

These are all temporary examples, but our thought is that perhaps the best opportunities lie in projects that leverage these rollups to do things that are impossible on L1, rather than replicating what already exists there.

If you made it this far, then congratulations.