Dialogue Celestia: Modularization, Ethereum, and the Future of Scalability in the Crypto World

Interview: Jack, BlockBeats, Vision, MetaStone

整理：Sharon，BlockBeats

Currently, the blockchain faces a triple dilemma: scalability, security, and a lack of decentralization—specifically, the absence of trustless cross-chain communication, insufficient Rollup scalability when transaction volumes grow large enough, and the inability to maintain high levels of security and decentralization while aiming to improve throughput. This has been a long-standing issue, and the essence of the problem lies in finding a way to securely store data in a smaller, lighter container without using excessively large or expensive storage devices.

Most blockchains today are monolithic, where the core functions of the blockchain, including execution and consensus, occur simultaneously and are executed by the same group of validators. This monolithic architecture is difficult to scale because every transaction must be executed by full nodes, leading to bottlenecks. In contrast, modular blockchains completely outsource at least one of the four components (consensus, data availability, execution, settlement) to external chains.

Celestia is the first modular blockchain network and a cloud computing network for Web 3. It is a pluggable consensus and data availability layer that allows anyone to quickly deploy decentralized blockchains without the overhead of bootstrapping a new consensus network. There is a view in the industry that Celestia is the most significant underlying innovation in the blockchain space since Ethereum. Both Ethereum and Celestia are building secure foundational layers. In this exclusive interview at the TOKEN 2049 conference, BlockBeats spoke with Nick White, co-founder and COO of Celestia, to explore the relationship between Celestia and Ethereum, as well as the story behind Celestia.

Without Celestia, Ethereum Cannot Scale Rollup

Among these three dilemmas, the impact of the lack of scalability is the greatest—only by enhancing the scalability of blockchains can we open the door to on-chain access for hundreds of millions of people. This is also the biggest dilemma faced by mainstream blockchains today, including Ethereum. Currently, Ethereum has scalability solutions like Optimism, ZKsync, and Starknet. However, these scalability solutions heavily rely on Ethereum itself for data availability. At the same time, Ethereum's gas fees remain high.

Previously, Ethereum founder Vitalik described what he envisioned as the ultimate form of the Ethereum blockchain, devoting significant space to depicting a new Ethereum built on Rollup and DA. This undoubtedly points to a modular solution as a way to break through Ethereum's challenges over the next decade.

Related reading: “Modularization: The Breakthrough Path for Ethereum in the Next Decade”

BlockBeats: Can you introduce yourself and your background?

Nick: Sure. I am Nick White, the COO of Celestia Labs. We are building Celestia, the first modular blockchain network, which means Celestia is a proponent of a new paradigm for building blockchains. We no longer try to accomplish all functions within a single protocol; instead, we break the protocol into different layers, each focusing on specific functionalities, and then we can recombine these layers to build blockchains and applications.

Thus, Celestia focuses on the consensus and data availability layers of the stack, without performing any execution. Execution is achieved through Rollup (one of the Layer 2 solutions). People can deploy Rollup on top of Celestia, and Celestia provides a scalable decentralized block space for people to build on. So you can think of us as the first layer designed specifically around Rollup to scale Rollup.

BlockBeats: When did you first start considering adopting a modular blockchain?

Nick: It all started with two white papers that emerged in 2018 and 2019. The first white paper was co-authored by Mustafa Albasan, one of Celestia's co-founders, and Vitalik, titled “Data Availability Sampling and Fraud Proofs.” In this paper, he demonstrated that it was possible to build a blockchain that could scale its locked space as the number of nodes in the network increased, thus addressing the scalability issue.

Then, building on previous work, he wrote another white paper titled “Lazy Ledger.” “Lazy Ledger” is a continuation and expansion of the concept of data availability scaling, where he proposed a new idea—a blockchain that is only responsible for data availability and does not execute any transactions. At that time, he referred to it as "client smart contracts."

Related reading: “Understanding Celestia and Fuel, Modular Blockchains to Watch in 2023”

The blockchain client would operate independently of the first layer executing transactions, which is now the Rollup. Rollup is essentially off-chain execution of smart contracts and applications. Therefore, “Lazy Ledger” indeed introduced the concept of modular blockchains. Subsequently, when Rollup emerged, he further demonstrated how the entire system could operate, as Rollup could make the execution layer scalable just like data availability sampling.

MetaStone: Will the launch of Ethereum's project sharding reduce Layer 2 costs, and has it impacted Celestia?

Nick: ETH sharding, that is, Ethereum, has actually shifted its roadmap to mimic the approach that Celestia is building. Before that, they were working on ETH 2.0, which is sharding technology, but at the end of 2020, they decided to pivot and follow the building approach of Celestia. Over time, they have gradually aligned their architecture more closely with Celestia's model. So, Danksharding is essentially a different implementation of similar ideas to those of Celestia.

However, there are several differences, the first being timing. Celestia will launch in a few months, while Danksharding is still in the design and research phase, and it's hard to know when it will launch. I think they haven't even set a date yet, but they do have Proto-Danksharding, which is EIP-4844, but that will only provide a one-time slight increase in Ethereum's block space.

Based on the demand we see for deploying Layer 2, I think this is far from sufficient to provide the needed throughput. Therefore, Celestia will launch when people want to deploy Rollup on a large scale. I believe that without Celestia, Ethereum cannot scale Rollup. In the long run, when Danksharding launches, the issue is that it resembles an additional data availability layer attached to a single first layer, which is the original Ethereum chain.

Thus, Ethereum has a lot of technical debt and burdens that need to be developed upon, while Celestia has the opportunity to start from scratch, so there won't be as much state bloat. We don't need execution; our network is very lightweight and simplified, whereas Ethereum does not have that luxury; they still need to carry the burden of Ethereum Layer 1. These are some of the differences I see.

DAS is More Trustworthy than DAC

Ensuring that users securely own their data and the assets represented by that data, alleviating ordinary users' concerns about asset security, is crucial for guiding the next billion users into Web3. Therefore, an independent data availability layer will be an indispensable part of Web3. Data availability (DA) essentially allows light nodes to verify data without participating in consensus, without needing to store all data, and without needing to maintain the entire network's state in real-time.

Currently, there are two mainstream methods for verifying data: DAS (Data Availability Sampling) and DAC (Data Availability Committee). The former verifies whether a block has been published by downloading some randomly selected blocks, while the latter confirms it has received data by signing each state update with its quorum.

The industry generally believes that when an independent data availability layer is a public chain, it is superior to a committee of individuals with subjective awareness. Because if enough committee members' private keys are compromised, making off-chain data availability unavailable, the security of users' funds and data will be greatly threatened. Nick points out that what Celestia is currently doing is making the data availability layer more decentralized—essentially providing an independent DA public chain with a series of validating nodes, block producers, and consensus mechanisms to enhance security levels.

MetaStone: In the current DA market, all DA layers mainly accept data from Layer 2 and Layer 3, but we know that most Layer 3s cannot send their data to the DA layer due to data staking. However, Polygon will use bridging to receive this data. I wonder what your thoughts are on this, and what methods Celestia will use to receive data from Layer 3?

Nick: Its role is to ensure that the bridge verifies the availability of data on Celestia. Therefore, third parties can publish their data to Celestia but publish their state updates to another chain, such as Ethereum Layer 1, Optimism, Polygon, etc. The aggregation contracts on those chains can verify the availability of data on Celestia through that bridge. Thus, we can help scale this.

MetaStone: In the current DA market, EigenLabs has also launched an EigenDA. At the same time, EigenLabs borrows Ethereum's original distributed nodes to secure other networks and reduce node operations. What are your thoughts on this?

Nick: Re-staking is an interesting idea that allows you to use existing funds, such as collateral, to stake a new protocol. However, it does not fundamentally scale the blockchain; it is merely a way to allow you not to issue new tokens when launching a new protocol. Regarding EigenDA, the problem is that their design is not actually about data availability. The data availability here refers to the kind of concept you think of when you think of Ethereum, Danksharding, or Celestia. Because EigenDA is just a data availability committee, which is a multi-signature account, where someone tells you the data is available, but you cannot verify it yourself. Therefore, EigenDA cannot actually be compared to Celestia; they are not entirely the same product.

Another issue is that if they use re-staked ETH or any non-EigenDA tokens to secure EigenDA, they cannot penalize data withholding attacks. A data withholding attack is an unaccountable failure, meaning you cannot prove to a smart contract on Ethereum Layer 1 or any other entity that data has been withheld. Therefore, if someone actually performs a data withholding attack, they will not be penalized for the re-staked ETH. This way, you can effectively attack EigenDA at zero cost. So, I think this is a deep design issue. That’s my view on EigenDA.

MetaStone: In the process of verifying data, some off-chain data availability layers choose to use DAC to protect their data, while others choose to use DAS. What are your thoughts on DAC and DAS?

Nick: Blockchains are essentially verifiable computers. Therefore, you do not need to trust others, such as a committee. Because the purpose of decentralization is achieved by allowing end-users to verify the chain. Thus, a data availability committee is not actually a blockchain because, by definition, you must trust a committee when using DAC. In contrast, data availability sampling is a method of directly verifying the chain through sampling. Therefore, from a verifiability perspective, it is a true blockchain. You do not need to trust Celestia's validators; you can verify it yourself. Even if they try to deceive you or collude, they cannot fool you. This is a fundamental difference and very important; people should be aware of this. This is also why I said earlier that EigenDA, being a DAC, is not the same as Celestia; you cannot really compare them.

BlockBeats: Does DAS also have more benefits when nodes are added or removed from the network?

Nick: Absolutely. One superpower of a network like Celestia is that due to the adoption of data availability sampling, you can increase the block size as the number of nodes in the network grows, which is very powerful. Because in a monolithic chain, no matter how many people run nodes, you can only use the same block size. In Celestia, as more nodes join and sampling begins, you can actually increase the block size.

We want to create a culture where users run nodes on their wallets or browsers. This means that as more users join the network, the number of nodes increases, so blocks can become larger, providing more block space for new users and new applications. Therefore, there is a positive feedback loop where users actually provide the scaling for their applications.

The Future May Use KZG Commitments

The Quantum Gravity Bridge (QGB) is a data availability bridge between Celestia and Ethereum, deployed by Celestia on Ethereum, allowing operators of Ethereum Layer 2 to publish their transmitted data to the Celestia network, which is then included in blocks by Celestia's proof-of-stake (PoS) validators. This data is then forwarded from Celestia to Ethereum in the form of data availability proofs. The proof is the Merkle root of L2 data signed by Celestia validators, proving that the data is available on Celestia.

The QGB contract verifies the signatures on the DA proofs from Celestia. Therefore, when Layer 2 contracts on Ethereum update their state, they do not rely on the transmitted data published to Ethereum but instead check whether the correct data is provided on Celestia by querying the DA bridge contract. This contract will respond affirmatively to any valid proof previously forwarded to it; otherwise, it will return a negative response. Nick points out that Celestia will provide high-throughput data availability for Ethereum Layer 2, with security levels higher than other off-chain data availability solutions, while being cheaper.

BlockBeats: Do you think the Quantum Gravity Bridge is more expensive or cheaper compared to EigenDA?

Nick: One issue with EigenDA is that they have not released any information about how they are actually building it. So, without code, it’s hard to know what it will look like. I think, depending on how they build it, EigenDA could incur expensive proof costs because you have to generate KZG commitments (Kate-Zaverucha-Goldberg polynomial commitment scheme) and verify signatures on Ethereum, meaning you have to verify a bunch of signatures for each batch. So this could actually consume a lot of gas. The benefit of QGB is that we designed it specifically to minimize gas costs.

First, we have batching. Just like having multiple Celestia blocks, they are all batched into one block, then a commitment is generated, signed, and published to Ethereum. So, you do not need to pass and verify each block; you only need to perform one operation in a batch, which significantly reduces the gas cost of verifying commitments.

Second, we are also building a zero-knowledge QGB, which will verify all these signatures through zero-knowledge proofs, further reducing the gas cost of verifying commitments on Ethereum Layer 1. Because the gas cost of verifying commitments on Ethereum Layer 1 is a significant expense for any off-chain DA. Then there are the actual DA costs of paying for data on Celestia and EigenDA, which are difficult to estimate right now. I think the costs will be very, very low, regardless of the situation, so low that I doubt it will become a differentiating factor unless Celestia suddenly becomes congested or other circumstances lead to very high costs.

BlockBeats: You just mentioned KZG, but why hasn't Celestia adopted KZG yet, and what are the considerations behind it?

Nick: Yes, the issue with KZG commitments is that they are still quite new and computationally slow. Therefore, using KZG commitments would make block creation more expensive. Moreover, as the block size increases, you have to compute more and more opening values, which slows things down. Therefore, Celestia made a very practical decision to use regular Merkle trees (hash trees) combined with fraud proofs.

But the problem is that if it becomes practically feasible, we can easily replace it with KZG commitments. Excitingly, a few weeks ago at the SBC (Blockchain Science Conference), Ethereum Foundation researcher Dankrad Feist shared some promising research on KZG hardware acceleration, and we are keeping an eye on that. If there are any changes and improvements, we will definitely consider making the switch. However, KZG would add a lot of complexity, so that is a challenge.

BlockBeats: I want to ask some questions about Rollkit (a modular rollup framework). What role do you think Rollkit will play in the future?

Nick: First, people should know that Celestia is completely neutral. In fact, we are currently collaborating with almost every rollup SDK to integrate Celestia as a DA option. We started Rollkit before there were open-source rollup frameworks because at that time there were Layer 2s, but they were all trying to build their own single thing rather than trying to build software SDKs that anyone could use to build their own rollup, which is why we incubated Rollkit.

I think one of the unique aspects of Rollkit is that it is the first designed without being tied to Ethereum and without involving the concept of settlement with smart contracts. Therefore, it is more suitable for running Sovereign rollups. Another important aspect is that Sovereign Rollkit is compatible with ABCI (Application BlockChain Interface), so any Cosmos SDK application or execution environment that is ABCI-compatible can work with it.

People have already used many different virtual machines and made them compatible with ABCI, and then they can launch them on Rollkit. This is very important because it opens up another ecosystem for building rollups. Another great thing is that the Rollkit team has built a fraud proof system for Cosmos SDK applications. Therefore, it is indeed possible to build an optimistic rollup on Rollkit, which is very exciting.

BlockBeats: Do you have anything to say to developers or practitioners in China?

Nick: We are very excited to hope for more presence in China, and we know that from very early on, China has played such an important role in the origins of blockchain and cryptocurrency. China has so many talented engineers and users, and the Chinese community is full of enthusiasm. So, we are very much looking forward to engaging and participating with them. I lived in Hong Kong for a year and a half and have traveled extensively throughout China. I love Chinese culture, and I really appreciate the mindset of the Chinese people; they are full of desire, with a builder's mindset and a fighter's mentality, which I really admire.