The Ethereum Rollups war has come to an end, and the DA track is filled with smoke

Author: Zuo Ye, Crooked Neck Mountain

Everything is modularized, Ethereum is self-modularized, and Bitcoin is modularized.

The narrative halts after the issuance of Rollups, and the narrative economics shifts to the DA layer/blockchain.

Orthodoxy and universality become the banners, while in reality, transaction fees and token issuance are the keys.

Marked by the StarkNet airdrop, the competition of Ethereum Rollups comes to an end, and it's time to talk about DA. In my view, the term Data Availability (DA) is an incomplete expression, lacking a clear subject and predicate, merely describing the importance of transaction data transmission beyond the execution layer. Furthermore, the DA mechanism involves the fundamental operational principles of blockchain, which I have elaborated on using Bitcoin as an example in my runes article.

Ethereum's Narrative Weakens, DA Takes Over Midway

Modularization is the premise of DA. Ethereum's horizontal modularization is sharding, while vertical modularization is layering. Rollups handle transactions, while the mainnet is responsible for DA and consensus. The popularity of DA means that the concept of layering has become consensus. Additionally, the Rollup battle has concluded, and what follows are mere fixes and patches.

The mainnet's upgrade plans have become daily updates and annual posts, providing limited confidence to the overall market. In this context, the narrative rhythm cannot unfold from the top-level Rollups and the bottom-level mainnet; thus, DA, which can link the two, becomes the best choice.

Let’s first complete the expression of DA: Data Availability, in a narrow sense, refers to how light nodes, such as wallets, efficiently verify full node data. There are two premises here.

Premise One: Light nodes do not download or cannot download complete full node data, especially when prioritizing user experience.

Premise Two: Full node data may be subject to falsification, with no access control mechanisms; whether PoS or PoW, malicious nodes may exist.

DA stems from actual demand

On single-chain systems like Bitcoin, this is not an issue because block headers already contain rich verifiable information, and the PoW mechanism ensures that a 51% attack is only theoretically possible. However, when it comes to modular chains, the problem becomes complex. Transaction execution, settlement, consensus, and DA may not exist on different layers, and they may not even be on the same blockchain.

It is important to note that, according to Vitalik, data availability ≠ data retrieval ≠ data storage, but is equivalent to data being published without tampering. As for storage and retrieval after publication, they do not become the focus of DA. The distinction lies in:

Data Publication: On Ethereum, light nodes can directly prove transaction validity without possessing all data.

Data Recovery: For Ethereum, using Ethereum for DA does not raise security concerns, so the term publication can encompass this. However, Celestia and others must prove that data existing with them is equivalent to being stored on Ethereum, hence there will be retrieval or recovery mechanisms.

From Vitalik's perspective, once data is published to the Ethereum mainnet, the entire process is complete, and later storage and retrieval do not require excessive concern. This makes sense; as the second most significant entity after Bitcoin, Ethereum's security does not need technical terminology to prove it.

However! There are always exceptions. If transaction data and consensus data do not circulate entirely within the Ethereum system, then the publication, retrieval, and even recovery of data need to be carefully considered. This is also a key point that Celestia and Near DA need to prove.

DA Relativity: Everything Can Be Modularized

Modularization is the direct driving force behind the rise of the DA narrative. Ethereum actively chooses to transform itself into a modular public chain, currently in a transitional hybrid architecture state. Bitcoin can be used as a modular layer; early practices like OmniLayer and current BTC L2s carry this implication.

The concept of modularization here is my own definition, where outsourcing or being outsourced of single-chain functions is considered a form of modularization, not equivalent to Ethereum's discourse system.

Any public chain can be modularized

Or understood this way: previous blockchains also had light nodes, partial nodes, and user verification of full nodes, but it was not a large-scale market demand. Only on modular chains does the separation of layers lead to significant issues in state synchronization, data storage, publication, and recovery. After all, no one wants to see a second rollback after The DAO incident.

First, let's understand modularization. The earliest practice should be the Lightning Network, which is another proof that modularization, like DePIN, is "practice preceding theory." Outsourcing part of the blockchain's functions or modules can be understood as a delayed settlement accounting system.

For example, the earliest issuance of USDT was on Bitcoin's OmniLayer, which ultimately published data on Bitcoin. This also indicates that UTXO model blockchains can be modularized.

Account model blockchains, like Ethereum's modularization, are simpler. Near DA and Celestia follow the same thinking. Since everything can be decoupled, the Ethereum mainnet does not possess the extreme sanctity of Bitcoin. Therefore, whether treating Bitcoin as a data publication object or "helping" Ethereum process data can both exist reasonably.

Without modularization, the concept of DA would still gain popularity, but it would not receive as much attention.

The End of Ethereum Rollup Wars, BTC L2 is Just Beginning

With modularization, there are champions. Before the concept of DA, the Rollup route won the expansion war, even showing trends of spreading to BTC L2. From a more radical perspective, modularization is the ultimate solution for expansion. Regardless of the demand for security, scalability, or decentralization, they can all be separated from the mainnet, built independently, and then connected to the mainnet.

However, this also raises an interesting question. In Bitcoin, where large-scale expansion solutions have not been implemented, various BTC L2 projects are flourishing, such as B² Network, which uses fraud proofs to relay data back to the Bitcoin mainnet, representing a way of treating it as a DA layer. Additionally, Alt L1s are entering the DA market with greater force. Why should Ethereum's DA monopolize the market? The orthodoxy must be overthrown and trampled upon by a thousand feet. Near DA says so and is prepared to do just that.

In a sense, Ethereum is an improvement over Bitcoin, with four major differences: PoW → PoS, UTXO → account model, single-chain → modularization, script → smart contracts. The interaction point between the two is precisely modularization, which is the converging evolution of expansion routes. The difference is that Bitcoin is passively modularized, with more and more L2s treating Bitcoin as a DA and settlement or consensus layer.

Thus, it must be acknowledged that "the modularized Ethereum created the market demand for Rollups regarding DA, leading to the heat of the DA layer." The implied premise here is that Rollups are no longer the main characters, at least not on Ethereum.

We can make a slight distinction, at least dividing it into Ethereum-related DA solutions, such as Ethereum, EigenLayer, Celestia, and Near DA, and Bitcoin's use of BTC as a de facto DA through the Lightning Network, OmniLayer, and B² Network.

The distinction here lies in that, for Ethereum, both Ethereum itself and EigenLayer's solutions still center around ETH and the Ethereum network, ultimately empowering ETH. This is rooted in the economic design of Rollups, which need to pay "tolls" to the mainnet to utilize the security provided by the ETH PoS network. The tolls here mainly refer to DA costs, which are the costs of publishing Rollup transaction data to Ethereum for final processing.

DA Economics

For Bitcoin, it is much simpler. Bitcoin has no smart contracts and no node review; you can write anything into transaction data as long as you pay the miner's fee. However, it must be noted that once written, there is no going back; data cannot be rolled back or any nodes slashed. BTC L2 must solve transaction conflict issues independently.

Ideologies on the Lips, Business in the Heart

Vitalik initiated a debate on the definitions and classifications of L2 and Rollup, distinguishing between Rollups, Validium, and Sovereign Rollups, with the main dimension being the choice of DA solutions. In the centuries following the end of the Middle Ages, one could still see familiar operations like "excluding your church membership."

Visa's Summary of Rollup Differences

We only need to remember that the issue of data availability is not purely a technical solution and definition dispute; the core lies in the cost of ETH in the PoS era, which is a matter of real money. The technical dispute is merely superficial, so I will provide a simple introduction.

In a narrow sense, data availability is "how light clients verify full node data," which can be deduced along the following logic, based on papers by Vitalik and the founder of Celestia:

Full nodes may be subject to falsification, meaning the data provided may have issues;

Among full nodes, at least one node is an honest node, preserving complete or truthful data;

Light nodes must have the ability to "distinguish truth from falsehood," able to timely correct falsified data. For example, multiple light nodes can cross-verify different data, which is a sampling mechanism.

The core here is the proof mechanism. Taking Celestia as an example, fraud proofs are central to DA operations, using fraud proofs to timely correct errors. Moreover, validating fraud proofs is faster than generating fraud proofs, allowing light clients to complete verification quickly without affecting user experience.

We will delve into fraud proofs. For fraud proofs, one only needs to remember that it is very close to the optimistic verification process of OP Rollups, which assumes truth first and then addresses the problematic ones.

The reasoning logic of fraud proofs:

At least one honest node exists among full nodes;

Broadcast mechanisms can operate normally, with delays below the network's effectiveness limit;

A certain number of light nodes exist, capable of combining to recover complete data or equivalent data proofs;

Under this logic, we can conclude: the security and effectiveness of light nodes are equivalent to full nodes.

Since there is OP, there is naturally a ZK route imitation. In fact, both Ethereum and EigenLayer follow the "validity proof" route, generating validity proofs in advance for distribution, but the generation itself requires substantial computational resource consumption.

To summarize, Celestia and Near DA consist of off-chain + fraud proofs (OP-like) + low cost + native token DA solutions, while Ethereum and EigenLayer consist of on-chain + validity proofs (ZK-like) + higher cost + ETH DA solution lineup.

Comparison of DA Solutions

It should be noted that completely developing DA solutions based on EigenLayer may not be as expensive as directly using Ethereum, and EigenLayer may not necessarily refrain from issuing tokens. However, the central position of ETH will not change.

Secondly, the costs of each DA solution, based on Near's calculations at the end of last year, cannot represent real-time and fixed prices. Moreover, Ethereum's ongoing upgrades will also speed up and reduce costs, but the overall comparative landscape will not change.

From the perspective of Rollups' interests, open-source and cost-cutting are two paths to profit. Transaction fees and token issuance are their sources of profit, which must not be relinquished. The only way to enhance profitability is through cost-cutting. If they continue to use Ethereum, security is sufficient, but the costs are too high. This is where opportunities for Celestia arise.

EigenLayer centers around ETH, while Celestia centers around TIA. From Vitalik's perspective, this is akin to a vampire attack, utilizing Ethereum's existing ecosystem while ultimately empowering its own token.

Orthodoxy and Universality, Discussing Bitcoin and Ethereum

I personally believe that a fragmented Ethereum lacks orthodoxy, but the on-chain DA layer still possesses the highest security, which is true for both Bitcoin and Ethereum. Orthodoxy can also be understood as the adaptability to Ethereum and the degree of dependency of expansion solutions on the Bitcoin mainnet.

In terms of universality, it is also necessary to deeply consider the design ideas of each DA. Some DA solutions are essentially specialized L2s or L1s. Even BTC L2s and L1 EVM chains like Near, as well as EigenLayer, regard EVM compatibility as an important development direction, thus equating EVM compatibility with compatibility.

Celestia's situation is somewhat special, as it introduces off-chain computation mechanisms, theoretically allowing compatibility with any virtual machine (VM), including EVM. Moreover, Celestia is actively expanding its ecosystem, with cross-chain dApp calls also in its plans.

Of course, the modularization and DA ideas of Bitcoin and Ethereum are not consistent, and doing so is merely for enjoyment.

Comparison of DA Solutions

Bitcoin's DA Transformation

To be precise, Bitcoin is forcibly treated as a DA layer. Whether it is inscriptions, runes, or BTC L2s, they all emphasize the importance of storing data on Bitcoin.

At this level, the Lightning Network and B² Network represent two extremes. The Lightning Network relies entirely on the Bitcoin mainnet for settlement and does not issue its own tokens. In daily operations, it requires BTC for staking. However, as I introduced in my BTC L2 article, the Lightning Network is merely a payment channel, lacking the capacity to support smart contracts, making it a historically significant product with high orthodoxy but poor EVM compatibility/universality.

In contrast, the DA orthodoxy of ETH, EIP-4844 ETH, and EigenLayer is similar, with the only difference being that the latter three inherently possess smart contract functionality. This indirectly proves that the ETH-centric approach is not only an economic consideration but also a responsibility for the long-term development of the ecosystem. Once ETH's value capture ability is lost, the entire EVM ecosystem faces the risk of collapse.

In comparison, OmniLayer has made more progress by using the Bitcoin mainnet as a data publication solution. Although it still requires nodes to download complete data and lacks an efficient proof mechanism, it does not support complex operations. This is also the main reason why USDT abandoned OmniLayer in favor of RGB. It is hard to call it a DA solution for Bitcoin, but considering this is already an "ancient" product, it is merely for comparison and does not impose strict requirements on the old-timer.

As a side note, RGB++ and CKB are exploring new ways to build BTC L2s. I will write a systematic update on the new developments in BTC L2s at an appropriate time, leaving a pit to fill later.

Next, using B² Network as an example, I will illustrate how the "new era" BTC L2s treat Bitcoin as a DA layer. Unlike the unconscious use of the Lightning Network and OmniLayer, B² Network plans to combine the data relay of the Rollup layer with a fraud proof mechanism, with an overall approach highly similar to Celestia.

B² Network Technical Architecture

In design, B² Network partially separates Bitcoin's DA role. The Bitcoin mainnet primarily serves as a settlement layer, while the data storage of the B² Network DA layer requires additional incentive mechanisms from B² nodes to cover the costs of decentralized storage.

The EVM compatibility of B² Network does not require excessive scrutiny, but it is highly likely to issue its own tokens. Additionally, how to reduce the interaction costs with the Bitcoin mainnet must also be considered, as the usage costs of Bitcoin are very high.

Overall, Bitcoin's DA transformation is still in its infancy. The inscriptions, runes, and BTC L2s on it need to be practically utilized on a large scale to generate real demand. However, it will not deviate from the path of Ethereum's practices; there will only be differences in implementation paths, considering the dual constraints of scripting languages and storage costs.

Ethereum DA: Encircling Celestia

DA is now well-known and is closely related to Celestia. It was Vitalik who, in 2018, co-authored a paper titled "Fraud and Data Availability Proofs: Maximising Light Client Security and Scaling Blockchains with Dishonest Majorities" with Celestia's founder Mustafa, which elucidated the operational mechanisms and implementation principles of DA.

Celestia's fraud proof mechanism, light clients, and the minimized number of honest full nodes are all reflected therein. Subsequently, Mustafa built the predecessor of Celestia under the name LazyLedger.

However, it was unexpected that after Celestia was truly launched in the market, it would face resistance from Vitalik, with economic disputes being the core reason, as analyzed earlier, and I will not elaborate further.

Celestia naturally lacks significant orthodoxy, being an external DA layer to Ethereum. Rollups choosing Celestia as their DA layer have also been stripped of their titles. However, under the law of cheap attraction, more and more various projects are flocking to it.

The operational mechanism of Celestia is not complex. The core lies in light nodes efficiently verifying full node data through the DAS (Data Availability Sampling) mechanism.

Celestia's low cost arises from transferring computation off-chain, which not only allows the DA layer to operate at high speed but also effectively accommodates any programming language and VM (virtual machine). The friendliness of developing dApps is also a clever trick for rapid ecosystem development.

Currently, various Rollup solutions, RaaS, Rollup development frameworks, settlement layers, cross-chain bridges, and wallets can all be developed through Celestia in a one-stop manner.

Celestia Ecosystem

In the face of foreign attacks, Ethereum emphasizes that it can also serve as a DA layer and that costs will continue to decrease with ongoing upgrades. However, constrained by its existing architecture, engaging in a price war with Celestia and Near is clearly unwise, and EigenLayer is thus pushed to the front lines of resistance.

Unlike Celestia, EigenLayer is essentially a collection of smart contracts on Ethereum. From this perspective, EigenLayer is Ethereum itself but can also be viewed as an abstract virtual chain. This duality allows it to maintain the central role of ETH while also extending its functions in different dimensions, such as DA, ordering, cross-chain bridges, and L2 bridges, which can all be built using EigenLayer. Eigen DA is one such example.

In simple terms, EigenLayer's so-called liquid restaking is a nested version of Lido. If ETH can earn yields while being staked and can be exchanged for stETH for use as a token, then stETH can also continue to be nested, with the tokens generated from further staking serving as both yield certificates and complete tokens for daily use.

After Ethereum transitioned to a PoS mechanism, the amount of ETH staked directly relates to the network's health and security. Currently, about 30 million ETH are staked in the network, valued at around $10 billion, with attack costs second only to Bitcoin.

Since staking ensures Ethereum's security, LSD/LRT theoretically can be infinitely nested, continuously amplifying the yield of staked tokens. Based on a foundational price of $10 billion, amplifying it tenfold would yield $1 trillion, which Ethereum's value can support.

The architecture of Eigen DA is not crucial; the sustainability of EigenLayer's economic model is key. Even if EigenLayer fails, using the Ethereum mainnet poses no issues.

Due to space limitations, I will not delve deeply into EigenLayer/ETH/EIP-4844 ETH, Near DA, and Avail. Just remember that they all deal with providing validity proofs without full node data.

Conclusion: DA is a Long-Term Competition

The Ethereum DA market will continue to compete for a while.

The Ethereum DA market has already begun, with Celestia having issued its token TIA first. Although EigenLayer centers around ETH, it is rare for projects not to issue tokens these days; we will see how it plays out.

While new DA solutions will emerge, the DA business on Ethereum has essentially completed its land grab and will not see much novelty.

Bitcoin's DA transformation is still in incremental competition, waiting for BTC L2s to determine their fate.

In my judgment, the likelihood of Bitcoin being treated as a DA layer similar to Ethereum is not very high. The lack of smart contracts is secondary; the main issue is the excessively high costs. Even with data compression of hundreds or thousands of times, it remains too expensive. Ultimately, Ethereum is not suitable for data storage, let alone Bitcoin.