Author: @Shaughnessy119
Translation: Huo Huo/Bai Hua Blockchain
1. Explore the competing views and contradictions between Ethereum Rollups, Cosmos Application Chain, and Solana
If application-specific Rollups are too expensive and do not provide the necessary technical customization, most developers will choose to release on a shared Rollup. This got me thinking about the perspectives between Ethereum Rollups and Cosmos’s AppChains and Solana.
My mental model for competing ecosystems is simple — the easiest and largest deployment platforms will have the most viral applications. Why? Because this community is most receptive to experimentation, which provides the highest chance of viral applications. I’ve held this view since 2019.
Having to deploy your own chain or Rollup will significantly slow down experimentation because you need to deal with both the infrastructure and the application.
That being said, for some developers, custom modifications are necessary to run high-performance applications. These applications cannot run on the shared L1. For example:
1) dYdX: requires each validator to run an in-memory order book to achieve high throughput and full decentralization.
2) Thorchain: No need to rely on any L1, and trustless cross-chain exchanges can be performed.
3) ChainFlip: An independent network is required to store assets through TSS solutions and other methods, and has other functions.
So how do you resolve the conflict between the competing view that the largest ecosystem (Ethereum) will attract the most developers/applications and the inability to build critical use cases within it?
I don't solve this problem.
Most developers/applications will be released on Ethereum/Solana, while specific cases that require custom modifications will be released on the app chain. My current view is:
1) Ethereum: Most developer/application experimentation, hands down.
2) Solana: The best choice in terms of the fastest shared global state.
3) Cosmos: Launching application-specific chains requires custom modifications, which are not supported on shared L1 or Rollup.
2. Some existing problems
Problem 1
Making the problem even more difficult is the emergence of zkEVM and ZkVM on Ethereum.
Zero-knowledge proofs are the ultimate scalability/privacy technology.
If they work, it would reduce the need to launch custom chains (speed, cost, privacy issues solved).
Problem 2
Both zero-knowledge proofs and optimistic techniques are in their early stages and are not fully decentralized.
Both have centralized serializers to order transactions.
Projects like dYdX need to be fully decentralized now. You can’t let a government shut down a node and stop violating operations ($1 trillion in historical transactions!).
Problem 3
Can Ethereum support Rollups at scale?
Currently, 83MB of data can be written on Ethereum, and 1.3MB on DankSharding.
If Ethereum is congested, Rollup data can be published to CelestiaDA and EigenDA.
Limited throughput could lead to high fees, but that seems unlikely here because the DA (data availability layer) is over-provisioned.
Problem 4
Is the argument for the Cosmos application chain only feasible on Ethereum?
With Celestia, teams can launch their own Rollups and use a cheap data availability layer while customizing their execution environment.
Ethereum is an anchor point, and users can dispute the proofs when something goes wrong.
Problem 5
Ethereum does not support unique/complex forms of experimentation.
Developers don’t like the limitations and inefficiencies of the Ethereum Virtual Machine (EVM). This also applies to zkEVM.
Arbitrum Stylus and RiscZero allow coding in more programming languages, which is an opposing view.
Problem 6
Timing is of the essence.
If the rollout of ZK L3 (or ETH’s data availability layer) takes too long, the next bull run of application development will happen elsewhere (e.g. Cosmos, Solana, etc.).
Timing plays a big role in determining where a project can realistically be released today.
3. What do you think?
If you can do these things on Ethereum using ZK technology or using CelestiaDA/EigenDA, then does the Cosmos argument come true?
Currently, Cosmos’ thesis is clearly being fulfilled through specific large-scale applications.
We will need to re-evaluate once Celestia/Eigen and ZK technology actually launches. It is not certain yet.
Some points are:
1) Ethereum will become the de facto global settlement layer, surpassing Bitcoin in the process and attracting most of the competition’s funding (not including Solana, Cosmos, or any strong contenders).
This isn’t a radical view, as it’s already happening.
2) For Ethereum, zero-knowledge proofs and optimistic Rollup (with a new data availability layer) make it less necessary to launch application chains in the longer term (currently not mature enough). A project can customize its launch on ZK Rollup with high throughput and low fees, which can solve most needs.
3) The narrative war will move upstream. The focus will be on where applications are released and which community is the best: ZkEVMs vs ZkVMs vs Optimistic Rollups. Eventually, the technology will mature and it will become a competition for commercial development.
Over time, there will likely be 1-2 ZkEVMs and 1-2 ZkVMs that win out.
4) There are few new applications built on Ethereum L1, mainly implemented through Rollup.
5) Solana will continue to be the fastest shared global state machine at scale. Despite facing adversity related to SBF, the community has become stronger and upgrades are underway (Firedancer, Jito, and ZK!).
6) Cosmos will certainly serve specific application chains, and in the years before Zk technology is fully established, these applications will likely create large communities that can communicate seamlessly. dYdX has reached $1 trillion in trading volume, and Thorchains' trustless cross-chain swaps have created their own world.
7) From what I understand, IBC on Cosmos is extremely underrated. As for trustless ZK L3 bridging, the development in this area is far from mature.
It’s a bit ironic that Cosmos has solved the bridge problem, while ZK L3 is solving the throughput and latency problems. They need each other’s success to achieve greater breakthroughs.