Inventory of teams that are bearish on Ethereum sharding and Layer2 solutions

Flow's former CryptoKitties crypto cat team, now it's out to build the Flow public chain

Flow was born almost at the same time as CryptoKitties due to the network congestion we caused. If a few thousand users can paralyze the network of our application, then we cannot ignore this platform risk.

Within a week or two of launching CryptoKitties, we knew we needed to find a solution to our scalability and usability issues. We did an in-depth study of dozens of blockchains, met with the top teams in the field, and came to the conclusion that we need our own platform.

Our main concern is that breaking up the network execution state (via sharding, second-layer solutions, or sidechains) would greatly increase the burden on developers. Furthermore, we recognize that a number of critical improvements are needed at the protocol layer, including resource-oriented programming, or human-readable security.

Comments: Flow is a typical case of Dapp applying their own public chain. When they saw the performance bottleneck of the Ethereum platform, they resolutely decided to build a new public chain, Flow. Its investment institutions are also top configurations: A16Z, USV, etc.

This passage does not look at the sharding scheme. It is just more objective to say that sharding is a very complicated technology, which will greatly increase the burden on developers. I believe that at least in the early days of the Flow blockchain, the team will not use the sharding scheme.

Professor Gun, founder of AVA Labs

I am not optimistic about Ethereum's sharding technology. It breaks one of the foundations of the Ethereum model by randomly sharding the network in a way that disregards contracts. It's like randomly dividing China's population into different provinces to realize communication among people in the same province, but if you and your neighbors are divided into two different provinces (shards), communication cannot be easily achieved. I don't think this division makes sense. I don't think sharding is the future, I think faster consensus protocols are the future. With a fast consensus protocol like Avalanche, there is no need for sharding.

I wish the best of luck for Ethereum 2.0, they are having enormous difficulty meeting the promised timeline. They have promised that Ethereum 2.0 will go online this year, but I think they will take a long time to fully deploy the protocol, because the protocol of Ethereum 2.0 is very complicated, and there are some problems that no one can solve in sharding, so I wish them good luck. But if you want a faster network, AVA is your choice.

Comments: I agree with Professor Gun on sharding. In short, sharding is complex and would break Ethereum's composability. For example, the reason why the current DeFi ecosystem on Ethereum is so prosperous is because of the mutual use and combination of various financial protocols, resulting in new financial products. This is often referred to as the composability of Ethereum. The sharding scheme will make the composability of Ethereum very uncertain, and even destroy this composability. At this point, it is an opportunity for other Ethereum killers: seize the gap when sharding destroys the composability of Ethereum, and seize their own market share.

But I don't agree with his nonsense that "with a fast consensus protocol like Avalanche, there is no need for sharding". This is a private product, and I am not optimistic about the Avalanche project.

Kyle Samani, Managing Partner, Multicoin Capital

For example, we invested in Nervos because they are really focused on Chinese user adoption and Layer 2 scaling. At the same time, we invested in Near because they are the leading team in building the sharding process. If I had to pick just one, it would be Solana. Solana is transitioning from a testnet to a mainnet, a project that really focuses on Layer 1 scaling through pure engineering. We see that Solana has an absolute lead in three aspects: 1) High throughput; 2) Lowest latency; 3) Lowest transaction cost.

The so-called "Layer 1 cannot scale" is very popular. When people saw that Solana could actually support 50,000 transactions per second in real world parameters, they thought there must be tricks in there. But counterintuitively, Solana doesn't actually try to deploy any new tricks. The fancy new gimmicks are things like Layer 2 and sharding, which are completely unproven. Whereas Solana is focused on working as we know it, on Layer 1, just on pure engineering. I consider the Solana protocol and implementation to be an engineering marvel.

Comments: Although Kyle is skeptical about sharding and Layer2, he still invested in Near and Nervos. Although you can have your own preferences and preferences for different technical solutions, invest in top teams and projects in different solutions/different fields.

This is basically the way institutions invest. This is in stark contrast to retail investors. Retail investors often judge projects in black and white. They will not invest a penny in projects that are not optimistic, and invest money in projects that are optimistic. But the lines between things are often not so sharp. To better grasp investment opportunities, we should make a variety of layouts for uncertain things and leave a little room for things we are not optimistic about.

Having said that, we don't think optimizing for extreme decentralization is a bad thing. This is an extremely complex issue that is important and will eventually be resolved.

For us, we believe that sharding does help decentralize the network, but it doesn't necessarily scale the network. The network is thought to scale as its capacity increases, but the fee for a transfer will be limited by the fee for a single transaction in a single shard, since that fee reflects the shard's resistance to malicious behavior. If the fee is reduced to a certain point and this resistance is lost, the shard can be hacked and reach its maximum performance. Therefore, all applications on this shard will need to go through the lengthy and cumbersome process of migrating to another shard.

Having said that, we don't think optimizing for extreme decentralization is a bad thing. This is an extremely complex issue that is important and will eventually be resolved.

Another challenge is the network complexity introduced by cross-shard communication. In terms of smart contract logic, it also includes how fees are distributed among participants. This is a problem that is not encountered in a single shard blockchain/homogeneous network like Solana.

A major disadvantage of Layer 2 solutions is that they are not fully general, or require significant trade-offs. We think this may limit the number of use cases where it can be applied. Another is whether we can flexibly transfer assets between things like state channels. For now this remains a challenge. While layer 1 solutions like EOS and Ethereum have proven to have a better developer experience, they are limited in performance as applications scale. That's why we think Solana can build on these two solutions and take further steps to address this issue.

Comments: Anatoly believes that sharding can bring decentralization, but it may not necessarily expand the network. More importantly, now that the Solana mainnet is live, its ultra-high performance seems to prove what the team claims: I think we have proved that the network can be expanded without sharding, low fees, high throughput, and low latency. Fragmentation.

Summarize

Summarize

I personally remain neutral on the sharding scheme. Currently, there are only two projects that I am looking forward to in the field of sharding: ETH2.0 and Near Protocol. These two projects are the final bottom line in the field of sharding. If the two fail, then sharding will fail completely.

Fragmentation technology is very complicated and will bring side effects, but the development of technology has always been tortuous. As an investor, you can only give sufficient time and patience. To say a cliché, it is to be friends with time.