This is Whistle’s second article, about why Aptos and Sui don’t work. Interview | Beichen Guest | Steven
Whether it is capital endorsement or technical endorsement, Aptos and Sui are very strong. But they actually have no blocks, no chains, and no future.
This podcast invited technical expert Steven, who interpreted Aptos and Sui as distributed networks from the perspective of a communications engineer. His final conclusion was that their technical logic was untenable.
The following is a condensed text version:
Beichen: Before we start chatting, please introduce yourself first.
Steven: I am a technician who has been working in communications for 20 years. I have a certain understanding of distributed systems, and all blockchain technologies can be analyzed from the perspective of distributed systems. I hope my views will be of some help to everyone.
Beichen: Let me add something. Steven is a technical expert who often speaks out against blockchain. This time, we invited Steven to systematically analyze some public chains for us. This will be a series. Today, we will start with several public chains in the Move language ecosystem.
Steven: My bold statement is that Sui is not the right direction for blockchain. What is strange is that in the past year, no one has pointed out that Sui and the three public chains Aptos and Linera that came out of Meta have problems, but in fact the problems are very obvious.
Beichen: This is also what confuses me. People in the cryptocurrency circle seem to be quite ignorant. For example, after Dfinity appeared, many technical people in the cryptocurrency circle were very excited. They thought its narrative was very grand and could solve many problems, but in fact, that was the story of cloud native, which Microsoft and IBM have been doing for many years.
Steven: That’s right. Most of the technical people in the cryptocurrency circle are software engineers. What they care about is whether new functions can be realized. The Move language runs more efficiently and has capital support, which is enough.
As a communications engineer, I view blockchain as a distributed system. I am concerned about the status of each part of the system and the overall efficiency and security that are ultimately formed, so I can see some big problems.
For example, Solana’s CTO and I used to work in the same company, so as a communications engineer, he created Solana, and as a communications engineer, I can understand the starting point of those technologies at a glance. If a cryptographer were to look at it, he would definitely have a different opinion.
In addition, I think no one criticizes Aptos and Sui for being in the wrong direction because the capital behind them is too strong.
Beichen: So no one dares to say it?
Steven: It’s not that they don’t dare, but ordinary software engineers don’t need to care about this, and communication engineers either have conflicts of interest or don’t care about it at all.
But I think someone in this industry should stand up and point out which projects have promoted technological progress and which projects are purely for profiteering.
The blockchain industry is now in a dilemma - since Ethereum, no new technology paradigm has emerged. This is a dangerous signal for the blockchain, but for such a long time, capital has been busy cutting leeks, and technology has also catered to capital in this regard. Of course, leeks are just following suit.
Satoshi Nakamoto did not create Bitcoin to do these things, but to continue the crypto-punk movement, which is to use technological progress to promote social progress, so as to avoid the erosion of personal freedom by large centralized institutions. But now the circle is the opposite, a small number of people with technology are exploiting and plundering the majority of people with worse technology than them, which is definitely not Satoshi Nakamoto's ideal.
Beichen: Let’s start with the first question. Many people are optimistic about the public chains that use the Move language because they think it solves the pain points of blockchain. So do you think the Move language is better than the solidity language?
Steven: There is no comparison between Move and Solidity because their foundations are different, so there is no question of which one is better.
Solidity is a compiled language created for Ethereum and must be executed in the Ethereum virtual machine. It is a Turing-complete language. Of course, it has many problems, which I will not go into detail here.
Move was created by Facebook when it was working on the Libra project. It is written based on the Rust language. Libra is essentially a standard consortium chain, and all nodes are officially approved nodes. Therefore, the Move language was born for the consortium chain. It inherits the hardware resource management of the Rust language, has good parallel capabilities, and then programs resources at the type level.
Rust is a compiled language, while Move is an interpreted language. It was later inherited by projects such as Aptos and Sui, but the technical logic of these projects is problematic, which determines that it has no future in technology and has nothing to do with the Move language itself.
For public chains, Move is not a good language because if you are separated from projects such as Aptos and Sui, Move will not work.
Beichen: For developers, the Move language is still very attractive.
Steven: What is attractive are Aptos and Sui. If you want to do projects on them, you must use the Move language, but the Move language cannot run on other blockchains.
Beichen: So do products developed on these Move language public chains have more advantages than other high-performance public chains (such as Solana)?
Steven: It has nothing to do with language. All public chains starting from Libra have problems. This is the root of all problems.
Beichen: Let’s compare the technical features of the three Move language public chains, Aptos, Sui and Linera. Which one do you think may be more technically powerful now?
Steven: As public chain projects, they are not real blockchains, or the direction of the public chains they represent is very bad, which will lead the entire blockchain world towards a negative technological trend.
Their entire system is inherited from Libra, and there are some improvements on this basis, but the improvements are not significant. Linera has relatively little public technical information, so I will mainly talk about Aptos and Sui.
Beichen: Before we talk about Aptos, let’s talk about Libra first.
Steven: Libera is an unfinished product. It is a relational database with a version number. It is a distributed transaction settlement system that uses transactions as a carrier, records transactions, and then outputs results and ledger status.
Note that it updates the ledger status for each transaction, which is completely different from the traditional blockchain. The blockchain ledger only adds new fields but does not modify existing fields.
This global state control method is suitable for consortium chains, but not for public chains. Since its high performance is achieved at the expense of decentralization and security, a centralized system with a little cryptography is enough, and there is no need to use blockchain.
Beichen: So you think there is no need for Aptos to inherit Libra?
Steven: Aptos has some good designs in actual use. For example, after the account address is generated, the private key can be changed and the wallet can be restored with multiple signatures, which is the same as the account recovery of traditional banks. However, implementing these functions will bring greater security risks because there are logical problems in the data structure, which reduces the security level of the distributed system.
Beichen: Specifically, what operations did Aptos perform that caused the security level to drop?
Steven: Aptos' accounting method is to use the key value in the entire account as an operational data structure, and then set up a timestamp to arrange the execution order on the blockchain, without the need to form a broad consensus among miners.
This is the practice of the consortium chain, which manages all nodes as a cluster, and its shared memory pool protocol is equivalent to the buffer in the traditional business cluster. So although Aptos has improved the operating efficiency, it is very easy to be attacked because the timestamp is proposed by the leader in the validator, and the block time is easy to estimate.
Although all BFT algorithms have this security risk, other public chains that use the BFT algorithm are still relatively cautious. For example, after Ethereum switched to the POS mechanism, all mining nodes also used the BFT algorithm in the merge phase, but all nodes were at the same level, and random numbers determined which nodes would package all transactions, thus ensuring the security of the entire chain.
Therefore, the basic logic of the data structure for parallel processing claimed by Aptos is incorrect - it does not determine the order of precedence according to the linked list structure and determine the global state in a packaged manner. There is no block or chain at all. It is just a relational database.
This is why Aptos can modify and delete data, because it is not a blockchain and is no different from a traditional relational database.
Aptos is a server cluster of a group of Web2 vendors that provides services to everyone. However, in order to make it look like a blockchain, state control is added to each operation. This will definitely make the efficiency very high, but the number of nodes must be controlled, because if there are too many nodes, the efficiency of state synchronization will decrease.
It can be said that Aptos has removed the decentralization and security in the blockchain's impossible triangle, so efficiency is meaningless.
Beichen: Next, let’s talk about Sui.
Steven: Aptos and Sui have similar problems, which is that not all nodes need to reach consensus.
At least Aptos is trying to move closer to the blockchain based on Libera (although it cannot change the nature of the alliance chain). Sui just laid it out directly - I am just a distributed smart contract execution platform, which has nothing to do with blockchain.
Sui divides transaction attributes into simple transactions and complex transactions. Simple transactions are completed as long as they are signed, while so-called complex transactions only require Byzantine consensus. In other words, as long as any transaction can be submitted in the system, it can theoretically be verified independently, and the final step is just a state synchronization. This is really letting go, no different from traditional clusters, so the efficiency must be very high.
Therefore, Sui did not adopt the usual blockchain approach to data consistency. This is not innovation at all, but a direct abandonment of the most basic consensus of blockchain.
Beichen: So is it necessary to add state control to the distributed clusters of traditional Internet?
Steven: No, it’s OK as long as the data on all servers is consistent.
Beichen: Aptos and Sui have added state control. Does it have any effect on security?
Steven: It has a certain effect, but it is not very strong. In theory, it will definitely be attacked. Hackers can attack as long as they can control a small number of miners. But to be honest, this situation is unlikely to happen, because Aptos and Sui will strictly control the entry threshold and number of nodes. It can only be said that the possibility of doing evil is much greater than that of traditional blockchains.
Beichen: So I can understand it as just a blockchain-like distributed network, not a blockchain?
Steven: Yes, it is a blockchain-like cluster system, but it adds certain state control to each operation. This is inherited from Libra, so which one is stronger in technology, Aptos or Sui? I can only say that Aptos has done more on the basis of Libra, while Sui is closer to the original state of Libra, so it is hard to say which one is stronger.
Although Solana also sacrificed the security and decentralization in the impossible triangle, it at least made some innovations and complied with the basic principles of blockchain such as global control and consensus algorithm. Aptos and Sui are simply traditional relational databases that abandon the basic properties of blockchain.
Beichen: Are all Move-related public chains essentially alliance chains?
Steven: Yes, and Sui doesn’t want to pretend to be a public chain.
Beichen: The use of alliance chain in China is mainly due to compliance considerations. I feel that it is a transitional product that will eventually die out and will die out soon.
In short, different people will pay for different IQ taxes. Aptos and Sui's narratives are very appealing to those in the Web2 world, because they can only understand these things. They may not even agree with the logic of Bitcoin and Ethereum, thinking that they are illusory visions, and the technology is far behind Web2. At this time, when a high-performance public chain appears, they will instinctively feel close to it and think "this thing is good."
Basically, every bull market will attract a new wave of people who have just come from Web2. Their IQ tax is basically locked in the two categories of "high-performance public chain" and "large-scale easy-to-use products", but their life cycle is destined to be short. It is undeniable that although it is an IQ tax, it can indeed attract hot money to the industry.
I came into contact with a capital some time ago. They were planning the Sui ecosystem. The logic was simple and crude, which was that the Sui ecosystem would definitely have a DEX or other basic suite. Then as long as you invest in the top one or two in the first batch, you would definitely get a good return.
How should I put it? It is indeed not wrong, but these so-called "high-performance public chains" have a life cycle. They are not real order outputters. They can only take over the hot money of Web2 and the technology of Web3 to make a quick buck. This is the case with EOS, BSC, Solana, etc. For example, I think Solana is not dead yet, but it is already brain dead, so it can actually be declared dead.
Steven: They don’t understand the essence of blockchain, and haven’t carefully read the Bible of the blockchain world - the Bitcoin white paper, or even the emails before that from Satoshi Nakamoto, which revealed how he thought and how to solve problems that traditional systems cannot solve.
Beichen: So how do you see the future of Aptos and Sui?
Steven: To be honest, this is difficult to predict. The technology doesn’t matter as long as it doesn’t affect making money. I think they are tools for the big capital behind them to reap the profits, so it ultimately depends on how long the capital can hold out (if FTX doesn’t collapse, Solana can hold out for a while longer).
Technically, Aptos and Sui are projects that were developed from the corpse of Libra. They directly use the existing technology as a selling point and then build a project.
Simply put, I have some technology in my hand, so I use this technology as a selling point to design a system, which cannot solve the problem of implementation in any scenario. If you do it without thinking it through, it is definitely not right, so I personally think they have no future.
Active participation in the short term should bring good returns, but it is not the right direction for the development of blockchain technology and may even have the opposite effect.
Beichen: Aptos and Sui are not some conspiracy or scheme of capital, but most of the participants (from project parties to capital to developers, media, and retail investors) really believe that it represents the future of blockchain. It should be said that it is the wishful thinking of a group of people who have just come to the Web3 world from the Web2 world.
Let’s talk about this for today. Can you summarize all the opinions expressed today in one bold statement?
Steven: To sum up, all of the Libra projects use consortium chain technology to build public chains. This is a technological degradation, and the technology itself is untenable.