Refuting the MOVE public chain as an alliance chain from 12 angles

Source: "Strike Back! A number of senior developers refuted some remarks criticizing Move as a public chain》

Author: Faust

Respondents:

• Wubuku—Founder of dddappp, a low-code Dapp development platform
• Nanne2022—Senior Move Ecological Developer
• Zhou Qi—Founder of EthStorage, an Ethereum ecological storage project
• Li—Senior Move Ecological Developer
• Jolestar — Founder of Rooch, a Layer 2 project that supports Move
• eternal—Move community member

Previous Summary: On June 21, 2023, a media called “Whistle” published an article titled "Reversing Technology? Aptos and Sui are actually alliance chains, and life depends on the patience of capital " article. The main content of the article is about the interview record of the author Beichen with Steven, a communication engineer suspected to be from the chip giant Qualcomm, and made many criticisms of Aptos and Sui. The article stated that Steven is a technical expert who often makes violent comments on the blockchain, and said that Sui is not the correct direction of the blockchain, and Aptos and Sui both use the technology of the alliance chain.

As soon as this article came out, it caused widespread dissemination and controversy in some Web3 communities. Many technical personnel in the industry, including the Ethereum community and the Move community, have criticized this article.

In the spirit of seeking truth from facts, we conducted interviews with a number of technical experts in the industry to discuss the 12 controversial arguments in “Reversing technology? Aptos and Sui are actually alliance chains, and life depends on the patience of capital”.

1. On the title and end of this article criticizing the Move-based public chain, it is repeatedly mentioned that “all Move-based public chains are essentially an alliance chain.”

**Nanne2022: There are many indicators to distinguish between alliance chains and public chains, the most important being the entry threshold of nodes. Consortium chain nodes need to obtain official permission to enter the network, and users are not allowed to build full nodes that synchronize data without permission. The data of the consortium chain can only be synchronized within the nodes. If the outside world wants to obtain data, it can only use the interface provided by the internal nodes of the consortium chain. **

However, there is no entry threshold for the full node of the public chain, and the consensus node Validator can enter with a pledge of sufficient assets.

In addition, it is technically possible to distinguish whether it is a public chain or an alliance chain, such as consensus and network communication, the design of the public chain and the alliance chain is different. The alliance chain only considers a fixed number of nodes. The design principle is that the nodes are officially licensed, not unlicensed. This will make the technical implementation of the consortium chain relatively simple. In fact, neither Aptos nor Sui use the technology of the consortium chain, but the public chain.

What is really criticized is that Aptos and Sui are not decentralized enough (compared to Ethereum), and Sui officials have the ability to control the liquidity of SUI in the secondary market.

Note: At present, most public chains including Aptos and Sui have no access threshold for all nodes. Aptos’ consensus node can run with a pledge of 1 million APTs, and Sui’s consensus node needs to pledge 30 million SUIs, neither of which requires official permission.

**2. Regarding the claim at the beginning of the article, Aptos and Sui actually have no blocks or chains. **

Wubuku: First of all, Aptos has blocks, just look at its Node API documentation.

Secondly, Sui does not have the concept of blocks because it uses a DAG structure for transaction data. However, the narrow “blockchain” is not so important. What is needed to realize the Web3 vision is a “decentralized ledger”, not a specific data structure.

3. The author of the article thinks, "People in the web3 circle are quite ignorant. For example, after the emergence of Dfinity (ICP), many people are very excited, thinking that the narrative is grand and can solve many problems, but in fact it is the story of cloud native. How many years have Microsoft and IBM been doing it.”

Jolestar: There is indeed a problem with the build path of ICP. But the author should feel that the path of “decentralized cloud computing” is wrong, but Web3 is decentralized cloud native.

4. Regarding the guest Steven in the article, he believes that “Since Ethereum, there have been no new technological paradigms.”

Zhou Qi: If it refers to the earliest Ethereum, in fact, it and other technical paradigms change very quickly. ZK, DA, Verkle Tree, interactive fraud proof, the earliest initiators of these technical paradigms were not the Ethereum Foundation, but they were quickly borrowed and absorbed by Ethereum.

For example, Ethereum borrows from Celestia in DA and Cosmos’ Tendermint in the PoS consensus protocol. ZK is not considered the technology of Ethereum itself, but the Ethereum Foundation has spent a lot of effort in promoting the concept of ZK.

In addition, the Move language can also be regarded as an innovation in the technical paradigm. The paradigm innovations of Algorand and Cardano on POS are also important, but they have not yet undergone large-scale tests.

5. Regarding Steven’s opinion, “Solidity is a compiled language, and Move is an interpreted language.”

**Nanne2022: If you want to use “compiled” and “interpreted” to divide, Solidity and Move should be divided into one category. **

In the traditional division, the compiled language is to compile the code into a form that is closer to the bottom layer of the machine, such as assembly language, and then run by the machine link. An interpreted language is to compile the code into some higher-level bytecode, and then directly interpret and run the bytecode by the interpreter of this bytecode, such as Ruby and Python.

From this point of view, both Solidity and Move are first compiled into bytecodes that can be interpreted by the virtual machine, and then the interpreter outputs machine codes that the underlying hardware can understand during specific execution. No matter how interpreted and compiled are understood, Solidity and Move are in the same category.

**eternal: **Contracts written by Solidity and Move are first compiled locally on the host where the IDE platform where the contract is developed is located, and when uploaded to the chain, all are stored in Bytecode. From the perspective of the entire process of the language being processed by the compiler and interpreter, Solidity and Move are not fundamentally different.

6. Regarding Steven’s opinion, “The Move language was created by the former Facebook alliance chain project Libra. It is implemented based on the Rust language and was born for the alliance chain. Move inherits Rust’s hardware resource management and has good parallel capabilities. .”

Wubuku: First of all, Starcoin uses Move as the contract development language, and the consensus algorithm uses the more classic POW, which has nothing to do with the alliance chain. Then, Rust is a language designed for systems programming, and Move is a language redesigned for “smart contract development”. The so-called “Move inherits the hardware resource management of the Rust language” is actually irrelevant. Furthermore, it cannot be said that “Move inherits the hardware resource management capabilities of the Rust language” just because Move’s “resource-oriented programming” is inspired to a certain extent by Rust’s ownership mechanism (used to achieve memory safety).

eternal: Whether it is the contract code written by Solidity or Move, it is finally put into a sandboxed virtual machine to run, and the underlying hardware resources are not perceived at all. Move doesn’t have “hardware resource management capabilities” like Rust does. In addition, there are already Ethereum Layer 2 such as Rooch and ZkMove that support the Move language. Writing Move contracts does not have to be on Aptos and Sui.

7. Regarding Steven’s opinion, “Move is not a good language, because if it is separated from projects such as Aptos and Sui, Move will not work.”

Nanne2022: This is exactly the same for Solidity. If Solidity is out of the chain of supporting EVM, it will not work. “XXX is not a good language, because if you leave the XXX ecology, the XXX language will not work.” This is the same for many programming languages.

**eternal:**Steven will say this, mainly because Solidity and Move require a dedicated virtual machine to execute, which requires a dedicated program, but Java, which dominates the entire Internet development field, also requires a dedicated virtual machine. If it were placed shortly after Java was launched at the end of the last century, some people would say: “If Java is separated from the JVM ecosystem, it will not be usable, so it is not a good language.”

8. Regarding Steven’s opinion, “The entire system of Aptos and Sui is inherited from Libra, and there are improvements based on Libra, but not much.”

**Wubuku: **Aptos mainnet was launched earlier, and it is actually relatively close to Starcoin, but the consensus is a more “green” PoS.

Sui’s improvement is actually quite radical. In terms of the state storage model of the public chain/network, Sui Move has carried out a fairly thorough rewrite, abandoning the account storage model of Starcoin and Aptos and adopting the object storage model. Move developers who have done applications on these different public chains must be deeply touched by the obvious differences. Just because someone has team members who came out of Libra, it is not natural to think that they are counting on their money.

9. Regarding Steven’s opinion, “Libra is a relational database with a version number. Every transaction will update the state of the ledger, which is completely different from the traditional blockchain. The blockchain ledger will only add new fields, not Modify existing fields. This global state control method is only suitable for alliance chains, not for public chains.”

**Li: **Blockchain is essentially a verifiable Key-Value database, and there is no need to compare it with relational databases.

**Zhou Qi:**Ethereum’s ledger state model, StateTree, can modify existing data. It is a Key-Value storage model that supports addition, deletion, modification and query, and has a version number. And other traditional blockchain ledgers can also delete data. For example, if the utxo on the Bitcoin chain is used up, it will be deleted.

Jolestar: I feel that Steven’s understanding of the state tree StateTree is not very accurate.

10. Regarding Steven’s opinion, “Aptos sets up timestamps to arrange the execution order on the blockchain, and does not need to form a broad consensus among miners. This is easy to be attacked…block time is easy to predict evaluation, leading to a decrease in the security level of Aptos.”

Li: Aptos’ accounting model is basically the same as that of Ethereum, and the BFT consensus used is not sequenced by timestamp. The so-called “block time (block time)” can be estimated, but it will expose the leader at different times in advance. In theory, there may be hidden dangers of DDoS attacks. But this is a usability issue and has nothing to do with security. Moreover, the exposure time of the leader at each time point of Aptos is only tens of seconds, and the leader is rotated every second.

**eternal: **Aptos, whether it is a block or a transaction, must go through at least 2/3 of the miners (Validator) consensus, and it does not mean that it does not need to form a broad consensus among miners. It’s just that the transaction data on Aptos is consensus, which is completed before being packaged into the block. The block does not need to contain complete transaction data, as long as there is a summary.

Zhou Qi: The block producer at a certain point in time is estimated, and theoretically it can be ddosed. But secret leader election can solve this problem.

11. Regarding Steven’s opinion, “Aptos is just a relational database…just like a traditional relational database.”

Wubuku: Based on the experience of a senior CRUD boy who has worked on traditional enterprise applications and Internet applications for more than 20 years, the experience provided by Aptos and “relational database” is very different.

**Li: **Blockchain is essentially a verifiable Key-Value database, and there is no need to compare it with relational databases.

12. Regarding Steven’s opinion, "Sui does not need all nodes to reach a consensus…

Sui divides transactions into simple transactions and complex transactions. Simple transactions only need to be signed, and complex transactions only need BFT to be consistent.

As long as any transaction can be submitted, it can theoretically be verified independently… Sui directly abandoned the most basic consensus of the blockchain. "

Nanne2022: Consensus means that most nodes in the system can confirm a certain message and reach a steady state within a certain period of time (that is, they can continue to process subsequent messages). Whether it is a simple transaction or a complex transaction, the transactions in Sui need to collect the signatures of the nodes. If more than 2/3 of the nodes respond, the consensus will be completed for simple transactions, and the next consensus process will be entered for complex transactions.

Simple transactions will not conflict with other transactions and can be parallelized naturally. That being the case, it is natural to separate simple transactions from complex transactions for consensus verification, which can speed up.

As for why Sui can achieve fast confirmation of transactions, it also has a lot to do with the Narwhal memory pool protocol it adopts. Narwhal ensures that the messages to be consensus can be directly obtained locally, so that nodes can directly complete the consensus locally, without needing to obtain similar block messages from the network.

**Wubuku:**Sui is quite special. It can adopt different consensus methods for simple transactions and complex transactions. This starts with the object model adopted by Sui.

Sui’s object model includes two types: Owned Objects and Shared Objects. Among them, Owned Objetcs is similar to Bitcoin’s UTXO, but it is an “enhanced version of UTXO”. In contrast, Shared Objects are closer to Ethereum contract accounts that handle global data.

Sui’s simple transactions often only involve Owned Objects. The latter is similar to UTXO, only its Owner can operate, and has a version number, which naturally supports parallel processing. Then, a simple transaction that only involves Owned Objects can of course take a shortcut in consensus verification, and does not need to be sorted with other transactions.