Move Gemini: How Sui and Aptos Challenge the Blockchain Landscape

Introduction

Recently, the market has been cooling down significantly, causing even industry veterans to question the purpose of the industry. I’d like to share some of my personal thoughts: I believe that many of the great visions from the past were “disproven” because they were never logically consistent from the start.

Non-financial Dapps often tried to emphasize decentralization to cover up the fact that the products themselves were not good enough. The reality is, they ask me to distrust Google, Twitter, and YouTube and instead believe that their multi-signature wallets and single-server setups are secure.

Many visions were not disproven but were never truly tested. I still believe that even if most visions aren’t grand, they have meaning—they may just need a solid foundation to support them. In the end, at least one of decentralization or user experience comparable to Web2 can be offered.

Just like how TON and Solana were once underestimated, yet are now catching up with their predecessors in many aspects. The public chains that support applications need innovation, which in every cycle drives industry progress. So today, we will explore a type of public chain that has long been overlooked—the Move ecosystem.

1. Move

The Move programming language was originally created for Meta’s abandoned project, Diem (initially called Libra), which aimed to develop a more stable and regulated stablecoin as the foundation for its metaverse vision.

However, the project faced strong opposition and ongoing pressure from global regulators. They feared that Diem’s scale and Facebook’s vast user base could threaten financial stability, monetary policy, and data privacy. The Biden administration’s leadership in applying pressure ultimately forced Meta to abandon the Diem project.

Fortunately, the core of Diem was not discarded. Different factions split from the original team continued to explore and develop Move, which has evolved into the Move twins we know today—Sui and Aptos. Additionally, other projects like Linera (a Rust-based blockchain inspired by Move) and the recently promoted Movement are still in their infancy.

Why has the residual impact of a discontinued project been so significant? Move, as a blockchain programming language developed by a leading Web2 company, has a solid foundation. It was designed with a focus on addressing performance and security issues prevalent in existing blockchain languages, especially Solidity. Its design goals were to create a type system tailored for asset management and access control. I summarize the key points as follows:

• Security: Move’s design prioritizes security, using static type checks and resource management to prevent common vulnerabilities like overflow errors and reentrancy attacks. Compared to other language virtual machines, Move supports various security features, as illustrated by the comparison below from Nansen.
  
• Composability: Move supports modularity and composability, allowing developers to easily create and combine different smart contracts to build more complex applications.
• Performance: Move’s virtual machine is optimized (supports parallel processing, memory management, compiler optimizations) to efficiently execute smart contracts, enhancing transaction speed and throughput.

In a market saturated with modular EVM chains, Move represents a brave attempt at something different. You might have encountered similar claims in the introductions of many public chain projects, but I suggest experiencing it firsthand to truly understand these concepts.

2. Sui

2.1 Architecture

As one of the Move twins, Sui faced criticism early on for issues related to airdrops and token release mechanisms. However, setting these issues aside and focusing solely on the project itself, Sui has demonstrated excellence in both performance and user experience, particularly in the gaming sector.

This success is closely tied to its architecture, which was specifically improved for mainstream adoption. Here’s a brief overview of Sui’s architectural innovations:

1. Object Storage Model: This component is the core of Sui’s enhancements to Move. The object storage model stores data as individual objects, each with a unique identifier. Unlike traditional database systems, the object storage model lacks a fixed data structure and can store various data types, such as text, images, videos, and audio. This model allows for parallel execution and horizontal scaling (adding nodes to expand storage capacity). Sui’s design revolves around this model.
2. Causal Ordering: Ensures that transactions are executed in an order that respects causal relationships, avoiding data conflicts and inconsistencies. This allows Sui to handle large volumes of concurrent transactions while maintaining data consistency.
3. Narwhal and Bullshark Consensus Engines: Sui uses Narwhal and Bullshark as its consensus engines. Narwhal is responsible for transaction ordering and validation by maintaining a local transaction pool, sorting transactions based on causal relationships, and broadcasting them. Bullshark, upon receiving the sorted transaction list from Narwhal, votes on the list using Byzantine Fault Tolerance (BFT) consensus to ensure that all nodes agree on the transaction order.
4. Sui Move: Sui extends the Move language by adding new features, such as support for NFTs, asset management, and data storage.
5. Sui Framework: Sui provides a complete framework to help developers quickly build and deploy applications, including tools and libraries like Sui Wallet, Sui SDK, and Sui CLI.

Sui’s architecture allows it to process large volumes of concurrent transactions while maintaining high speed, low fees, and security. The Move language and Sui framework also provide developers with powerful tools to build secure, scalable, and user-friendly applications.

2.2 Consensus

The Sui blockchain uses a consensus mechanism called Mysticeti, a Byzantine Fault Tolerant (BFT) consensus designed for low latency and high throughput.

Mysticeti allows multiple validators to propose blocks in parallel, fully utilizing network bandwidth and providing resistance to censorship. Moreover, the protocol only requires three rounds of messages to finalize a block from the Directed Acyclic Graph (DAG), matching the theoretical minimum just like pBFT.

The submission rules allow for parallel voting and block leader certification, further reducing median and tail latency. The submission rules also tolerate unavailable leaders without significantly increasing latency.

Mysticeti had been running on the testnet for three months before Sui mainnet launch, showing significant results like an 80% reduction in latency. Now, the Sui network can process tens of thousands of transactions per second, with end-to-end latency well below one second.

The Sui blockchain also employs a specific type of Proof-of-Stake (PoS) consensus called Delegated Proof-of-Stake (DPoS). When a transaction involving shared objects (referred to as complex transactions) occurs, Sui uses the aforementioned Narwhal & Bullshark consensus engines for transaction ordering. Compared to other BFT consensus mechanisms, Sui’s advantages and disadvantages can be summarized in six points:

Advantages:

1. Low Latency and High Throughput: The Mysticeti protocol significantly reduces consensus latency and increases network throughput by parallel block proposals and optimized message flow, enabling Sui to handle tens of thousands of transactions per second with end-to-end latency well below one second.
2. Censorship Resistance: Mysticeti protocol allows multiple validators to propose blocks in parallel, improving the network’s resistance to censorship.
3. Leader Failure Tolerance: The submission rules tolerate unavailable leaders without significantly increasing latency, automatically electing a new leader to take over when the leader node fails.

Disadvantages:

1. Complexity: The Mysticeti protocol’s design is relatively complex and requires a deeper technical understanding to fully grasp its operational mechanisms.
2. Security: Although the Mysticeti protocol performed well on the testnet, its security still needs further validation in real-world applications.
3. Scalability: The scalability of the Mysticeti protocol needs further observation to ensure it can adapt to future growth in network size and transaction volume.

2.3 Abstract Accounts

Sui’s abstract account model (Account Abstraction) is a mechanism that allows users to manage their accounts and transactions more simply and securely by abstracting account and transaction logic from the underlying blockchain protocol, achieving a higher level of account management and transaction processing.

In Sui’s abstract account model, accounts are no longer simple public-private key pairs but objects with richer attributes and behaviors. Each account has a unique identifier, called an account ID, which is associated with the account’s public and private key pair.

Sui’s abstract account model includes the following key components:

• Account Object: The fundamental unit of accounts in Sui. Each account object has a unique account ID and contains the account’s attributes and behaviors.
• Account Data: The core component of the account object, containing basic account information such as account ID and public-private key pair.
• Transaction Context: The basic unit of transactions in Sui, containing transaction-related information such as transaction ID, account ID, and transaction data.
• Account Logic: The collection of account behaviors and rules in Sui, defining how accounts handle transactions and manage their state.

The process of handling transactions in Sui’s abstract account model involves the following steps:

1. Transaction Creation: Users create a transaction and send it to the Sui network.
2. Transaction Validation: The Sui network validates the transaction’s validity and integrity.
3. Account Lookup: The Sui network finds the corresponding account object based on the account ID in the transaction.
4. Account Logic Execution: The Sui network executes the account logic to process the transaction and update the account state.
5. Transaction Confirmation: The Sui network confirms the transaction result and writes it to the blockchain.

In simple terms, Sui’s abstract account model is an innovative mechanism that simplifies account management and transaction processing, making applications more application-like.

2.4 Gaming

For a blockchain to stand out, it must first build a solid foundation. I referred to the Move ecosystem as a bold experiment earlier for two main reasons.

First, in an era where the concept of modularization is becoming widespread, the native Move ecosystems (specifically the Move twins) represent the last attempt at Layer 1 solutions, going against the current trend.

However, the recent rise of several heterogeneous chains may be proving that modularization is not the only answer.

Second, rebuilding a blockchain using a new programming language is akin to trying to create a new operating system to challenge iOS and Android in today’s smartphone market—a path inevitably filled with challenges.

Whether the Move ecosystem will shine like Solana in the coming years depends significantly on the direction it chooses. Sui’s answer to this challenge is gaming.

Gaming is one of the key gateways to Web3, yet most blockchains do not support gaming well. This is because blockchain technology was originally designed around financial applications, and its decentralized architecture inherently has lower performance, making it ill-suited for gaming.

Sui, however, is different. Its model is well-suited for both DeFi applications and non-financial applications, including games. As previously mentioned, in Sui, everything is an object. In Sui, objects can own other objects, which makes it possible to model complex asset hierarchies typical in games or applications.

Imagine playing a game where a hero character has an inventory, and that inventory holds other digital assets belonging to the character. Sui can model these data hierarchies in ways that other blockchains cannot. This gives developers the opportunity to build applications without having to work around the inherent limitations of the chain.

Additionally, Sui is actively collaborating with traditional Web2 giants. Last year, Sui established partnerships with three of the four major South Korean gaming giants (Netmarble, NHN, and NCSoft). This year, Sui partnered with TikTok to develop blockchain games and SocialFi projects, bringing traditional giants into the Web3 space.

3. Aptos

Aptos, another Layer 1 blockchain based on the Move language, is also committed to building a high-performance, scalable Web3 infrastructure. Its architectural design shares many similarities with Sui, but it also introduces some unique features.

3.1 Architecture

1. Modular Design: Aptos adopts a modular architecture that allows developers to independently develop and upgrade different modules, enhancing development speed and flexibility.
2. Parallel Execution Engine (Block-STM): Unlike other blockchains that require pre-declared data dependencies, Aptos’ parallel execution engine can process transactions in parallel without prior knowledge of data locations, thus increasing throughput and reducing latency.
3. Pipelined Transaction Processing: Aptos divides transaction processing into multiple stages, such as propagation, metadata ordering, and batch storage, executing these stages in parallel through pipelining to maximize throughput and minimize latency.
4. Move Programming Language: Aptos uses the Move programming language, focusing on refinement rather than innovation compared to Sui. For example, it standardizes the language, introduces more robust function support, and enhances customization capabilities.
5. Flexible State Synchronization: Allows nodes to choose different state synchronization strategies, such as syncing the full history or only the latest state, improving node flexibility.
6. AptosBFT Consensus Mechanism: AptosBFT is a Byzantine Fault Tolerance consensus mechanism used by Aptos. It optimizes communication and synchronization between validators to improve throughput and reduce latency. Compared to Sui, which can be seen as an improved version of DiemBFT, AptosBFT has made certain advancements in efficiency and crash recovery, so it will be briefly mentioned here.

Aptos’ architecture enables it to handle a large number of concurrent transactions while maintaining high speed, low fees, and security. Additionally, the Move language and Aptos framework provide developers with powerful tools to build secure, scalable, and user-friendly applications.

3.2 Block-STM

Let’s delve deeper into Aptos’ core innovation, the parallel execution engine Block-STM:

Core Principles of Block-STM:

1. Predefined Sequential Execution: Block-STM relies on a predefined sequence of transactions within a block. All transactions must be executed in this order to ensure consistency in the final state.
2. Optimistic Concurrency Control: Block-STM optimistically executes transactions in parallel, assuming no conflicts will occur. This control method is based on the assumption that conflicts are rare, allowing transactions to access and modify data without locking. It hypothesizes that the likelihood of simultaneous conflicts is low, so modifications can proceed, with checks for conflicts before the final commit.
3. Multi-Version Data Structure: To support optimistic concurrency control, Block-STM uses a multi-version data structure to store data. Every write operation creates a new data version, while read operations access the corresponding data version.
4. Validation and Retry: After executing a transaction, Block-STM verifies whether the versions of the data it read are still valid. If validation fails, indicating a conflict, the transaction is marked invalid and re-executed.
5. Collaborative Scheduling: Block-STM employs a collaborative scheduler to coordinate the execution and validation tasks of various threads, maximizing parallelism.

Block-STM Workflow:

1. Transaction Grouping: Group transactions within a block and assign them to different threads for parallel execution.
2. Optimistic Execution: Each thread optimistically executes its assigned transactions and records each transaction’s read/write set.
3. Validation: Once a thread completes a transaction, it verifies whether the data versions in its read set are still valid.
4. Retry: If validation fails, indicating a conflict, the transaction is marked invalid and re-executed.
5. Commit: After all transactions pass validation, their results are written to the blockchain state, completing the transaction commit.

Advantages of Block-STM:

1. High Throughput: By leveraging optimistic concurrency control and collaborative scheduling, Block-STM fully utilizes multi-core processors, achieving high throughput.
2. Low Latency: Since transactions can be executed in parallel, Block-STM significantly reduces transaction confirmation time.
3. Security: Block-STM’s predefined sequential execution and validation mechanisms ensure consistency and security in the final state.

In summary, Block-STM is an efficient parallel transaction execution engine that combines optimistic concurrency control, multi-version data structures, and collaborative scheduling to maximize blockchain throughput while ensuring security and correctness.

3.3 Account Abstraction

Unlike Sui’s more direct approach to account abstraction, Aptos supports a more limited abstraction dimension and lacks specific predefined standards. Its account abstraction capabilities primarily manifest in the following areas:

1. Modular Account Management: Uses Move modules to define and manage accounts, allowing developers to create custom modules for different account types and functionalities.
2. Flexible Key Management: Allows users to use different keys for different account operations, such as using one key for transaction signing and another for account management.
3. Programmable Transaction Verification: Developers can define custom transaction verification logic in Move modules, such as multi-signature or limits, to meet different application scenarios.

3.4 Partnership with Microsoft

Compared to Sui, which focuses more on game development, Aptos does not have a specific development goal. Its slogan is to be the blockchain most suited for production.

One notable aspect is Aptos’ ongoing collaboration with Microsoft, aiming to integrate Microsoft’s AI technology into blockchain. Their first collaborative product, Aptos Assistant, a generative AI assistant built on the Aptos network, is already live on the official website. Additional AI products will be announced over the coming months.

4. The Move Ecosystem

Although Sui has recently performed well, the Move ecosystem still requires time to mature compared to EVM-based chains, Solana, Ton, and other heterogeneous chains. Despite Sui and Aptos being in the spotlight and making technological breakthroughs, the overall scale and activity of the Move ecosystem are not yet on par with other mature ecosystems.

The number of developers, types of applications, and user base still need time to grow. In terms of external collaborations and operations, both Sui and Aptos exhibit a strong Web2 mindset, lacking some of the Web3 essence, which has kept their projects relatively low-key within the industry.

However, considering the potential of the Move ecosystem, it holds much promise. Some developers are already recognizing Move’s future value. As mentioned in the preface, there are already projects incorporating Move into Ethereum Layer 2 solutions. In the future, the Move ecosystem may shine within the ETH Layer 2 space. The current focus is on how to further promote the Move ecosystem.