Skip to main content

Exploring the Diversity of Decentralized Ledger Structures: DAG, Hashgraph, Block-lattice, and Tempo

Join us as we venture further into the cutting-edge world of Web3. After delving into the performance dynamics of mainstream public chains and Ethereum's upgrade path, we're expanding our view to explore the innovative approaches to decentralized ledgers that go beyond the traditional "blockchain" concept.

1. DAG / Tangle (Directed Acyclic Graph)​

Core Idea: "Transactions → Form a web → The whole network queues for itself"

  • Operation: Unlike the linear "train" model of blockchain, DAG resembles a constantly spreading spider web. Whenever a new transaction is created, it must first verify and "reference" (link) two or more existing transactions in the network. Thus, the transactions themselves form the nodes and edges of the network, eliminating the need for miners to package them into blocks.
  • Advantages: Since transactions can be integrated into the network in parallel, theoretically, concurrency is extremely high, with no "idle period" of block time. It is very suitable for scenarios requiring the processing of massive, small transactions, such as IoT (Internet of Things) micropayments.
  • Challenges: Effectively choosing which "old transactions" to reference (i.e., "Tip selection") is a challenge. Poor management or "selfish referencing" attacks may lead to some transactions not being confirmed for a long time. IOTA 2.0 introduces Mana (a reputation system) to resist such spam transaction attacks (Data Source: mdpi.com, pmc.ncbi.nlm.nih.gov).
  • Progress: Avalanche cleverly uses DAG for transaction ordering and combines it with its unique "avalanche" sampling consensus, which has been stably running on the mainnet for years, proving the feasibility of DAG in high-performance public chains (Data Source: medium.com).

2. Hashgraph​

Core Idea: "Gossip-about-Gossip + Virtual Voting, synchronizing the same timeline in everyone's mind"

  • Core Mechanism: Hashgraph is a purely consensus algorithm. Nodes continuously and randomly broadcast to neighbors the latest "events" they know (including transactions and timestamps), a behavior known as the "gossip protocol." Furthermore, they gossip not only about the events themselves but also about "who received whose gossip at what time" (Gossip-about-Gossip). Through this rich communication history, each node can independently compute a globally consistent, timestamped event order locally without any actual network voting (Data Source: docs.hedera.com).
  • Features: Fast transaction finality (usually within 5 seconds) and theoretically no forks, with fair and predictable ordering results. Its only public implementation, Hedera, offers the consensus engine as a service, with a single call costing as low as $0.0001 (Data Source: reddit.com).
  • Limitations: The Hashgraph algorithm is patent-protected, and the Hedera network is managed by a council including giants like Google and IBM, making its permissioned governance model controversial in terms of "absolute decentralization" (Data Source: solulab.com).

3. Block-lattice​

Core Idea: "Everyone has their own small ledger, with both parties referencing each other for confirmation"

  • Structure: In the Block-lattice structure, each user account has its own independent, mini blockchain. A complete transfer consists of two blocks:
    1. The sender records a send block on their chain and deducts their balance.
    2. Upon observing this send block, the receiver creates a receive block on their chain, increases their balance, and references the sender's send block. (Data Source: coinmarketcap.com)
  • Advantages: Since transactions occur and are confirmed asynchronously without waiting for global consensus, transfers are almost instantaneous (sub-second). The network has no unified blocks, thus no fees, and extremely low energy consumption. Its representative project, Nano, has been stably running for over 7 years with this architecture (Data Source: coinunited.io).
  • Challenges: When conflicts like double-spending occur, consensus decisions rely on "representatives" delegated by users for weighted voting. If voting power is overly concentrated in a few representatives, it may pose a centralization risk (Data Source: reddit.com).

4. Tempo (Radix Consensus)​

Core Idea: "Stamping logical clocks on all events, allowing them to naturally queue in a sharded world"

  • Principle: Tempo is a consensus protocol designed for large-scale sharding environments. It assigns a logical clock sequence number to every event in the network. Within a shard, all events are strictly ordered by logical clock. When a transaction needs to cross multiple shards, the system analyzes the causal relationships between these events to ensure they are correctly aligned, achieving cross-shard atomicity (i.e., either all succeed or all fail), a problem that traditional sharding solutions find extremely difficult to solve (Data Source: radixdlt.com, komodoplatform.com).
  • Highlights: Theoretically, by increasing the number of shards, its throughput can linearly scale to millions of TPS.
  • Current Status: The proposer of its core idea, Radix, has launched its Babylon mainnet, seen as a partial implementation of its ultimate consensus protocol Cerberus (an iteration of the Tempo idea), with the entire architecture still transitioning to its final form (Data Source: blocmates.com).
  • Challenges: As a highly complex system, its long-term resistance to attacks and network partition tolerance in a fully open, highly adversarial public network environment still requires time and practice for further verification.

Selection Considerations​

  • High-frequency micropayments → DAG / Block-lattice (Zero fees and second-level confirmation are key advantages).
  • Enterprise-level accounting and auditing → Hashgraph (The certainty of transaction ordering and fast finality are core requirements).
  • Large-scale DeFi / Gaming → Tempo (Theoretically unlimited linear scalability and native cross-shard atomicity are most attractive) or DAG + Avalanche Consensus (A high-performance solution validated by the market).
  • Pursuit of extreme low energy consumption → Block-lattice + Representative Voting (Asynchronous structure and lightweight consensus make its energy consumption far lower than other solutions).

Further Reading​

  • Performance Comparison of DAG-Based Ledgers — MDPI Computers 2023, an academic paper comparing the performance of various DAG ledgers.
  • Hedera Docs "Gossip-about-Gossip" — Hedera's official detailed explanation of the core mechanism of the Hashgraph consensus protocol.
  • CoinMarketCap Glossary "Block-lattice (Nano)" — A concise introduction to the block-lattice structure and how Nano works.
  • Radix Blog "Tempo Consensus: Lessons Learned" — A review and reflection by the Radix team on their early Tempo consensus design ideas.
  • Amber Group "Demystifying DAG Architecture" — An excellent industry research article that explains the advantages and challenges of DAG architecture in a simple and clear manner.
Share on Twitter