Understanding the Current Status and Trends of DAG Technology in One Article

Author: Li Qiang / Founder of DAG Application Bsure

This article aims to introduce readers to the architecture, development trends, and typical applications of DAG technology, hoping to provide a new perspective for thinking about the future development of blockchain.

1. What is DAG?

DAG: Directed Acyclic Graph.

DAG is originally a commonly used data structure in the field of computer science. Due to the excellent characteristics brought by its unique topological structure, it is often used in various algorithm scenarios such as dynamic programming, seeking the shortest path in navigation, and data compression.

Now, let's look at the differences between traditional blockchain and DAG, simply put:

1) Unit: The unit of blockchain is Block, while the unit of DAG is TX (transaction);

2) Topology: Blockchain is a single chain composed of Block blocks, which can only be written sequentially according to block generation time, similar to a single-core single-threaded CPU; DAG is a network composed of transaction units that can asynchronously and concurrently write transactions, similar to a multi-core multi-threaded CPU;

3) Granularity: Each block unit in blockchain records multiple transactions from multiple users, while each unit in DAG records a single user's transaction.

Several issues exist in traditional blockchain technology:

1) Efficiency issue: Traditional blockchain technology is based on Block blocks, and Bitcoin's efficiency has always been relatively low. Due to the chain-like storage structure of BlockChain, the entire network can only have one single chain at the same time, and block generation based on the POW consensus mechanism cannot be executed concurrently; for example, Bitcoin generates a block every ten minutes, and it takes about an hour to confirm after six blocks are generated; Ethereum has significantly improved this, but block generation speed still takes several seconds.

2) Determinacy issue: Bitcoin and Ethereum have the 51% hash power attack problem. The biggest hidden danger of POW consensus is that there is no definite, immutable final state; if a group controls 51% of the hash power and launches an attack, the Bitcoin system will definitely collapse; considering the reality of miner groups in the real world and the rapidly developing quantum computers with extraordinary computing power, this danger is real.

3) Centralization issue: In the POW consensus based on blocks, miners can form centralized mining groups on one hand, and on the other hand, miners who gain the right to package transactions have enormous power to choose which transactions enter the block and which transactions are not processed, and they can even only package transactions that benefit themselves. This risk already exists in reality.

4) Energy consumption issue: Due to the traditional blockchain's reliance on POW proof of work for consensus, Bitcoin's mining energy consumption has reached the same level as that of an entire country like Argentina. The IMF and several governments have criticized the energy consumption of cryptocurrency mining.

Digiconomist data indicates that global mining operations generate approximately 290 million tons of carbon emissions annually.

Due to the above issues, some have raised the question: Why is it necessary to have blocks? DAG technology is being used to attempt to solve the aforementioned problems of blockchain.

2. Origin of DAG

The concept of DAG was first introduced in blockchain as a consensus algorithm in 2013 by an Israeli scholar from the Hebrew University, ID avivz78, on bitcointalk.org, known as the GHOST protocol, as a scalability solution for Bitcoin's transaction processing capability; Vitalik described the POS consensus protocol Casper in the Ethereum white paper, which is also a POS variant based on the GHOST POW protocol.

Later, someone in the NXT community proposed using the topological structure of DAG to store blocks to solve the efficiency problem of blockchain. Blockchain has only one single chain, and block packaging cannot be executed concurrently. If the chain-like storage structure of blocks is changed to a DAG mesh topology, transactions can be written concurrently. Under the condition that the block packaging time remains unchanged, the network can parallel package N blocks, allowing the transactions in the network to accommodate N times.

At this point, the combination of DAG and blockchain still remained at a similar sidechain solution, where transaction packaging could be performed in parallel on different branch chains to achieve performance improvement. At this time, DAG still had the concept of blocks.

DAG of blocks proposed by the Nxt community

In September 2015, Sergio Demian Lerner published an article titled "DagCoin: a cryptocurrency without blocks," proposing the concept of DAG-Chain, which elevated the DAG network from block packaging to the transaction level for the first time. However, DagCoin itself was a paper without code implementation.

The idea of DagCoin allows each transaction to directly participate in maintaining the transaction order of the entire network. After a transaction is initiated, it is directly broadcast to the entire network, skipping the block packaging stage, achieving what is called Blockless. This eliminates the time spent on packaging transactions into blocks. As mentioned earlier, the initial combination of DAG with blockchain was to solve the efficiency problem; now, without the need for packaging confirmation, transactions are directly broadcast to the network for confirmation after initiation, theoretically achieving a qualitative leap in efficiency. DAG further evolved into a solution that completely abandoned blockchain.

In July 2016, based on the genesis post published on the Bitcointalk forum, IOTA emerged, followed by ByteBall, marking the first true technical implementation of a DAG network, and they became the most prominent leaders in this field; at this point, the prototype of the so-called blockless chain, the DAG chain family, was basically formed.

In summary: DAG is a new generation of blockchain technology aimed at the future, evolving from a single chain to tree-like and mesh structures, from block granularity to transaction granularity, and from single-point transitions to concurrent writes. This represents an innovation in blockchain from capacity to speed.

3. Current Development Status of DAG

The current representative projects of the DAG system are undoubtedly the "three horse-drawn carriages" of DAG: IOTA, Byteball, and Nano (formerly RaiBlocks). As the latest distributed ledger competitive technology, DAG began to attract significant attention when IOTA surged to the fourth place in market capitalization rankings in the second half of 2017, followed by a continuous influx of new projects based on DAG technology into the public eye.

1) IOTA

The main innovation behind IOTA, Tangle, is a newly designed distributed ledger structure based on DAG, which is a blockchain without blocks or chains. In Tangle, each node represents a transaction. There is no concept of blocks in IOTA, nor is there a concept of mining and miners, which means there are no transaction fees, and the overall throughput of the network is very high, which is one of the most attractive features of IOTA.

The core principles of Tangle are consistent with those of blockchain; it is still a distributed database, a P2P network, and a consensus algorithm to verify transactions. The main difference between Tangle and traditional blockchain lies in the data structure and consensus mechanism.

In IOTA, there is no concept of blocks; instead, there is a transaction network where each transaction references the hashes of the two previous transactions. This way, the previous transaction proves the legitimacy of the past two transactions and indirectly proves the legitimacy of all previous transactions.

Thus, the entire network participates in verifying the legitimacy of transactions, unlike traditional blockchain, where only a small number of nodes (miners or PoS stakeholders) verify transaction legitimacy. Therefore, the consensus of IOTA is its internalized characteristic, allowing it to be used on a large scale without transaction fees. In IOTA, there is no longer a concept of blocks; the minimum unit of consensus is the transaction.

Another powerful aspect of Tangle is that transactions can be freely detached from or merged back into the network. This ability for offline asynchronous processing is particularly important in applications within the Internet of Things (IoT) field.

Current issues with IOTA include:

First, a MIT report pointed out that IOTA uses its own developed hash algorithm, curl, which is prone to collisions, allowing for the forgery of digital signatures.

Second, since consensus is determined by the entire network's transactions, theoretically, if someone can generate 1/3 of the transaction volume, they can turn invalid transactions into valid ones. On the other hand, since IOTA has no transaction fees, there is no incentive for miners, making IOTA susceptible to denial-of-service attacks and spam attacks, similar to a community that does not collect property fees, where it is difficult to eliminate illegal elements relying on self-governance.

Third, IOTA introduces a closed-source centralized component called Coordinator to check transactions across the network (e.g., double spending). How to effectively remove the Coordinator and establish a decentralized "Coordinator group" with a healthy incentive mechanism is still unresolved by IOTA.

2) Byteball

Byteball is known as the representative of blockchain 3.0. It has the most complete application ecosystem within the DAG family. The Byteball wallet has rich built-in features, including a BotStore similar to an Appstore, where developers can freely develop applications, and developers are very active.

Building on DAGCoin, Byteball innovatively introduces the concepts of a main chain and witnesses, encouraging the verification of multiple parent transaction units, forming a digital signature hash network that grows with transactions and strengthens security through mutual verification. Byteball creatively invented the concept of a "main chain," which is the shortest path MC's Parents optimization algorithm recognized by witnesses. The main chain creates a transaction time sequence determined by consensus across the network, elegantly avoiding the double spending problem.

In Byteball, the "witness" truly forms the consensus mechanism; the transaction units published by 12 "witnesses" delineate a deterministic transaction time sequence in the theoretically infinitely broad DAG concurrent transaction network. It is this infinitely extending time-based deterministic transaction sequence that creates the main chain in Byteball, forming a robust and orderly unique backbone in the vast and unordered directed acyclic hash world. Based on the witness + main chain consensus mechanism, issues like double spending are easily resolved.

Byteball eliminates the concepts of blockchain and proof-of-work mining, opting for DAG data storage technology instead. Compared to traditional blockchain-based cryptocurrencies, this has significant advantages, as all transactions in Byteball are cryptographically linked. Newly generated transactions will be added behind the tips transaction units. This allows all nodes (users) on the network to participate in verifying transactions, achieving complete decentralization.

This not only allows for faster payment verification but also keeps the network sufficiently decentralized, avoiding some issues seen in Bitcoin, such as large centralized mining pools that could threaten the network. Additionally, Byteball reduces spam on the network by charging a storage fee for each byte of data stored in the DAG network, similar to a Gas mechanism.

Some argue that Byteball's Witness has a centralized dependency; however, it should be clarified that in Byteball's Witness node design, witness nodes are highly secure and can only issue witness units without accessing transactions, so Witnesses are not miners. Witnesses act as WatchMen, helping the system anchor the order of transaction occurrence without leaving room for malicious actions; moreover, the number of Witnesses can be set and selected as needed, not limited to the current 12 witnesses.

Due to each transaction having the initiator's private key signature, and each transaction verifying and referencing previous transactions, Byteball weaves a massive network. Tampering with the network would affect the entire system, and it is impossible for anyone to possess all users' private keys across the network, thus Byteball possesses bank-level finality.

The issues with Byteball are: due to the relationship between the main chain algorithm and the frequency of witness publication, the time for transaction confirmation is uncertain; since Byteball uses a relational database to store data, the SQL language is too tightly coupled with algorithm logic, which somewhat limits Byteball's current scalability and speed.

3) NANO (XRB)

Nano (formerly RaiBlocks XRB) is a new type of cryptocurrency based on a block lattice structure.

Nano innovatively adopts a user-specific chain approach, recording only its own transactions, and only the user can modify the records, not sharing data with other accounts. This allows all transactions to be executed in parallel, providing second-level transaction speeds and infinite scalability, while allowing them to asynchronously update to the rest of the network, achieving fast transaction confirmations with minimal resource overhead.

A Nano node can store the historical ledger of all accounts or only store the last pruned record of each account. When a transaction occurs, the sender generates a send tx block that includes the deducted amount; the receiving account generates a receive tx block to record the corresponding amount received. The sending and receiving of transaction data can occur asynchronously, so even if multiple amounts are sent to one account simultaneously, it is not an issue; the final amount is the sum of the received amounts. If the recipient is offline, it doesn't matter; the unreceived amount will be separately marked, and once the receiving account comes online, this amount will be transferred from the unsettled area to the receiving block, completing the transaction.

NANO uses a DPOS consensus mechanism, where accounts can designate representatives to vote. The representative with the most votes will handle forks, broadcasting the fork to the network and observing the voting results from high-stake account nodes within a fixed time to determine which block to retain. DPOS ensures the reasonable low-energy operation of blocks. NANO also employs a POW mechanism, requiring very little proof of work (PoW) to confirm transactions.

The issues with NANO are: it has not been thoroughly tested and lacks peer review, posing risks of serious flaws in the consensus algorithm. For example, what happens if there are not enough votes to resolve network conflicts? Another major issue: what happens if certain parts of the NANO network are separated for a long time, and when the separated network rejoins, will it paralyze during the inevitable voting process?

4. DAG Innovation and Trends

DAG technology is rapidly developing and innovating. In addition to the three major projects, new DAG projects are making progress in consensus algorithms, decentralization mechanisms, speed, and concurrency. Recently, projects that have attracted significant attention include:

1) HashGraph

Hashgraph is a gossip protocol consensus algorithm developed by Leemon Baird. All nodes randomly share their known transactions with other nodes, allowing all transactions to eventually propagate to each node. Hashgraph is extremely fast (over 250,000 transactions per second) and, due to being closed-source and patented, is suitable for private or consortium chains, and will not be applied to public chains or receive large-scale validation in the short term.

Hashgraph innovatively implements asynchronous BFT consensus in a public chain environment. One major issue with traditional BFT is that message complexity is too high, consuming a lot of the system's network bandwidth, making it difficult to cope with dynamic networks. Here, Hashgraph introduces the traditional Gossip Protocol and adds unique innovations, along with a virtual voting mechanism, so that when consensus is needed, it does not trigger a sudden large-scale message transmission storm.

Hashgraph and Algorand have improved the application scenarios and conditions of BFT by addressing different aspects, allowing BFT consensus to be applied to public chain systems. Hashgraph reduces the instantaneous communication requirements traditionally needed for consensus to a minimum through gossip propagation of the hash graph and virtual voting based on the hash graph, while local computation ensures the efficiency of consensus.

The latest commercial introduction of Hashgraph mentions plans to switch to POS and support DOPS, allowing holders who do not run full nodes to choose agents and share profits.

Hashgraph combines the strengths of various approaches, achieving significant breakthroughs in scalability, security, and consensus cost, but the technical difficulty is high, and it has not yet operated in a large-scale public chain environment. If it can achieve rigorous mathematical and application verification as described in the Hashgraph white paper, then Hashgraph could become an important milestone in exploring trust on the internet.

2) SPECTRE/PHANTOM

The SPECTRE Protocol adopts the "block directed acyclic graph" technology of Block+DAG, allowing for parallel mining, thus bringing greater throughput and faster transaction confirmation times. In February 2018, the scaling protocol of SPECTRE—Phantom—was released, significantly expanding the network's transaction capacity and being compatible with smart contracts. This technology is a "generalization of the blockchain proposed by Satoshi Nakamoto," solving the trade-off between security and scalability, making it more suitable for establishing faster or larger-scale blocks.

Unlike off-chain solutions like the Lightning Network, PHANTOM is an on-chain scaling solution. At the same time, PHANTOM's linear ordering sacrifices some of the transaction confirmation speed achievable by SPECTRE.

3) Hycon

Hycon is a DAG project from South Korea, positioning itself as a platform public chain, aiming to create an ecosystem that includes a decentralized exchange for value exchange media, preparing to raise nearly $100 million in funding, with 70% to be mined in the future.

The establishment of the entire Hycon ecosystem is divided into three phases: value exchange media, blockchain platform, and decentralized exchange, aiming to create a value ecosystem that integrates value exchange, commercial applications, and token circulation. Among them, the blockchain platform is the core of the entire ecosystem, aiming to solve the performance bottlenecks of blockchain in terms of low transaction confirmation speed and limited throughput, thereby achieving commercial-grade applications.

The main features of the Hycon public chain platform include: fast transaction confirmation time, on-chain transaction scalability (up to 3000 TPS transaction throughput in a 2MB/S connection), synchronized block generation (linking blocks based on DAG structure location rather than chronological order), and smart contracts.

4) Algorand

Recently, the star project Algorand has gained popularity both domestically and internationally, aiming to establish a low-energy, high-speed, democratized, scalable distributed ledger that almost never forks. Algorand does not introduce incentive mechanisms or issue digital cryptocurrencies. Algorand was developed with $4 million raised by Turing Award winner and MIT professor Silvio Micali.

GHOST, SPECTRE, and Meshcash are some recent proposals aimed at enhancing Bitcoin's throughput by using trees or directed acyclic graphs to replace the underlying blockchain ledger structure, addressing the conflict issues of blockchain forks. These protocols use POW to prove consensus and can significantly improve throughput through carefully designed selection rules between tree/graph branches.

China's new technology sector is not slow to follow the script of innovation; blockchain systems based on DAG are emerging rapidly. However, there is still a gap in the innovation of underlying consensus design and methodological innovation. We have compiled some typical projects based on DAG technology currently in China:

1) ITC (Wanwu Chain)

A lightweight operating system for secure IoT based on blockchain, the solution integrates blockchain technology with cryptographic asymmetric encryption technology, semi-homomorphic encryption ciphertext computing technology, and a distributed architecture without data centers, aiming to solve the serious security issues currently facing IoT and meet the highly concurrent usage scenarios of IoT, achieving interconnectivity among all things.

2) TrustNote

TrustNote is a DAG public chain that supports POW mining, featuring an innovative dual-layer consensus mechanism, targeting applications such as digital token issuance, blockchain games, and social networks. The base token is called "TTT," focusing on the development of the underlying public chain, with applications in financial credit, information security, IoT, gaming, social networking, and more. It has already made some inroads into application fields.

3) Bsure

Bsure is a professional digital insurance and health blockchain smart platform, building a public chain for the digital insurance and healthcare industry based on DAG technology. The test chain has already gone live, and development is underway for the Bsure.cloud chain cloud infrastructure to empower innovation in healthcare and insurance technology products and services.

The main features in development for the Bsure industry public chain platform include: fast transaction confirmation time (instant payment), transaction scalability (optimizing the main chain algorithm through in-memory computation to enhance transaction concurrency), Package+DAG consensus innovation (combining local partial order packages with notarized main chain total order algorithms), and a layered empowerment architecture that includes an independent smart contract layer, decentralized storage layer, DAPP application layer, and homomorphic cross-chain capabilities.

4) Nerthus

A low-profile project initiated in the second half of 2017, Nerthus is based on an improved Byteball DAG structure and uses the GO language to implement a three-layer architecture consisting of service layer, core layer, and application layer. It has not yet been listed on exchanges. Nerthus is likely the first platform public chain project in China developed based on DAG technology, and the project is currently under development.

5) CyberVein

CyberVein is based on a DAG architecture, incorporating its own Vein programming language, virtual machine, and new type of smart contracts, dedicated to solving the data value definition and management issues faced in the era of big data from both technical and business logic perspectives.

CyberVein is completed through a combination of DAG + PoC mechanism + database virtual machine. The DAG architecture is just part of CyberVein; the platform also allows for the creation of its own smart contracts, has a virtual machine for operating databases, and features innovative consensus mechanisms. If we compare the project to a well-decorated building, the DAG architecture is merely the foundation of CyberVein, the database virtual machine is the concrete and steel of the building, and the programming language and smart contracts are the decoration of each room, while PoC contribution proof is the intelligent access control system of the building!

6) Mixin

Mixin plans to establish the largest developer-friendly mobile blockchain network, linking all existing blockchains with infinite throughput, based on DAG.

Mixin has grand plans, including cross-chain capabilities, infinite throughput, and account and asset management based on phone numbers, aiming to become the public chain with the largest user base in China in the future. A test version of a Dapp based on it, Mixin Messenger, has already gone live. Other projects based on Mixin, such as Pressone, Motion, and Show.one, are also under continuous development.

7) Travelflex

A decentralized social travel network and payment system based on DAG technology, aimed at solving the scalability issues currently faced by Bitcoin and other tokens.

5. Conclusion

Every advantage comes with a disadvantage. DAG's speed and high throughput are impressive, but as a very young data structure, its security and consistency still require more validation and recognition, and its application scenarios are not yet as widespread as traditional blockchain; however, the advantages and innovative speed of DAG technology have already emerged, with more and more subsequent innovative projects and DApps based on DAG rapidly surfacing.

Currently, people like to compare DAG with traditional blockchain, but we believe that DAG is merely one technical implementation of distributed ledger applications, and some innovative blockchain projects are already exploring the integration model of Block+DAG.

Broadly speaking, DAG is just one component technology of a broader blockchain. The blockchain and token economy, as key components, will jointly embrace the arrival of the fourth wave of the value internet alongside AI, big data, AR/VR, and 5G high-speed wireless networks.