Author: duoduo, LD Capital

Kaspa is a POW public chain built on the GHOSTDAG protocol. Compared with Bitcoin, Kaspa mainly changes the structural model of the blockchain. Bitcoin uses a single chain structure, and GHOSTDAG uses a directed acyclic graph (DAG) structure, in which one block can point to multiple blocks.

The token KAS was launched in November 2021, with a total supply of 28.7 billion, current circulation of 19.8 billion, accounting for 69% of the circulation, market value of US$750 million, and FDV of US$1.08 billion. Since the token was launched, it has increased hundreds of times.

1. Team

Kaspa’s team is well-known. Its founder Yonatan Sompolinsky is currently a postdoctoral fellow at Harvard University, and his research interests include transaction ordering and MEV. As early as 2013, he and his then doctoral supervisor conceived the GHOST protocol, and the related papers were cited in the Ethereum white paper.

Source: Kaspa official website

Here is a quote from the Ethereum white paper:

Source: Ethereum Whitepaper

In addition to the founders, there are 5 core developers. Michael Sutton studies parallel algorithms and distributed systems. Shai Wyborski is one of the authors of the GHOSTDAG paper and studies classical and quantum cryptography. Mike Zak and Ori Newman study distributed system development. Elichai Turkel is an applied cryptographer and blockchain high-performance developer.

Source: Kaspa official website

2. Technical Principle

The current technical principles of kaspa are mainly described in its paper "PHANTOM GHOSTDAG: A Scalable Generalization of Nakamoto Consensus" published in 2021.

Bitcoin is essentially an open and anonymous network of nodes that jointly maintain a public transaction ledger. The ledger uses the "longest chain" principle to link honest blocks to each other and protect the security of the network. This design artificially suppresses the network's throughput and reduces the scalability of the protocol. Currently, a Bitcoin block is produced every 10 minutes, with 3-7 transactions per second.

Structural model: directed acyclic graph (DAG) structure

Kaspa proposed the PHANTOM protocol, a permissionless ledger protocol based on proof of work that generalizes the blockchain defined by Satoshi Nakamoto to a directed acyclic graph (blockDAG). PHANTOM can reference multiple previous blocks, provide a total ordering of all blocks and transactions, and output a consistent set of accepted transactions.

PHANTOM contains a parameter k that controls the protocol's tolerance for blocks created simultaneously, which can be set to accommodate higher throughput. When k=0, it means there is no fork, which is the single, longest chain structure of Bitcoin.

来源:《PHANTOM GHOSTDAG:A Scalable Generalization of Nakamoto Consensus》

Let's first understand the different types of blocks in DAG. The corresponding concepts will be used in the examples below. In the above figure, take block H as an example:

past(H)={Genesis, C, D, E} — — past blocks that directly or indirectly point to H before H was created;

future(H)={J, K, M} — — H is created and directly or indirectly points to H ’s future block;

anticone(H)={B, F, I, L} — — blocks other than past(H) and future(H) that have no direct or indirect relationship with H;

tips(G)={J, L, M} — — Leaf blocks or end blocks, which will become the block header references of new blocks.

Identify honest and malicious blocks

PHANTOM solves the problem of identifying honest blocks and malicious blocks. Malicious attacks have a characteristic: the connectivity between blocks generated by malicious nodes and blocks generated by honest nodes is low, while the connectivity between blocks generated by honest nodes is high.

The judgment standard is the parameter K value mentioned above. For a specific block X, if the number of intersections between anticone(X) and honest blocks is higher than the k value, it means that the connectivity between X and honest blocks is low, and X will be judged as an attack block; otherwise, it means that the connectivity between X and honest blocks is high, and X is considered to be an honest block.

The following figure shows a judgment of honest blocks and attack blocks. The K value here is 3. After checking, the blue part is the honest block and the red part is the attack block.

来源:《PHANTOM GHOSTDAG:A Scalable Generalization of Nakamoto Consensus》

Linear sorting

In order to solve the double-spending problem, the project team used the GHOSTDAG protocol. The principle is to score each block according to its connectivity (the number of elements in the past block set), select the block with the largest total score to form the main chain, and the main chain will form the initial subset. The remaining blocks will vote in the order of the main chain. The entire network will vote according to the trend of connectivity from high to low.

The following figure shows how GHOSTDAG completes the sorting process when the parameter K = 3. The small circle next to each block X represents its score, which is the number of blue blocks in the past DAG.

Step 1, starting from the highest-scoring block M, select K, H, D, and the genesis block in turn, marked with a blue background and a black border, which forms the initial subset. Visit block D, whose only past block is the genesis block.

Step 2, visit block H. The past blocks of H are C, D, and E. After using the above-mentioned identification method for honest blocks and attack blocks, C, D, and E belong to honest blocks and are added to the subset and marked with a blue border.

Step 3, visit block K. K’s past blocks include H and I. After identification, they are all honest blocks and are marked with blue borders.

Step 4, visit block M. The past blocks of M include K and F. K is an honest block and is added to the subset, marked with a blue border.

Step 5, block V is a virtual block whose past is equal to the entire current DAG.

来源:《PHANTOM GHOSTDAG:A Scalable Generalization of Nakamoto Consensus》

At this point, Kaspa has completed the discussion of the new consensus architecture and put it into practice. The official website shows the visual production process of DAG:

Source: Kaspa official website

3. Computing power

The KAS mining algorithm is kHeavyHash, which supports GPU single mining or dual mining with ETHW and ETC, and supports mining with some FPGA and ASIC mining machines.

According to the official blockchain browser, Kaspa’s hashrate is 2.6–2.7 PH/s. According to Mining Pool data, Kaspa’s hashrate ranks around 30th, after BCH, BSV, and DASH, and before DOGE and LTC.

Kaspa's computing power shows a trend of continuous growth. In October 2022, December 2022, February 2023, and July 2023, it experienced four relatively obvious computing power increases. In March this year, a mining machine manufacturer launched a professional mining machine to improve the mining efficiency of miners.

Source: miningpoolstats

Judging from the distribution of computing power, the concentration of computing power is not too high. In the latest 999 blocks, the output of the top five mining pools accounted for 37.1%; more than 56.7% of the blockchain was produced by unmarked addresses.

Source: miningpoolstats

Token Economic Model

Token Allocation

KAS was launched in November 2021, with no pre-mine, zero pre-sale, and no token allocation; the total supply is 28.7 billion, the current circulation is 19.8 billion, accounting for 69% of the circulation, the market value is US$750 million, and the FDV is US$1.08 billion.

Token Release

According to the emission plan, KAS reduces its output in a given way every month, so that the output can be reduced by half every year. The figure below is a schematic diagram of the release. It can be seen that there is a higher release rate in the early stage, and early miners have accumulated a large number of chips. The specific release date table can be checked on the official website (https://kaspa.org/wp-content/uploads/2022/09/KASPA-EMISSION-SCHEDULE.pdf).

According to the token emission table, the monthly emission and release value of KAS tokens from July 2023 to June 2024 are calculated, as shown in the following table. If calculated at a price of 0.037, KAS will emit tokens worth $19 million in July 2023, and then gradually decrease; by June 2024, the emission value of tokens will be about $10 million.

Source: Kaspa official website, LD Capital

According to data from f2pool, Kaspa's 24-hour output value is currently fourth, second only to Bitcoin, Dogecoin and Litecoin, and higher than ETC and BCH.

Source: f 2 pool

Positions

There are 267,000 addresses holding 1 or more KAS. The concentration of tokens is also relatively high, with the top 10 addresses holding 17.299% of tokens, mainly exchange wallets; the top 100 addresses holding 26.13%; and the top 1,000 addresses holding 61.35%.

From the perspective of token liquidity, in the past 30 days, addresses holding 100-10K tokens have been reducing their positions and outflows, while addresses holding 0-100 tokens and above 10k have been inflows. In the past 7 days, addresses holding 100-10K and Humpback addresses holding 100m to 1b have been reducing their positions and outflows, while other addresses have been inflows.

Source: kaspa blockchain browser

V. Current Progress and Development Plans

The official website disclosed some important progress and near-term development plans since 2023.

completed

In February 2023, project core developer Michael Sutton published a paper on DagKnight Consensus, a consensus mechanism that is an evolution of GHOSTDAG and could theoretically lay the foundation for faster transaction and confirmation times.

We are tested

  • Rewrite the code using the Rust programming language

Currently, Kaspa is written in the GoLang programming language. Michael Sutton is rewriting the code using the Rust programming language, which will improve Kaspa’s performance and transaction speed.

  • Mobile wallet development

Users have a strong demand for high-performance mobile wallets, and the development time is estimated to take 3-4 months.

  • Integrating Kaspa on Ledger

Users are able to send and receive KAS using the Ledger hardware wallet.

Under research and development

  • Upgrade the consensus mechanism according to DagKnight Consensus

  • Further increase the number of blocks per second and transactions per second

Currently, Kaspa produces 1 block per second, and the goal is to increase the output to 32 blocks per second. The current test network can produce 10 blocks per second.

  • Release project white paper

There are several research papers surrounding Kaspa’s technology, but the official white paper for the project has not yet been released and is being compiled.

  • Improved Archive Node

Currently, Kaspa's standard nodes can only access transactions from three days ago. By improving the archive nodes, more historical data can be retrieved.

development plan

  • Implement smart contracts and build an ecosystem

Kaspa hopes to implement smart contracts, Defi and Layer 2 applications on its public chain and establish a corresponding ecosystem.

The development and deployment of smart contracts is the most important factor for its further development. If smart contracts can be deployed in a timely and smooth manner, and an active ecosystem is established, Kaspa’s market value will still have room for further growth. However, if smart contracts are not deployed smoothly and the ecosystem fails to develop, there will be obvious bottlenecks in future development.

VI. Conclusion

The project team has strong technical strength and a good development foundation. It has proposed a new blockchain model and tried a new development direction. It obtained the computing power transferred from Ethereum in the early stage, and the computing power continues to increase.

The market value is high, having risen more than 100 times. Currently, KAS has a market value of 750 million US dollars, ranking around 60th. The market valuation of the technical team and computing power growth is already relatively sufficient. Future growth requires stronger expectations to support it.

The potential selling pressure of tokens is relatively large. Early miners hold a large number of low-priced chips, and there is also continuous daily output. If they are sold, it will cause a large price impact.