Basic information of the project

Operation Mode

Bittensor aims to create an AI algorithm aggregation platform to stimulate the development and decentralization of artificial intelligence. Bittensor adopts the design concept of Polkadot parallel chain, and connects to the mainnet through different subnets through Bittensor API to form the entire blockchain network. Miners and validators are the core modules of the subnet. Miners provide operation and upgrade support for the AI ​​model of each subnet, and validators evaluate the work quality of miners, and the evaluation results are used for the reward distribution of TAO tokens.

Subnet creators can earn token rewards by providing various AI-powered services, and users can purchase these services to accelerate their business development. Bittensor aggregates different AI models together, and the AI ​​type is determined by the algorithm, so Bittensor aggregates various types of AI algorithms in one place to better meet user needs, while encouraging knowledge sharing between subnets to promote collaborative research between different algorithms. Ultimately, the vision of the project is to create a more powerful decentralized AI model.

The operating mode of the Bittensor blockchain at the current stage of development is shown in Figure 1-1. Currently, 32 subnets have been launched on the network.

project team

Ala Shaabana, PhD in Computer Science, Machine Learning Researcher, previously worked at Instacart and VMware.

Jacob Steeves, a BS in Mathematics and Computing, is a machine learning researcher who previously worked at Google.

From the public information of Bittensor team members, it can be seen that the core team has rich development experience in the field of AI and provides solid support for the operation of the project.

Decentralization trend

Public information shows that the development of Bittensor has only been supported by the OpenTensor Foundation, and has not received any investment from individuals or institutions, which shows that the team intends to keep the project decentralized. It is worth noting that the Opentensor Foundation operates the largest verification pool at present, and is the largest validator of the Bittensor network, with a stake of 23.41%, which is twice as much as the second largest stake.

Bittensor's protocol governance is divided into three stages:

1. The foundation is centrally managed;
2. Senate model (current stage): The Senate is composed of staking pools with a staking amount exceeding 2% of the circulating market value of TAO tokens, with a total of 12 seats;
3. Community management;

According to the operation of the Senate model, the development control of the project is still concentrated in the hands of the Opentensor Foundation. Therefore, the Bittensor project is moving slowly towards decentralization, and its decentralization tendency is not significant. This centralized control structure may affect the in-depth practice and development of the project in terms of the concept of decentralization.

Development Strength

The Bittensor project was established in 2019 by founder Jacob Steeves and co-founder Ala Shaabana. A mysterious developer with the online nickname Yuma Rao assisted in writing the project white paper. The key events in the development of the project are shown in Table 1-1:

Table 1-1 Bittensor development key event information statistics

From the key time points of the official release, it can be seen that the team has a solid technical foundation in the field of technology development. However, the economic model design of the project, especially the distribution mechanism of TAO tokens, all came from a mysterious developer named Yuma Rao. Therefore, the distribution consensus mechanism of TAO tokens, Yuma Consensus, also got its name, making Yuma Rao known as the "Mr. Satoshi Nakamoto" of TAO tokens to highlight his core position in the progress of the project.

Innovation compared with similar projects

The AI ​​track in the cryptocurrency market can be divided into three sub-tracks: computing power, algorithms, and models. Bittensor is the pioneer of the algorithm track. Bittensor has many innovations that are different from other projects:

1. Business model innovation: With many projects focusing on computing power aggregation trading, Bittensor innovatively proposed the algorithm aggregation trading model for the first time, marking a breakthrough in this field and providing a new path for the development of related markets.
2. Decentralized Mixture of Experts (MOE): Each subnet focuses on a specific aspect of the data. When new data is introduced, the subnets work together to generate a collective and obtain an optimal answer. The speed and quality of this answer are much higher than the traditional single problem-solving model.
3. Digital hive mind : Bittensor encourages collaborative learning between network nodes to improve performance and accuracy. Similar to how neurons in the human brain work together, this process involves exchanging data samples and model parameters between nodes, forming a network that optimizes itself over time to achieve more accurate predictions.

Project Model

Business Model

The Bittensor economic system consists of three roles: users, validators, and miners

1. Miners: Miners are tasked with hosting AI models and providing them to the network. They are responsible for maintaining and ensuring the operation of the models. Miners will receive corresponding rewards based on their contribution to the network. The key success factor for miners is to reduce costs and provide high-quality services.
2. Validators: Validators are divided into subnet validators and mainnet validators. The task of subnet validators is to judge the quality of miners' products and services, while mainnet validators are used to evaluate the quality of results given by each subnet. They are also routing nodes between user needs and miners' services. The mainnet validator is the subnet validator with the largest stake, so the key success factor of the validator lies in the number of staked tokens. The larger the stake, the greater the influence of the validator on the final evaluation result.
3. Users: End users seeking high-quality results and developers building applications using AI services.

The Bittensor business process is shown in Figure 2–1.

Figure 2-1 Bittensor business process diagram

The official document does not mention whether users will be charged for using services within the network. The distribution plan for future service revenue is also not clearly stated. Therefore, the current operation of the network is completely dependent on the issuance of TAO tokens, which shows that the project is still in its early stages.

YUMA Consensus

Yuma Consensus is a consensus mechanism used to determine the distribution of newly produced TAO tokens. Its innovation lies in providing a complete evaluation system and rules for evaluating the output results of artificial intelligence models and the work quality of model supporters.

The distribution of TAO tokens is divided into 3 steps:

1. The mainnet validators score the 32 subnets and calculate the number of tokens that can be allocated to each subnet through Yuma Consensus.
2. The rewards allocated to each subnet are 18% to the mainnet creator, 41% to the subnet validator, and 41% to the subnet miner. The amount allocated to each miner and validator is determined by Yuma Consensus.
3. On the subnet, the subnet validator will score the miner's service quality. Based on this, Yuma Consensus uses the amount of TAO staked by each subnet validator as the weight to calculate the number of tokens each miner can get. The reward that each subnet validator can get is based on whether the evaluation result of the subnet validator is close to the final consensus result. The closer the result is, the higher the reward ratio is, and vice versa. This also reflects the mechanism of Yuma Consensus to prevent validators from doing evil.

Token Model

Token Allocation

The total amount of TAO is 21 million. According to the white paper, there is no pre-mining of tokens by the project party. From the first token, they are all mined by miners. Every token in circulation must be earned by actively participating in the network.

As of now, 6.64 million TAO tokens have been circulated, accounting for 31.64% of the maximum supply. The token distribution statistics are shown in Figure 2–2.

Figure 2-2 Statistics of TAO token distribution ratio (data source, Bittensor blockchain browser)

According to the white paper, the Bittensor network will halve every 10.5 million blocks, and plans to halve 64 times in about 45 years, and complete it after 256 years. The specific halving time is shown in Figure 2-3. The reward for each block is 1 TAO, which is issued approximately every 12 seconds, and a total of 7,200 TAO tokens are issued daily. These rewards will be distributed to miners and validators.

Staking Model

Only by staking TAO tokens can one become a subnet validator. Only a maximum of 64 subnet validators on any particular subnet are considered to be licensed validators, and only licensed subnet validators can receive rewards.

Ordinary users can stake their TAO to subnet validators and enjoy TAO token rewards.

85.3% of the circulating supply has been staked, which indicates the following:

1. Opentensor Foundation mined a large number of tokens at the start-up stage, and these tokens were staked. As an official mining pool, it attracted more TAO stakes, which is why it has always been at the top of the stake rankings;
2. New users are optimistic about the future price prospects of TAO tokens and pledge the tokens they received;

Therefore, although the staking model of TAO tokens is simple, it reflects the optimistic tendencies of ecosystem participants.

TAO's value judgment

In the Bittensor network, there is no centralized or regular destruction of TAO, so the total amount of tokens will continue to be in a state of inflation, but the inflation rate will continue to decrease over time, which is similar to the inflation mechanism of Bitcoin.

The circulation pledge rate has become the main reference factor for measuring token prices. The circulation pledge rate = the total number of pledged tokens/the total number of issued tokens.

• When the pledge rate is around 85%, TAO’s prospects can be considered stable;
• When the pledge rate is above 90%, TAO's prospects can be considered optimistic;
• When the pledge rate is below 80%, the TAO outlook can be considered pessimistic;

In summary, the current investment value of TAO lies in the staking rewards, and the long-term value lies in users using TAO to purchase services. As the price of TAO increases, the income of miners and validators increases, which attracts better AI models (miners and validators) to join, improves service quality, and thus attracts more users, forming a growth flywheel.

Token Price Performance

According to Coingecko statistics, since TAO began trading in April 2023, its price has increased 10 times (US$50-500). The main trading venues are Gate, MEXC, and Binance, with a daily trading volume of US$16 million. Its circulating market value is approximately US$450 million (market value of issued tokens - market value of pledged tokens), and its turnover rate is less than 5%, which is on the low side. This shows that most investors in the market are not familiar with TAO and lack the motivation to trade.

Project Risks

1. Centralization risk: As mentioned above, the Opentensor Foundation controls 23.4% of the staked amount, and this proportion continues to rise, so the degree of decentralization of Bittensor needs to be improved.
2. The flaws of Yuma Consensus: The mainnet validator is concurrently served by the strongest of the 64 subnet validators, and the objectivity of the model evaluation results does not entirely depend on the true evaluation of TAO holders. It is even possible that the validator is biased towards the subnet to which it belongs, resulting in the evaluation result obtained by the user not being the best choice, but a non-objective judgment made by the validator in order to obtain more rewards.
3. Subnet number limit: To compete with big companies like OpenAI, it is not enough to have only 32 simple AI models. However, due to the limit on the number of validators, it is impossible to add more subnets and thus more models, and the project still needs to be upgraded in this regard.

Recent Developments

According to the discussion summary of the official Discord, the following two proposals may be implemented in the future:

1. Discuss the establishment of a Dynamic TAO mechanism, through which the power to evaluate the quality of subnet work is transferred to all TAO holders rather than a few validators.
2. Discussions were held on removing the limit on the number of mainnet validators and increasing the upper limit on the number of subnets to 1024, making it easier for new models to access Bittensor.

If these two proposals are successfully implemented, Bittensor's decentralization will be greatly improved, the types of models will be richer, and its commercial competitiveness will be further enhanced.

Summarize

At present, all kinds of investors are paying unprecedented attention to AI concept assets. Computing power and algorithms have become two major areas that urgently need breakthroughs in the development of AI. As the pioneer of the algorithm aggregation concept, Bittensor has established a consensus evaluation system for model quality through Yuma Consensus, bringing together various types of AI algorithms to better meet user needs. At the same time, Bittensor encourages knowledge sharing between different subnets and promotes cooperative research between different algorithms, aiming to create a more powerful decentralized AI model. This project is expected to promote innovative development in the field of AI and drive the industry forward, and it deserves close attention.

Website: Frontier Lab

Follow Frontier Lab on X .

Telegram channel: https://t.me/frontierlab