# Why we need our own L2?

## Very short

1. **Technical | Security and Privacy**
- **Existant L2**: have no advanced cryptographic technologies.
- **Our L2**: Incorporates FHE, MPC, zk-SNARKs, and Monero-style ring signatures to protect both inputs and outputs comprehensively. Coldkey-Hotkey native support to enhance security.

2. **Economic | Funding and Investment**
- **Existant L2**: tokenomics do not keep AI in the focus.
- **Our L2**: tokenomics is focused to support AI.

3. **Legal | Regulatory and Operational Autonomy**
- **Existant L2**: usage will force us to share the legal and market risks.
- **Our L2**: independency ensure regulatory and operational stability.

4. **Economic | Transactions Efficiency and Scalability**
- **Criteria**: stable transaction costs and scalability.
- **Existant L2**: experience unpredictable transaction fees and scalability issues due to shared network resources. Nodes do not prioritize AI.
- **Our L2**: Offers predictable transaction costs and dedicated resources for AI-centric tasks.

5. **Strategic | Innovation and Market Leadership**
- **Criteria**: Leading the Market with our own platform.
- **Existant L2**: Competitors like BitTensor, CommuneAI, and Eternal-AI already have blockchains but may not offer sufficient innovation in AI functionalities.
- **Our L2**: Develops technologies that push the boundaries of AI and blockchain integration, leading the market with innovative solutions.

6. **Technical | Innovation | Data Handling and Storage | Decentralized Data Markets**
- **Criteria**: Efficient ML dataset management.
- **Existant L2**: relying on generic IPFS solutions that do not meet AI needs.
- **Our L2**: Provides solutions designed for AI Data-Ocean Marketplace.

7. **Technical | Task Scheduler & Adaptive Gas Fee Structures**
- **Criteria**: tasks scheduling and dynamic fee adjustments.
- **Existant L2**: lack dynamic task management and have rigid fee structures, leading to inefficiencies and elevated costs during peak demand.
- **Our L2**: Implements a native task scheduler and an adaptive fee model.

## Longer

1. **Technical Perspective | Advanced Security and Privacy Technologies**
- **Why Old L2 is Bad**: Existing platforms often lack the integration of advanced cryptographic technologies.
- **Why Our L2 is Good**: Incorporates FHE, MPC, zk-SNARKs, and Monero-style ring signatures to protect both inputs and outputs comprehensively. Coldkey-Hotkey native support to enhance security.
- **Going Deeper**: This ensures that all AI-related data processes remain confidential and secure against surveillance and unauthorized analysis, providing a robust environment for sensitive AI applications.

2. **Economic Perspective | Funding and Investment**
- **Why Old L2 is Bad**: Existing platforms tokenomics do not keep AI-projects incentivization in the focus..
- **Why Our L2 is Good**: Our BlockChain tokenomics will be focused to support AI development on the platform.

2. **Legal and Compliance Perspective | Regulatory and Operational Autonomy**
- **Why Old L2 is Bad**: Launching on third-party L2 blockchain will force us to share the legal and market risks with this L2 and other projects on that.
- **Why Our L2 is Good**: Independent L2 will ensure regulatory and operational stability.

3. **Economic Perspective | Transactions Efficiency and Scalability**
- **Criteria**: Maintaining stable transaction costs and efficient scalability.
- **Why Old L2 is Bad**: Competing platforms often experience unpredictable transaction fees and scalability issues due to shared network resources. Nodes do not prioritize AI.
- **Why Our L2 is Good**: Offers predictable transaction costs and dedicated resources for AI-centric tasks.

4. **Innovation and Market Leadership**
- **Criteria**: Introducing novel functionalities and setting industry standards.
- **Why Old L2 is Bad**: Competitors like BitTensor, CommuneAI, and Eternal-AI already have blockchains but may not offer sufficient innovation in AI functionalities.
- **Why Our L2 is Good**: Develops proprietary technologies that push the boundaries of AI and blockchain integration, leading the market with innovative solutions.
- **Going Deeper**: By offering unique features such as AI-specific smart contract languages and on-chain AI model training, the platform not only leads in technology but also influences future development trends.

5. **Strategic Perspective | Data Handling and Storage**
- **Criteria**: Efficient and secure data management for AI services.
- **Why Old L2 is Bad**: Existing L2 solutions relying on generic IPFS-based solutions that do not meet AI-specific needs.
- **Why Our L2 is Good**: Provides robust, scalable solutions designed specifically for complex AI data handling, from ingestion to processing and storage.
- **Going Deeper**: With optimized data structures and storage solutions, the platform ensures high performance and security, essential for effective AI deployment.

10. **Innovation and Research Perspective | Decentralized Data Markets**
- **Criteria**: Facilitating secure and compliant data exchange.
- **Why Old L2 is Bad**: Lack of secure marketplaces for data can hinder the development and deployment of AI models.
- **Why Our L2 is Good**: Creates a platform for buying and selling data in a secure, anonymized, and compliant manner, enabling vibrant data exchange ecosystems.
- **Going Deeper**: This marketplace not only supports the growth of AI services but also ensures that data transactions comply with privacy regulations, fostering trust and participation from a wide range of data providers and consumers.

6. **Technical Perspective | Decentralized AI Task Scheduler**
- **Criteria**: Efficient management and scheduling of AI tasks.
- **Why Old L2 is Bad**: Many platforms do not support decentralized task management, leading to inefficiencies and bottlenecks.
- **Why Our L2 is Good**: Implements a decentralized system to manage and schedule AI tasks, optimizing resource allocation and ensuring timely task execution.
- **Going Deeper**: This system helps in managing the computational load effectively, distributing tasks across the network without centralized bottlenecks.

7. **Technical Perspective | Adaptive Gas Fee Structures**
- **Criteria**: Dynamic adjustment of transaction fees based on network load and AI task complexity.
- **Why Old L2 is Bad**: Fixed fee structures on existing platforms can lead to high costs during peak times or for complex AI operations.
- **Why Our L2 is Good**: Utilizes an adaptive fee model that adjusts based on actual network conditions, ensuring fair and economic transaction costs.
- **Going Deeper**: This model supports a sustainable economic environment where AI developers and users can predict and control their operational costs better.

## Draft

8. **Node Diversity**
- **Criteria**: A diverse range of nodes to enhance network resilience and decentralization.
- **Why Old L2 is Bad**: Some platforms suffer from node centralization, which can lead to vulnerabilities and biases in AI computations.
- **Why Our L2 is Good**: Encourages a geographically and organizationally diverse node ecosystem, promoting robustness and unbiased AI operations.
- **Going Deeper**: This diversity ensures that the AI services are accessible globally and operate reliably, mitigating risks associated with centralized node control.

9. **Public and Private Hybrid Models**
- **Criteria**: Support for hybrid blockchain models blending public transparency with private execution.
- **Why Old L2 is Bad

**: Existing solutions often force a choice between public and private, limiting use case flexibility.
- **Why Our L2 is Good**: Offers the flexibility of hybrid models, allowing users to choose public or private execution based on their specific needs.
- **Going Deeper**: This approach enables secure, private AI operations while still benefiting from the transparency and security of public blockchains, ideal for diverse applications.

10. **Decentralized Data Markets**
- **Criteria**: Facilitating secure and compliant data exchange.
- **Why Old L2 is Bad**: Lack of secure marketplaces for data can hinder the development and deployment of AI models.
- **Why Our L2 is Good**: Creates a platform for buying and selling data in a secure, anonymized, and compliant manner, enabling vibrant data exchange ecosystems.
- **Going Deeper**: This marketplace not only supports the growth of AI services but also ensures that data transactions comply with privacy regulations, fostering trust and participation from a wide range of data providers and consumers.

This list reflects a structured approach to addressing key challenges in the decentralized AI blockchain space, emphasizing security, innovation, and efficiency while aiming to lead in market development and operational excellence.

зачем свой L2

вообще что такое L2 блокчейн - это не что-то сакральное и неприкасаемое
- это просто ноды + распределенный сторад + горстка контрактов

0) технология может быть реализована хоть на тамагочи и эксель таблицах
критерии
А - удобство создания платформы
Б - удобство бизнес менеджмента
В - удобство юзеров на платформе

Technical Perspective Criterias
- Customization: Does your project require specific features that only a custom L2 can provide?

- Smart Contract Capabilities: Are the smart contract functionalities of existing L2s sufficient for your project's needs?

coldkey - hotkey - as in TAO, its not supported by EVM

можно сказать что мы добавим в EVM инструкцию AI или DATASET и все

Economic Perspective
- Transaction Costs
- Resource Allocation
- Funding and Investment
- Revenue Models

Legal and Compliance Perspective
- Legal Risks
- Regulatory Compliance

Strategic Perspective
- Market Trends

Innovation and Research Perspective
- Research and Development Opportunities: What opportunities for innovation does building a new L2 offer?
- Knowledge Building: Consider the learning curve and knowledge acquisition in building a new L2.
- Long-Term Viability: Assess the sustainability and future-proofing of creating a new L2 versus using an existing one.

1) реальная причина - мы хотим сделать миллиардный проект
свой блокчейн это статусно

1.0) биттензор и comai и eternal-ai все они сделали свой БЧ чем мы хуже

1.1) у нас крупный проект критичный для человечества
мы не хотим быть завязаны рисками на БЧ
что если на говночейне будет какойто overload - gas will go up etc
what if they switch from PoS to PoW idk

1.2) нам удобнее управлять всеми правилами функционирования системы на нашем БЧ

2) ИИ-специфика
есть следующие роли
- L2 node
- Project relayer / validator
- Project ai miner

!!! Зачем нужен майнер и релеер отдельно
relayer
- receive request
- N miners
- N results
- X
-> blockchain

хочется чтобы исполнение запроса происходило вне блока - без задержек и быстро
и хочется чтобы L2 ноды выполняли роль Project-relayer, иначе будет energy-consumption
а именно ноды л2 для скорости и безопасности переиспользуются для сабнетов
для этого нам нужно адаптировать код ноды для эффективного переспользования их ресурсов в качестве безопасных релееров ИИ под-сетей

brainstorm:
если нужно сохранить модель приватно
и нужны доверенные узлы
https://vitalik.eth.limo/images/cryptoai/mpcnets.png

FHE + MPC
Secure multi-party computation

3) наш L2 также будет служить Data-Ocean
а именно поддерживать

4) давайте реально рассмотрим альтернативу

Смарт-Контракты на Метисе
люди разворачивают свои шаблоны, они подключаются к маркетплейсу
за запросы платят токенами ИИпроекта
часть токенов как комиссия уходит в маркетплейс
либо же платят и токенами ИИпроекта и токенами маркетплейса
- не нужен блокчейн
- у нас теперь участвует аж 3 токена так или иначе - МЕТИС, АИТХЕР, ИИ

- нет возможности кастомизировать код нод и правила блокчейна
- почему это важно?

Economic Perspective
- Transaction Costs
it is uncontrollable since there are thsousands of other project

- Resource Allocation
AI first!

- Funding and Investment
Metis tokenomics is not based for AI innovative projects incentivization

Legal and Compliance Perspective
- Legal Risks
some "bad" project could be stored on Metis
it will lead to Metis ban in Z country
our project cannot remove "bad" project by DAO voting and will be banned in Z country

Innovation and Research Perspective
- Long-Term Viability: Assess the sustainability and future-proofing of creating a new L2 versus using an existing one.

our AI project will be very dependent of the popularity and operation if the BC

НЕКОРРЕКТНАЯ АНАЛОГИЯ (как я люблю)
представьте что вы хотите создать Амазон, а вам говорят, идите на Ебей там ведь уже можно выставлять товары
поинт - наша технология слишком большая чтобы быть вписана в рамки существующего блокчейна и быть ограничена его возможностями и правилами которые мы не в силах будем изменить

## Draft

**Why We Need Our Own L2 Blockchain for Decentralized AI Services**

1. **Customization and Smart Contract Capabilities**
- **Criteria**: Custom features tailored specifically for decentralized AI, such as unique security mechanisms.
- **Existing L2 Shortcomings**: Lack of support for complex security features like coldkey-hotkey setups.
- **Our L2 Solution**: Our blockchain integrates coldkey-hotkey directly, enhancing security specifically for AI operations.
- **Deep Dive**: Coldkey-hotkey security mechanisms involve separating wallet keys that handle transaction approvals from those that can spend funds, reducing the risk of unauthorized transactions. This setup is vital for managing AI transactions securely, especially when handling large volumes or sensitive data.

2. **Regulatory and Operational Autonomy**
- **Criteria**: Swift adaptability to regulatory changes and maintaining operational stability independently.
- **Existing L2 Shortcomings**: Dependency on external platforms exposes operations to regulatory disruptions.
- **Our L2 Solution**: We are focused on providing AI services and do not bound with other risks.

3. **Economic Efficiency**
- **Criteria**: Stable transaction costs and dedicated resource allocation for AI-centric tasks.
- **Existing L2 Shortcomings**: Fluctuating transaction fees and resource competition on shared networks.
- **Our L2 Solution**: Predictable costs and prioritized AI processing optimize both cost and performance.
- **Deep Dive**: By controlling the blockchain infrastructure, we can optimize block sizes, transaction processing times, and fee structures to suit our specific needs, unlike existing L2 solutions which are optimized for general use cases and may not provide the best cost-efficiency for high-demand AI operations.

4. **Advanced Security and Privacy Technologies**
- **Criteria**: Integrating cryptographic technologies like FHE and MPC.
- **Existing L2 Shortcomings**: Lack of support.
- **Our L2 Solution**: Embedding advanced cryptographic techniques to ensure data integrity.
- **Deep Dive**: Implementing FHE allows computation on encrypted data, ensuring that AI operations remain confidential and tamper-proof. MPC facilitates data sharing among multiple parties without exposing the underlying data, crucial for collaborative AI without compromising privacy.

5. **Innovation and Market Leadership**
- **Criteria**: Driving innovation and establishing a distinct market presence.
- **Existing L2 Shortcomings**: Standardized frameworks limit differentiation.
- **Our L2 Solution**: Facilitating new AI integrations to set industry standards.
- **Deep Dive**: Developing proprietary blockchain solutions allows us to introduce novel functionalities not yet available on existing L2s, such as specialized AI orchestration layers or AI-specific opcode that can directly execute AI algorithms on the blockchain, pushing forward the frontier of what's possible in integrating AI with blockchain technology.

6. **Legal and Compliance Transparency**
- **Criteria**: Establishing transparent and non-censored AI operations.
- **Existing L2 Shortcomings**: Fixed compliance measures restrict transparent AI operations.
- **Our L2 Solution**: Customizable compliance frameworks ensure adherence to global standards.
- **Deep Dive**: Our blockchain will enable the implementation of jurisdiction-specific rules and auditing protocols that are transparent and modifiable by the community, fostering trust and ensuring compliance with international data protection regulations.

7. **User-Centric Design**
- **Criteria**: User-friendly interface for decentralized AI interactions.
- **Existing L2 Shortcomings**: Interfaces not optimized for end-user AI interactions.
- **Our L2 Solution**: Prioritizing intuitive user experiences for broader accessibility.
- **Deep Dive**: Developing a dedicated UX/UI framework that supports both novice and advanced users, integrating features like real-time AI performance monitoring, easy access to AI services, and interactive tutorials to guide users in managing AI-driven tasks.

8. **Scalability and Flexibility**
- **Criteria**: Efficiently scaling operations and adapting to demand without performance loss.
- **Existing L2 Shortcomings**: Rigid architectural designs limit scalability.
- **Our L2 Solution**: Building a scalable architecture for flexible adjustments and expansions.
- **Deep Dive**: Our blockchain will employ modular design principles, allowing for components such as consensus algorithms and data storage methods to be swapped or upgraded as needed without disrupting existing operations, thereby accommodating growth and technological advancements.

11. **Data Handling and Storage**
- **Criteria**: Efficient and secure data management tailored for AI applications.
- **Existing L2 Shortcomings**: No data handling capabilities for complex AI tasks. Only IPFS integrations.
- **Our L2 Solution**: Specialized data storage and retrieval systems designed for AI.

Quantum Resistance: Preparing for post-quantum cryptographic standards.
Enhanced Privacy Features: Additional privacy options beyond standard encryption.

**Enhanced Transaction Privacy Using zk-SNARKs and Ring Signatures**
- **Criteria**: Ensuring absolute privacy for both prompts to and outputs from decentralized AI services.
- **Current L2 Problems**: Existing L2 solutions typically fail to adequately protect the confidentiality of both input prompts and AI-generated outputs, leaving them vulnerable to interception and analysis.
- **Our L2 Solution**: Integration of zk-SNARKs for prompt confidentiality and Monero-style ring signatures for output anonymity. This combination ensures that both the details of the queries and the nature of the responses remain private.
- **Deep Dive**: By implementing zk-SNARKs, we enable users to submit queries to AI services without revealing the content of their requests. The system verifies the validity of the query without needing to know its specifics, thereby maintaining the confidentiality of the prompt. On the output side, ring signatures are used to anonymize the transactions related to the AI responses. This means that while the AI can send results back to the user, the transaction does not reveal who received the information, similar to how Monero obscures the sender in a transaction. This dual approach not only protects the data integrity but also ensures that both the input and output of the decentralized AI interactions are shielded from any potential surveillance or analysis, thus enhancing user privacy and system security.

AI-Developer Incentives: Encouraging ecosystem growth through developer rewards.

Adaptive Gas Fee Structures: Implementing dynamic fee adjustments based on network load and transaction type.

Node Diversity: Encouraging a diverse range of nodes to enhance network resilience and decentralization.
TODO explain - AI should be accessable everywhere nodes should be diversed

Public and Private Hybrid Models: Support for hybrid blockchain models that blend public transparency with private execution.

Peer-to-Peer (P2P) Messaging Protocols: Secure direct communication channels between nodes.

State Channels for AI Computation: Off-chain state channels to handle complex AI computations.

Conditional Transactions Based on AI Output: Transactions that are executed based on AI model outputs.

Distributed File System Integration: Seamless integration with distributed file systems for efficient data storage and retrieval.

Data-Ocean

Robust Metadata Handling: Detailed metadata management to enhance the traceability and usability of stored data.

Zero-Downtime Networks: Designing the blockchain to achieve zero downtime through robust redundancy and failover strategies.

Data Streaming Capabilities: Real-time data streaming support for dynamic AI applications

Decentralized AI Task Scheduler: A decentralized system for scheduling and managing AI tasks efficiently across the network.

Blockchain-Enforced SLAs: Smart contracts that enforce service level agreements for AI services provided over the blockchain.

Participant Reputation Systems: Implementing systems to track and evaluate the reputation of participants within the blockchain ecosystem.

Decentralized Data Markets: Creating platforms for buying and selling data in a secure, anonymized, and compliant manner.