Token for Image Tokenizer (TiTok) & Onchain Storage, Arweave&AO

On June 15, 2024, Ethereum co-founder Vitalik Buterin posted a warpcast endorsing TiTok as an innovative solution for future blockchain storage. TiTok, not to be confused with TikTok, stands for Token for Image Tokenizer compression. It leverages minimal data to generate high-quality images, addressing the blockchain’s space constraints and opening new approaches in AI integration and diverse storage scenarios. No wonder Vitalik is so enthusiastic and even envisioned its integration into Farcaster.

Challenges of On-chain Storage

Blockchain’s on-chain storage has long faced challenges due to limited block space. One development trend is “on-chain computation, off-chain storage”. As EIP-4844 pushed Ethereum into the Blob era, all L2 transaction data will be stored in Blobs, relieving the main network’s computational burden. According to the ultimate vision, this part of the data will no longer participate in the main network’s consensus mechanism, thereby reducing the computational pressure on Ethereum.

Recently, mysterious transactions sent substantial Bitcoin to Satoshi Nakamoto’s address, sparking discussions and expectations. Later, an anonymous billionaire spent 1.5 BTC, approximately $66,000, to inscribe 8.93 MB of garbled data onto the Bitcoin mainnet, highlighting the demand for dedicated on-chain storage networks.

In this context, dedicated on-chain storage networks have emerged. For example, Ethstorage provides professional Blob data storage to address Ethereum’s storage challenges. On a broader scale, Arweave and Filecoin have become the two giants in the storage sector.

Filecoin, originating from IPFS, distinguishes itself with a token-incentivized system rewarding users for content uploads with $FIL. Its recent compatibility with Ethereum’s Virtual Machine (EVM) and the introduction of Filecoin VM (FVM) are transforming storage operations. The introduction of smart contracts will fundamentally change the operational logic of Filecoin’s on-chain storage:

Compared to Web 2 storage solutions like AWS, it offers the advantage of flexible data set operation permissions. FVM (Filecoin Virtual Machine) will enable Web 3 data access customization. Its smart contracts will change current access and control permissions while still ensuring permanent data storage.

With highly customizable data access, data DAOs will be able to tokenize data after defining permissions and use DAOs to manage and allocate funds.

Before FVM, data persistence required manual renewal before the expiration date, and due to various reasons, it was practically effective for only one to two years. However, after FVM, users can use smart contracts to achieve automatic renewal, thereby extending the period.

Smart contracts will make it easier to assess the service levels of different node service providers. Based on the FIL token, an on-chain lending market can be created to improve overall network storage efficiency.

Leveraging EVM compatibility, tokens that comply with the ERC-20 standard can be bridged to the Filecoin main network. Assets on it can also cross-chain to other asset pools, ultimately activating the on-chain storage market.

Once FVM is fully implemented, Filecoin will effectively become the most fundamental development infrastructure on-chain. Its full EVM compatibility will allow direct interaction with any mainstream asset. There will be no need for separate storage settings, but rather, developers can directly deploy any required modules from Ethereum’s development logic using smart contracts.

Arweave offers “pay once, store forever” through its endowment model, ensuring over 200 years of data longevity via treasury and linear funding mechanisms. Recently, Arweave introduced AO infrastructure, aiming to integrate AR, AI, and on-chain/off-chain computation and storage, potentially revolutionizing decentralized computing.

Arweave’s New Era: AO

AO’s framework can be divided into three parts: distributed architecture, parallel computing capabilities, and communication scheduling components. Unifying the three is a complete supercomputer function.

For a distributed architecture, AO uses a single-system image (SSI) to organize the decentralized system of the AO network.

For parallel computing capabilities, AO uses the Actor Model theory, which handles high-concurrency environments and effectively integrates blockchain-related knowledge. The term AO comes from Actor Orient (analogous to OOP’s object-oriented).

For scheduling units, AO has designed three key parts: Messenger Units (MUs) for information transmission, Scheduler Units (SUs) for process scheduling, and Compute Units (CUs) for computing tasks. Among these, MUs and SUs are the most important, while CUs handle the final parallel computations.

AO’s computing power combined with Arweave’s permanent storage capability creates a global, high-concurrency server that is censorship-resistant and highly applicable. Any type of dApp, regardless of language or public chain, can connect to Arweave, enjoying cheaper yet more efficient public chain services than Ethereum.

A major focus of AO’s development is its integration with AI. Specifically for Arweave, its unique permanent storage feature can be further optimized and expanded through AI technology. For example, AI can be used for data analysis and processing, improving data retrieval efficiency and accuracy. Additionally, AI can be used for developing and executing smart contracts, ensuring transaction security and transparency.

Currently, AO has already achieved Web Assembly 64-bit support, allowing developers to create applications that use more than 4GB of memory, significantly enhancing computational power. Secondly, AO now supports WeaveDrive, enabling developers to access remote data stored on Arweave as if using a local hard drive and efficiently transfer data to the execution environment.

Lastly, AO has kept pace with large language model inference engines. For instance, it has ported the Llama.cpp system, supporting the direct operation of various open-source large language models in smart contracts. An example of this capability is the creation of a simulator named Llama Fed, using the Llama 3 model to implement a fully autonomous, trustless monetary policy system, showcasing the potential and use cases of the new technology.

Conclusion

From on-chain to off-chain storage, major blockchains like Ethereum are evolving to manage expanding data states. As blockchain acts as a state machine, the migration of data to off-chain solutions becomes imperative, making integrated solutions bridging on-chain and off-chain functionalities necessary.