Vitalik Buterin introduced an approach where the blockchain gradually moves away from the need to repeatedly perform the same operations on different nodes. According to him, the development of cryptography allows for the separation of computation execution and its verification: one participant processes the task and forms proof of correctness, while the other nodes verify the already compact cryptographic result. This principle relies on SNARK and STARK technologies and is intended to simultaneously reduce computational load and expand programmable privacy capabilities. While early blockchains mainly solved the task of confirming the right to manage assets, the next architecture may allow programs themselves to determine which data is accessible to various participants. Buterin places particular importance on recursive STARKs, which allow for the aggregation of a large number of proofs into one, thereby reducing the amount of information transmitted and processed by the network's base layer.
This idea is associated with changing the entire computational process structure of Ethereum. Among the considered directions are preliminary aggregation of signatures and proofs, separation of transactions into individual actions and dependencies, as well as the ability to perform independent operations in parallel. An additional element of the architecture is distributed block formation, reducing dependency on a single participant when determining their content. In a broader view, the blockchain transforms from a single execution layer into a multi-stage computational pipeline. This can include operations on the user's device, mempools with various privacy parameters, proof aggregation systems, second-layer solutions, and the network's base layer. Part of the complex computations in such a model is performed outside the main execution, and the blockchain receives compact confirmation of their correctness. This should allow the system to scale without proportionally increasing the workload of each node.
Artificial intelligence occupies a special place in Buterin's concept. In his assessment, the complexity of cryptographic protocols makes manual development, testing, and code verification increasingly labor-intensive tasks. AI can be used for program preparation, implementation analysis, specification creation, error detection, and formal verification, where code compliance with specified requirements is checked by mathematical methods. Several independent specifications and automated verification can further reduce the likelihood of errors when creating complex systems. According to this logic, AI will be able to work not only with individual code fragments but also interact with infrastructure from L1, L2, and intermediate services, selecting the necessary tools for performing specific operations. As a result, the development of cryptography and AI is seen as part of a unified transition to blockchains where computations are performed more distributedly, and their correctness is confirmed by compact proofs.
Related articles
