Eric Winer

Here’s a high-level stack overview of Shade Agents that @proximityfi has been working on. It is an extremely powerful design because with Chain Signatures, a decentralized group of worker agents can control the same key (derived from the smart contract) across various blockchains, so there’s no risk of key loss or compromise from using a single TEE. Ultimately, it enables AI agents to execute, transact, and manage assets but in a fully decentralized and trustless manner, which has been a real technical challenge for our space. For a dev, that means you can build a fully autonomous dapp for prediction markets, lending optimizers, trading agents, whatever. For an end user, that means you have verifiable and private automation without worrying about the risks associated with centralized solutions. Handy at user level and will also be crucial for large-scale autonomous businesses like an AI-driven hedge fund.

Trying to compete with @ilblackdragon for attention is hard but here I am. @NEARProtocol mainnet just finished upgrading from 6 shards to 8 shards today. Many have talked about the significance in terms of 33% capacity improvement etc and here I want to share some insights from a more technical perspective on why I think this is an important upgrade: - Even though we did a few shardings before, we never did it in a way where resharding happens within one block. Past reshardings rely on validator nodes building the entire state of new shards from scratch, a process that may take an hour to finish. It is no longer feasible after Nightshade 2.0 launch when the protocol switched to stateless validation. The state witness for such a state transition would be humongous and even if it can be distributed to validators, they need to completely rebuild the shard themselves to validate, which would mean that we get stuck for over an hour for one block. Instead, resharding now works by splitting the state trie of a shard along the boundary and in O(logn) time computes two new state roots. The state witness for this state transition is also O(logn) which is small enough for validators to validate. All of this happens in one block time. It is significant because it brings us much closer to dynamic resharding where a shard would automatically split on demand within a very short amount of time. - We did two shard splits in this release. While it sounds insignificant, it allows us to test splitting of multiple shards to quickly increase the number of shards on mainnet. If we can only split one shard at a time, then it would take a while before we get to 50 shards on mainnet Overall it was a great release and everything went smoothly. The team did a lot of hard work and I am excited to see more shards appearing on mainnet in the coming months!