Breakthroughs coming to attestation broadcast too. Three principles: shaving off wire bytes, reducing latencies, increasing information efficiency per committee. One Hegotá proposal that made several Core Devs happy is Batching Attestations at Source, authored by yours truly, in collaboration with @nero_eth and @mkalinin2.

Privacy should be the universal default, not a cypherpunk flag. Every on-chain read and write leaks metadata to multiple observers: ISP, DNS, RPC, indexer, bundler, chain. Hermetic is a modular playground for access-layer privacy. Still very alpha. A self-contained local binary that stacks privacy-preserving layers and exposes sound solutions, with stated tradeoffs, over simple APIs. Today: - Railgun shielding, unshielding, queries. - Every egress request over an isolated Tor circuit, including DNS (next: DoH). - Rust host + embedded Tor via Arti, no separate sidecar. How: - SDK code runs sandboxed in @deno_land, stripped of most permissions. - No fetch, no node:net, no host writes. Modules are cut off from the world. Untrusted. - Every SDK network call crosses the Deno boundary, Hermetic intercepts and routes via Tor. No escape possible. Doesn't yet defend against broadcaster trust, query-pattern inference, timing leaks, or mempool exposure. Exploring: - Aztec and other on-chain privacy backends. - Wasm modules, light clients, messaging. - Account isolation. - Local, agent-friendly APIs. Would appreciate help. DM if interested!

non-headliner proposal period for Hegotá is open. we're proposing at least 4 EIPs for the networking stack: 1. ethp2p erasure-coded broadcast for exec payloads => supercharged RS, speeding up p95 delivery of 2MiB payloads by 6x (9s -> 1.5s), at half the bandwidth. 2. attestation broadcast redesign => targeting sub-500ms worldwide propagation. 3. batched attestations at origin => operators can send one aggregate for validators scheduled on the same committee, without consolidating balance in advance. 4. ethp2p/libp2p coexistence => plan for gradually migrating to Ethereum's new networking stack, and making QUIC compulsory (looking at you, Teku, Nimbus, and Lodestar 👀) together with cell-level deltas and sparse blobpool, these changes free up bandwidth while making everything faster. on the execution/data front, this capacity can be reallocated to increase gas limits, scale blobs, carry optional zkEVM proofs, and transmit BALs sooner for exec parallelization. on the consensus front, we can consolidate subnets and reduce the committee count. because we hear from more validators sooner, we can shorten epoch durations and speed up finality to 16, 8, or even 4 slots. things are looking gud for Ethereum 🫡