real-time proving Q&A :) What is it? ELI12. A: An army of nerds just cracked a sci-fi problem: proving every Ethereum mainnet block in real time. These zk proofs are exponentially cheaper to verify than re-executing transactions—just a few milliseconds, no matter the gas used. Thicc blocks, tiny validators. Decentralisation maxis rejoice. Wasn't real-time proving impossible moon math? A: It was. Five years ago we were five orders of magnitude away. But the tech kept compounding 10x/year, and now we're here. Don't fade exponentials. Also, on exponentials: → don't fade quantum computers breaking ECDSA → don't fade halvings breaking Bitcoin security How does Ethereum L1 become a "based and native rollup"? A: Let's break it down. → based: L1 proposers order transactions. Ethereum L1 is the base layer—it's tautologically based. → native: L1 executes transactions via the enshrined EVM state transition function. By definition Ethereum L1 is native. Always will be. → rollup: A rollup has onchain data and offchain execution. By snarking mainnet EVM blocks Ethereum L1 turns into a zk rollup. Rollups unlock big gas limits. Base is already at 35 megagas/sec—20x Ethereum L1. Base is aiming for 1 gigagas/sec, Ethereum L1 can get there too. Wen gigagas L1? A: Ethproofs call #3 will be dedicated to gigagas L1—do tune in. With EIP-9698 Dankrad proposes ~3x/year gas bumps, or ~10x every 2 years. That's 1,000x in 6 years. Respectable—but still rookie numbers. My optimistic take? Now that zkVMs match EVM performance gas limits can track zkVM improvements. If we keep hitting 10x/year that's 1 gigagas/sec in 3 years. I worked with Dankrad for over a decade—reality usually lands between his caution and my hopium. Aren't zkVM buggy? What about formal verification? A: Formal verification is the endgame—not step one. The short-term strategy: zkVM diversity. By end of Q2 a handful of zkVMs will hit real-time proving. Just like CL and EL clients, any individual client is assumed buggy. Security comes from diversity. Can low-latency proving scale to gigagas? A: Yes. SNARK proving is massively parallelisable. Only the initial x86 execution and chunking is sequencial. EIP-7825 proposes to cap the maximum amount of gas used per transaction. This allows for easy chunking at transaction boundaries. Parallelism scales. Latency stays low. What's the next zkVM bottleneck? Power. Credible 1-of-n prover liveness means home proving should be accessible to enthusiasts. A reasonable target may be 10kW or less per proving cluster. How do we grow the gas limit 1,000x on a 10kW budget? → software: proof systems, circuits, algos keep compounding → hardware: Moore's law + SNARK ASICs → distributed proving: spread the load, trustlessly
68,75K