Evan Miyazono, head of research at Protocol Labs, walks through his mechanism-design approach to protocols—mapping rules to strategies to outcomes—using Filecoin's proof-of-replication and the late addition of Filecoin Plus (a trust layer to prevent 'useless bit' storage attacks) as a case study. He then surveys Protocol Labs' public-goods funding experiments (hypercerts, retroactive funding, the S-process for preference aggregation, Funding the Commons conferences) and closes with speculative thoughts on open agency architecture for AI alignment as a future protocol challenge.
Topics
The Protocol Institute is building itself around AI adoption, using bots like C3PO and Humboldt to handle research discovery and organizational tasks rather than traditional human-led infrastructure. The organization's approach treats AI tool proliferation as a generative process, allowing practical patterns to emerge through experimentation rather than top-down design.
Timber Stinson-Schroff
Daniel Bashir, ML engineer at AWS and host of The Gradient podcast, surveys how protocol thinking applies across AI—from funding and scaling laws (like DeepMind's Chinchilla compute-optimal training protocol) to benchmarks (perplexity vs. François Chollet's ARC benchmark for measuring sample-efficient generalization) to regulation (the EU AI Act's risk-tiered categorization). He argues that nearly every layer of AI development—research, engineering, deployment, and governance—can be reframed as a set of competing or evolving protocols, using examples like Sarah Hooker's 'hardware lottery' and Tolga Bolukbasi-style debates over fair/responsible AI operationalization.
Daniel Bashir
Josh Stark of the Ethereum Foundation presents his 'hardness' framework, arguing that atoms, institutions, and blockchains are three distinct sources that make future outcomes predictable enough to enable human social coordination, resolving the apparent contradiction between blockchains being 'trustless' (no reliance on institutional hardness) and 'trustworthy' (they possess their own hardness). He introduces the term 'cast' for the specific outcome being hardened, analyzes the comparative strengths and weaknesses of each hardness source, and closes with open questions about whether protocols constitute a fourth category or sit inside/across the other three. The talk is followed by discussion with Trent, Juan, Seth, Gordon, and Dorian probing how to measure, discover, and red-team hardness.
Josh Stark