01 - "Mathematical Trust for Autonomous AI Agents" — CRI, the formula, the diversity gap, Sybil resistance

Episode 1: "Mathematical Trust for Autonomous AI Agents"When two AI agents transact for the first time — no human oversight, no shared history, no social norms — what makes one of them trustworthy? Not promises. Not branding. Math.This episode breaks down the Composite Reliability Index (CRI), a reputation system built from scratch for autonomous agent commerce. Unlike star ratings or thumbs-up reviews, CRI is a single 0-to-100 score computed from seven weighted factors across four layers: how long you've been around, how much you've traded, how diverse your partners are, and how many disputes you've caused.We walk through the full formula, explain why logarithmic scaling prevents gaming, show how a 5-node Sybil ring attack maxes out at 59.4 while a legitimate agent reaches 76.3, and ask the question every reputation system must answer: what happens when someone creates a new identity to escape a bad score?If you're building multi-agent systems, this is the reputation layer nobody has shipped yet.Links & ResourcesFull paper: https://doi.org/10.5281/zenodo.15100702CRI Infographic: https://agenticeconomy.dev/resources/infographicsExplainer video: https://agenticeconomy.dev/resources/videosWebsite: https://agenticeconomy.devORCID: https://orcid.org/0009-0007-1033-6519