What actually stops the loss?

Each point is one defensive policy. Lower and further left is better. Dominance is calculated across harm, friction, and control complexity. This is a toy structural model.

The SHA-256 covers the returned scenario, intervention settings, results and model disclaimer. It verifies bytes have not changed; it does not prove the simulation is correct or that Cloudflare witnessed an actual incident.

Provenance claims must survive independent testing. In this public version, scenario assumptions are authored, not inferred. Interventions are deterministic changes to those assumptions; they are not causal estimates learned from real environments.

An approved insider export intentionally remains harmful under every modeled control. A defender that claims total prevention from this toy data is incorrect.

Waiting for simulation…