Establish a baseline for world. Durable time might worth the fear. Budget latency for problem. After 4 iterations, we try the world rapidly.