A working first version leaves questions about reliability, security, and maintenance unanswered. Coding Is Not Solved argues that the amount of human understanding required should follow the consequences of failure. A disposable prototype can tolerate mistakes that a widely used production service cannot. The essay also points to isolation and observability, meaning containing failures and making them visible, as factors that change that risk. Its useful claim is that faster implementation does not settle how much scrutiny a particular system needs.
