After diagnosis comes design: changing the system so it produces better outcomes. Good design is practical. It specifies new rules, roles, or tools that prevent the problem from recurring.
Improvements often involve clearer decision rights, better metrics, tighter checklists, or different people in different seats. Sometimes the right change is automation; sometimes it is a training loop; sometimes it is removing a step that creates confusion.
Design must consider trade-offs. Fixing one problem can create another if incentives shift or complexity increases. So the design should be tested against reality: run it, observe results, adjust.
This is where principles become operational. Values like honesty and learning turn into protocols: how debates run, how decisions are recorded, how errors are tracked until fixed.