Debugging and Assumptions

The first useful debugging question is rarely "what broke?" It is usually "what am I assuming to be true?"

When a system feels unpredictable, there is often an invisible contract being violated:

• a response shape changed
• state was stale
• retries duplicated work
• a background job ran later than expected

Good debugging is a process of making assumptions explicit, then removing them one by one.

A reliable loop

1. Reproduce the bug with the smallest surface area possible.
2. Inspect boundaries: input, output, timing, persistence.
3. Validate every assumption with evidence, not memory.
4. Keep notes while debugging so the path is reusable.