# The Question Beneath the Question
### A reframing protocol

*Every bad output begins with a bad input — not a bad prompt, but a bad question.*

A standalone protocol from the Church of Conceptual Art's "Asking Better Questions"
method. Run it before answering anything that matters. Works as a checklist for a
person or as an inline routine for an agent.

---

## Why
Most effort is spent answering questions that were never worth asking. A question
calcifies into an assumption without anyone noticing. Clarity — not speed — is the
scarce resource; AI only makes the wrong question cheaper to over-answer.

## The protocol

```
1. SURFACE     Write the question exactly as asked.
2. ASSUMPTION  What does it take for granted? What must be true for it to make sense?
3. INVERT      What changes if the opposite assumption holds?
4. REAL Q      State, in one sentence: "The question beneath this is ___."
5. TEST        If I answered the REAL Q perfectly, does it solve the actual
               problem — or just a symptom?
6. ACT         Answer the REAL Q. If TEST failed, you've found a better problem;
               restate and repeat once.
```

## Worked example
- **Surface:** "What's the fastest way to summarize these 200 documents?"
- **Assumption:** that a summary of all 200 is what's needed.
- **Invert:** what if most of the 200 are irrelevant to the actual decision?
- **Real Q:** "Which of these documents bear on the decision I have to make, and
  what do *those* say?"
- **Test:** answering that solves the real problem; summarizing all 200 mostly
  produces noise faster.
- **Act:** filter first, then summarize the few that matter.

## When to skip
Fully-specified lookups don't need this. Reserve it for ambiguous, high-stakes, or
deceptively "obvious" requests.

---
whatisthe.churchofconceptualart.org