Observe the moment research stalls and you find something in common. Data are accumulating. Analysis is progressing. And yet what you want to show stays vague — the question is absent. Stalling is caused not by ability or by a lack of data, but by the absence of a question.
This is different from the data-first strategy described in Chapter 02 (searching data for regularities). That strategy has a question — "what regularity can be read from these data?" — and a design for collecting unbiased data to answer it. The failure here is having only data, with no question. Praying that meaning will emerge afterward is not design.
Idea → Framing → Figures → Draft
Framing is the key. Everything downstream — which figures to make, what to write — follows from Framing. Framing has two stages. This chapter covers the first half, Framing-A: question design; the next chapter covers the second half, Framing-B: claim design.
The Paper Card's Framing-A consists of 2 Steps and 5 outputs.
Do not write results here. Do not write methods either. Framing-A is the step that fixes only "where you stand and what you ask." The following explains each in turn.
Questions do not arise in a vacuum. Only once you know what the field currently presupposes can you say whether your question updates it, extends its range, or refutes it.
What you collect is literature for positioning, not for completeness. Grasp the field's beliefs from the major review papers; grasp what types of questions the field asks, and its implicit premises, from five or more representative papers of the last 3–5 years. Keeping the question archetypes (Appendix 03-A) in mind here means you will not hesitate when choosing your own archetype in the next Step. The practicalities of finding and managing literature are summarized at the end of Appendix 03-A.
Compress what you read in Step 1 into 1–3 lines. "On this problem, the field currently understands things this way." Not your opinion — the field's current position. If you cannot write this, Step 1 is not done.
Put prior studies into a table. The columns: prior study / their claim / hidden assumption / limitation / our update.
| Prior study | Their claim | Hidden assumption | Limitation | Our update | |
|---|---|---|---|---|---|
| 1 | Suzuki et al. 2022 | Effective fracture surface area can be estimated by thermal tracer inversion | Estimation in simple systems generalizes to complex ones | Verified mainly in simple systems | Verify in branching multi-path systems |
The criterion is not how many papers you read, but whether you can say "which premise I update." Once the table fills in, the positioning sentence emerges.
"Compared to X, we Y." — Whereas Suzuki et al. (2022) verified in simple systems, we verify in branching multi-path systems.
This is not a question. It is a one-sentence positioning of where you stand. The Core Question, next, is raised from this position.
Only now do you write the question. Four conditions.
Choose the archetype. A good research question always fits an "archetype."
| Archetype | Form of the question |
|---|---|
| Estimation | How far can X be estimated from Y? |
| Mechanism | What controls X? |
| Validity | Under what conditions is the model or assumption valid? |
| Uncertainty | How sensitive is the prediction to uncertainty? |
| Scale | How does micro affect macro? |
| Dynamics | How does X change over time? |
| Design | How should we design to achieve X? |
If you cannot choose the archetype of your question, the question is still vague. Check the correspondence with Chapter 02 as well. What decides the archetype is what the claim commits to. Commit to explanation → mechanism; to prediction → uncertainty or dynamics; to control → design. Estimation, validity, and scale appear under any commitment. The entry strategy (hypothesis-first or data-first), on the other hand, does not decide the archetype. What the entry decides is "how you answer that question": for the same estimation-type question, whether you set a model and invert it or learn a mapping from data changes the data you need and the form of the evidence. Only validity and uncertainty presuppose the existence of a model, so hypothesis-first is natural for them.
Decide the elements the archetype requires. Each archetype has elements that must be decided for the question to stand (Appendix 03-A). For estimation: what is X (the thing to estimate), what is Y (the observable), is it identifiable, how is uncertainty handled. Until these are decided, the question does not become one sentence.
Make it one sentence.
✕"Can thermal tracer inversion estimate fracture surface area?" — Ends in Yes / No. No conditions
○"Under what conditions does thermal tracer inversion recover effective fracture surface area in branching multi-path systems?" — What (X), from what (Y), in which system, under which conditions — all in one sentence
If you cannot state it, either the archetype or the elements are not yet decided.
One or two lines. Write what changes if this question is answered. For the field, for applications, or for society. It should emerge naturally from the Field Belief and the Comparison Table; if it does not, the positioning is weak.
Put the Previous belief (the understanding the field implicitly presupposes) side by side with the Updated belief (the new understanding this study shows). This is the "Hidden assumption" and "Our update" of the Comparison Table, raised to the level of the field's belief. This becomes the seed of the Key Contributions in Framing-B.
The "elements the archetype requires," decided in Step 2-3, decide the data you need. For estimation, once X, Y, and identifiability are decided, which observations to take under which conditions follows by working backward. In research as design, the order is this.
✕Data → pray that meaning emerges ○Question → Required Data → Method
Framing-A is the second tier of the Paper Card. Fill in the five outputs, receive the "Next Judge" at Team Meeting, and you can move on to Framing-B in the next chapter. → Lab Tools "Paper Card"