In Framing-A, the question became one sentence. But answering the question alone does not make a paper. A paper is the form in which you say "we showed this" and others can question it. Framing-B is the step that designs that "we showed this" — before the results are in.
Here, too, you do not write results. It is a conditional design: "if the results come out this way, we claim this."
The Paper Card's Framing-B consists of three Steps.
Order matters. Write the story once the contributions are fixed; draw the structure once the story is written. Go the other way, and you end up inflating the contributions to fit the story.
Four rules.
Why start with "We show that…". "We propose a new method" is not a contribution. Proposing alone says nothing about what changed. A contribution states up front what is shown, in the form "We show that ○○." The method is the means of showing it, not the contribution itself.
Why three. One makes a thin paper; with four or more, readers cannot remember them. If you exceed three, either the claims are not focused or you have two papers' worth. That so many papers have three contributions is not convention; it has reasons in both readers' memory and logical structure (Notion: "Why Do Many Papers Have Three Contributions?").
Strong and weak contributions.
✕"We propose a machine-learning framework for reservoir modeling." — A description of the method. Nothing about what was shown
✕"We show that our method achieves significant improvement." — Vague words. No what, no how much
○"We show that permeability distributions can be recovered from tracer responses with less than 20% error in branching multi-path systems." — What, under which conditions, how far — all present
Contribution Check:
When writing Key Contributions, the distinctions of Chapter 02 apply directly.
Entry strategy (Chapter 02 I)
| Entry | Form of the claim | Form of the evidence |
|---|---|---|
| Hypothesis-first (model-driven) | We show that the proposed law holds for … / breaks down at … | Agreement between prediction and observation. The boundary of the range of validity |
| Data-first (data-driven) | We show that the data reveal … / that the index predicts … | Distribution and unbiasedness of the data. Reproducibility of the regularity found |
Kind of inquiry (Chapter 02 II)
| Kind | Where the claim reaches |
|---|---|
| Passive inquiry (understanding) | The claim may end at "understanding" |
| Active inquiry (design) | The claim must reach "usable for design." The evidence is the effect when used for design |
Commitment (Chapter 02 II)
| What the claim commits to | The content of "We show that…" |
|---|---|
| Description | … was observed |
| Explanation | … controls … (after ruling out alternative explanations) |
| Prediction | … predicts … even under unseen conditions |
| Control | manipulating … changes … |
Claims that straddle entries — writing an index found data-first as if it had been generalized hypothesis-first — are what reviewers attack most. If you were data-first in your master's and move to hypothesis-first in your PhD, switch all three: the question (what regularity is there → how far does this law hold), the evidence (unbiased data → prediction matching observation), and the form of the claim (reveal → holds / breaks down). Without the switch, you make hypothesis-first claims on data-first evidence, and the logic does not close.
Likewise, claims that exceed their commitment — writing "controls" when you have shown only correlation — get attacked. Look at what kind of evidence you have, and choose the verb in "We show that…".
Once the contributions are fixed, write the skeleton of the paper's Introduction in seven sentences. This is a draft of the abstract and the blueprint of the Introduction.
The outputs of Framing-A go straight into 2, 3, 4, and 6, and Contribution 1 from Step 1 goes into 5. The Narrative Skeleton, in other words, is the step that stitches Framing-A and Framing-B into a single story. If the sentences do not connect here, something is misaligned between Framing-A and the Contributions.
Make "Here we show" match Contribution 1. The most-read sentence in a paper is the "Here we show that…" at the end of the Introduction. If it says something different from Contribution 1, readers cannot tell which is the paper's claim.
If you have two or more contributions, draw how they depend on one another. This is not a workflow diagram. It is a claim-structure map.
How to draw: boxes and arrows for the dependencies. Hand-drawn is fine.
Serial structure:
[ C1: foundational claim ]
↓
[ C2: extension that depends on C1 ]
↓
[ C3: final implication ]
Parallel structure:
[ C1 ] [ C2 ]
\ /
↓ ↓
[ C3: integrating claim ]
Collapse test: which box, if removed, makes the paper collapse? The box that collapses it is the core claim, and it should be Contribution 1. A box whose removal changes nothing may be a supplement, not a contribution.
Provability check: can each box be shown by a figure? A box that cannot is a wish, not a contribution.
Figure 0 does not appear in the paper. But it becomes the blueprint for deciding "which figures are needed" in Figures-A (the next stage). One box corresponds, roughly, to one main figure.
When Framing-B is done, you should be able to say "if this figure comes out, the claim stands." For each box of Figure 0, check whether you can name the form of the figure that supports it.
If you cannot, the claim is not yet designed. Experiments and data collection are done in order to make these figures. Main figures: one figure, one claim — three Contributions means roughly three main figures.
But a paper's figures are not only main figures. Figures that show the object or the apparatus, and figures that show the validity of the method, do not carry claims, but readers need them to evaluate the claims. Only main figures correspond to boxes in Figure 0; the roles and design of the other figures are covered in Lab Forum Lectures "Writing Papers — Figures".
Framing-B is the third tier of the Paper Card. Fill in the Key Contributions (with the Contribution Check), the seven sentences of the Narrative Skeleton, and Figure 0; receive the "Next Judge" at Team Meeting, and you can move on to Figures-A. Once Framing-A and Framing-B are filled in, the paper's claim structure is fixed. Structure before data — No Card → No Paper. → Lab Tools "Paper Card"