Paper Card

Why Paper Card

The biggest reason research stalls is not a shortage of data, and not a lack of ability. It is the absence of a question (Lab OS 03: Question Design). Research that collects data first and asks for meaning later always goes astray.

Paper Card reverses this order by force. It asks you to design the structure a finished paper will need — question, gap from prior work, contribution, figures, and narrative — before you collect data. Paper Card is not a progress report. It is your research blueprint.

Three Roles of the Paper Card

1. Blueprint — Structure first, data second. Once the question is fixed, the data you need can be back-calculated. Once the figures are fixed, the text to write becomes clear. Filling in the card in order is the research design itself.
2. Common language — Anyone who opens your card can immediately see where your research stands and what decision comes next. Team Meetings run on "Paper Card → decision → next action" (Lab OS 09). With the card, discussion never starts with "wait, what are we talking about?"
3. Inquiry history — Research is the updating of beliefs (Lab OS 01). The strikethroughs and dates on your card are the record of how your beliefs have been updated.

Three Rules

1. This is not a ranking — A person further along the stages is not more valuable. Someone with 10 Ideas and someone polishing one Draft are simply at different phases of inquiry.
2. Reversing, stopping, and dropping are normal — The inquiry cycle does not move in a straight line. Returning from Framing to Idea is not failure — it means the precision of your design just improved.
3. Revisions are not deleted. They are crossed out. — When you revise after a meeting, leave the old text with a strikethrough and a date: like this (revised 2026/02/02). The reason: research advances through belief-updating. If you erase the history of updates, you cannot see why you arrived at the current claim — and you repeat the same deliberations. A strikethrough is evidence the inquiry cycle turned, not a mark of confusion.

Overview: Four Stages and Their Gates

Each stage ends with a ➡️ Next Judge. The decision to advance, return, or stop is made in Team Meeting, and sparring also takes place within this Next Judge. Doing decision and refinement together in the same venue is the role of Team Meeting; the decision and its reasoning go into the Decision Log (The 5-Layer Structure).

🌱 Idea — Accumulate discomforts, then commit

Idea-0 | Spark: A place to accumulate instinctive discomforts and questions — without judging them. One or two sentences of bullet points is fine. No need to share at this stage. Do not ask "will this become a paper?" yet. Do not search prior literature prematurely — scrutinising a seed before it sprouts prevents germination.

Idea-A | Check & Commit: Take one Spark and bring it to a state where someone else can judge that it "looks designable as a paper." Write five things:

1. Likely field
2. Close papers (1–2; no need to understand them fully)
3. What seems new or different
4. Why you want to pursue this topic
5. A provisional target journal (journal name + why the scope and readership seem right)

Note: No results or numbers needed. Do not try to prove novelty. No complete literature review, no figure plans, no experiment designs yet. The only question here is: "Is this a topic worth designing?"

🧭 Framing — Build a positioned question in one sentence

Framing is not idea generation. It is the process of constructing a Core Question — one sentence that is positioned within the field. Everything downstream — figures, manuscript, journal — flows from this one sentence.

Framing-A | Positioned Question Design

Start by surveying the field (key review papers + 5+ representative recent papers; your own notes are fine). Grasping the Question Archetypes common in the field prevents later wandering. Then organise five points into a form you can explain at Team Meeting:

1. Field Belief — How does the field currently understand this problem?
2. Comparison table — Line up prior papers' claims, hidden assumptions, and limitations; show which assumption your research updates (Compared to X, we Y)
3. Core Question (one sentence) — No results, no methods, no Yes/No endings
4. Why This Question Matters — Why this question is important (1–2 lines)
5. Belief Update — Previous belief (the field's implicit assumption) → Updated belief (the new understanding this research will demonstrate)

Framing-B | Key Contributions and Story Design

Once the question is fixed, design what the paper will claim.

• Key Contributions (max 3) — All in "We show that …" form. No vague words like "important" or "significant." Only include what a figure can support.
• Narrative Skeleton — Write the story skeleton in 7 lines: Broad Context → Field Context → Gap → Core Question → Here we show → Field Meaning → Broader Positioning
• Figure 0 (Story Map) — Draw the dependency relationships among contributions using boxes and arrows. This is not a flow chart — it is a structure diagram of your claims. Identify the core of your research by asking: which box, if removed, collapses the whole thing? (the collapse test)

Once Framing is settled, sparring and the decision take place at the Next Judge of Team Meeting. Break and sharpen the question, positioning, and expression, and decide whether to advance, return, or stop.

🖼 Figures — Build the claim through figures

The rule is 1 figure = 1 claim. A figure is not an illustration of results — it is the claim itself. Design the "empty figures" before you have data.

1. Figures-A | Design empty figures — Assign a figure to each Contribution (C1 → Fig.1 …). For each figure, decide: what does it claim, axes, comparison axis, panel layout, and one-sentence caption. No real data yet.
2. Figures-B | Preliminary verification — Use minimal data to confirm "can this figure actually support its claim?" Check whether the comparison axis is coherent and whether figures are proliferating.
3. Figures-C | Freeze (Freeze & Reduce) — Freeze the figure set at minimum count and maximum claim strength. Cut weak or unnecessary figures. Limit remaining work to a "minimum additional-work list."

The frozen figure set becomes the skeleton of the paper. Here, too, sparring is built into the Next Judge of Team Meeting.

→ For details, see the Lab Forum lecture "Writing Papers — Figures".

✍️ Draft — Do not start with the Introduction

The writing order is fixed:

Methods → Results → Discussion → Introduction → Conclusion → Abstract → Polish

Why this order: first fix the facts (Methods / Results), then give them meaning (Discussion), then place them in the context of the field (Introduction). Writing the Introduction first means previewing a claim that does not yet exist.

Results = "what happened." Discussion = "what it means."
If Introduction is a declaration of absence, Discussion is a declaration of update. If they do not mirror each other, the structure is broken.

→ For details, see the Lab Forum lecture "Writing Papers — Draft".

📝 Decision / Log — Never delete

Record each gate decision (Decision) and its reasoning (Reason), with a date, at the end of the card. "Why did I keep pursuing this topic?" "Why did I abandon that direction?" — these are records for yourself six months from now, and for whoever eventually continues this work. Like the strikethrough rule: the history of decisions is part of the inquiry itself.

Get Started

Lab members: Everyone's Paper Cards live in the Paper Progress Board (Notion database). Create a new card from the board template and update it every week (Lab OS 10: Paper Card is a required weekly habit). Make sure it is current before every Team Meeting.

* Viewing and editing the board requires lab membership login.

Outside the lab: The template structure is exactly as described on this page. It works in Notion, Word, or on paper. The tool is not the point — the sequence "structure first, data second" is.