paper-planning
Guides pre-writing planning for academic papers with 4 structured steps: story design (task-challenge-insight-contribution-advantage), experiment planning (comparisons + ablations), figure design (pipeline + teaser), and 4-week timeline management. Includes counterintuitive planning tactics (write a mock rejection letter to identify weaknesses before writing, narrow before broad claims, design ablations first). Use when: user wants to plan a paper before writing, design story/contributions, plan experiments, create figure sketches, set a writing timeline, or write a pre-emptive rejection letter for planning purposes. Do NOT use for actual writing (use paper-writing), running experiments (use experiment-pipeline), self-reviewing a finished draft (use paper-review), or finding research problems (use research-ideation).
CamusGIT · 已有 AI 审核记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add CamusGIT/EvoQuant --skill paper-planning技能说明
name: paper-planning description: "Guides pre-writing planning for academic papers with 4 structured steps: story design (task-challenge-insight-contribution-advantage), experiment planning (comparisons + ablations), figure design (pipeline + teaser), and 4-week timeline management. Includes counterintuitive planning tactics (write a mock rejection letter to identify weaknesses before writing, narrow before broad claims, design ablations first). Use when: user wants to plan a paper before writing, design story/contributions, plan experiments, create figure sketches, set a writing timeline, or write a pre-emptive rejection letter for planning purposes. Do NOT use for actual writing (use paper-writing), running experiments (use experiment-pipeline), self-reviewing a finished draft (use paper-review), or finding research problems (use research-ideation)." allowed-tools: "write_file edit_file read_file think_tool" metadata: author: EvoQuant version: '1.0.0' tags: [core, research, writing, academic-writing, experiment-design]
Paper Planning
A structured approach to planning academic papers before writing begins. Covers four key activities: Story design, Experiment planning, Figure design, and Timeline management.
When to Use This Skill
If you don't yet have an idea, use the
research-ideationskill first to find a problem and design a solution.
- User wants to plan a paper before writing
- User asks about structuring a paper's story or contributions
- User needs to plan experiments (comparisons, ablations)
- User wants to design pipeline figures or teaser figures
- User asks about writing timelines or submission schedules
Planning Overview
Paper planning follows four steps, ideally completed before writing begins:
Step 1: Story Design → What is the narrative? What are the contributions?
Step 2: Experiment Plan → What experiments prove our claims?
Step 3: Figure Design → How do we visually communicate the method?
Step 4: Timeline → When does each section get written?
Counterintuitive Planning First
Prioritize these counterintuitive rules before regular planning:
- Write your rejection letter first: Draft the top-5 likely rejection comments ("limited novelty", "missing baseline", "not robust", etc.), then plan experiments that directly preempt each one.
- Narrow claim before broad claim: Define the smallest defensible core claim first. Expand only after evidence is strong. Over-broad claims fail review more often than narrow strong claims.
- Design ablations before polishing method text: If a module cannot be ablated cleanly, its contribution claim is weak.
- Allocate compute to stress tests, not only benchmarks: A single convincing stress-test figure often contributes more than multiple small benchmark gains.
- Plan a fallback narrative now: If SOTA gain is marginal, predefine a secondary value proposition (efficiency, robustness, fewer assumptions, wider applicability).
See references/counterintuitive-planning.md
Step 1: Story Design
The "story" is the logical narrative that connects the problem, insight, method, and results.
Reverse Engineering the Story
Work backwards to build the story:
- What is the technical problem? — The specific challenge that existing methods cannot solve well
- What are our contributions? — The concrete technical novelties
- What are the benefits and new insights? — What advantages does our approach provide?
- How do we lead into the challenge? — How to frame the task and previous methods to naturally arrive at the challenge
Then write forward: Task → Previous methods → Challenge → Our contributions → Advantages
Core Elements to Define
Before writing any section, clearly articulate:
| Element | Question | Example |
|---|---|---|
| Task | What problem does this paper address? | "Real-time 3D scene reconstruction" |
| Challenge | Why can't existing methods solve it well? | "Cannot handle dynamic objects efficiently" |
| Insight | What key observation drives our approach? | "Motion patterns are temporally sparse" |
| Contribution | What do we propose? | "Sparse temporal attention for dynamic regions" |
| Advantage | Why is our approach better? | "Reduces computation while preserving quality" |
Starting Point: Pipeline Figure Sketch
Start by drawing a pipeline figure sketch. This forces you to clarify the overall method before writing.
The pipeline figure sketch serves as the paper's visual backbone:
- Draw it before writing anything
- It reveals whether the method is clear enough to explain
- It identifies the novel modules vs. standard components
- It determines subsection structure for the Method section
See references/story-design.md
Step 2: Experiment Planning
Plan experiments before writing to avoid discovering gaps late.
Two Categories of Experiments
Comparison Experiments — Prove our method is better:
- Which baseline methods to compare against?
- Which datasets and metrics?
- What is the evaluation protocol?
Ablation Studies — Prove each module is effective:
- Part 1: One big table showing impact of core contributions
- Part 2: Several small tables for design choices and hyperparameters
Planning Checklist
- List all comparison baselines (recent, relevant, SOTA)
- Define evaluation metrics (standard for the task)
- Identify datasets (standard benchmarks + challenging demos)
- List ablation configurations (remove each core component)
- Plan design-choice tables (hyperparameters, input quality, alternatives)
- Plan demo scenarios (challenging data to showcase upper limit)
See references/experiment-planning.md
Experiment Plan Template
Use the template at assets/experiment-plan-template.md
阅读完整来源 ↗