Survey

Literature Survey Skill — Intent → Triage → Deepen → Synthesize

This skill helps PhD students build a deep, synthesized corpus of research insights through a structured, cognitively-aware workflow. It combines Kahneman's dual-process theory, Keshav's three-pass reading method, first-principles analysis, and NotebookLM as a backend query engine.

Architecture: All student work lives locally (Obsidian/filesystem). NotebookLM is a query backend only — papers go in, grounded answers come out.

Modes:

• /survey intent — Capture what you're trying to learn and where you're starting from
• /survey triage — Map the landscape, prioritize reading depth
• /survey deepen — Structured reading with craft and visualization extraction
• /survey synthesize — Cross-paper analysis for deliverables
• /survey expand — Structured corpus growth proposals

If the student invokes /survey without a mode, ask which mode they want. If they seem unsure or are starting a new survey, begin with intent.

Prerequisites: NotebookLM MCP CLI must be configured. See reference/notebooklm_tools.md for setup.

Mode 1: Intent — "What are you trying to learn?"

Purpose: Capture the student's survey goals, expertise level, and success criteria BEFORE any paper is read. This shapes all subsequent modes.

Do NOT create any NotebookLM notebooks in this mode. Do NOT ingest any papers. The goal is purely clarifying intent.

Step 1: Identify the survey archetype

Ask the student which best describes their situation:

Before we look at any papers, I need to understand what kind of survey this is. Which best describes you?

A. Explorer — "I'm entering a new area and need to understand the landscape." B. Investigator — "I have specific questions I need answered from the literature." C. Validator — "I think I've found a gap/idea and want to confirm it's novel." D. Examiner — "I need to demonstrate comprehensive mastery for an exam or survey paper."

Each archetype has different defaults:

Step 2: Assess expertise level

Ask calibration questions for the chosen topic:

1. Vocabulary check: Can you name 3 key technical terms in this area and define each in one sentence?
2. Landmark papers: Can you name any papers, authors, or research groups you associate with this area?
3. Current mental model: In 2-3 sentences, what's your current understanding of the main problem and approaches?
4. Known unknowns: What specific questions do you hope the literature will answer?

Cognitive purpose: These questions create a baseline for later comparison. After triage, the student will see how their mental model changed — making System 1's invisible anchoring effects visible.

Step 3: Define success criteria

1. Deliverable: Related work section? Area exam presentation? Gap analysis? Idea validation?
2. Scope: Approximately how many papers do you expect to cover?
3. Timeline: When do you need the output?
4. Time budget: How many hours per week can you invest?

Step 4: Check for advisor input

Has your advisor or collaborator recommended specific papers or threads to explore? Any "must-read" papers with suggested reading depth?

Step 5: Generate the intent profile

Write a survey_intent.md file using the template from templates/survey_intent_template.md. Save it to the student's survey directory:

literature-survey/surveys/<topic-slug>/survey_intent.md

Also create the directory structure:

surveys/<topic-slug>/
├── survey_intent.md
├── pdfs/
├── papers/
│   └── figures/
├── synthesis/
├── corpus_log.md
├── backlog.md
└── nlm_config.md

Tell the student: "Your survey intent is captured. Run /survey triage when you're ready to start mapping the landscape."

Mode 2: Triage — "What's in this pile?"

Purpose: Rapid Pass 1 over a corpus of papers using NotebookLM. Map the landscape and decide where to invest deeper reading. The archetype from intent mode shapes scope and clustering.

Step 1: Set up NotebookLM backend

Read the student's survey_intent.md and nlm_config.md. If no notebook exists yet:

notebook_create(title="survey-<topic-slug>")
chat_configure(notebook_id=<id>, goal="custom",
    custom_prompt="You are a research corpus query engine for a PhD student
        surveying [topic]. The student is an [archetype] with goals: [from intent].
        Always cite specific papers and quote relevant passages.",
    response_length="longer")
tag(action="add", notebook_id=<id>, tags=["survey", "<topic>", "<term>"])

Save the notebook_id to nlm_config.md.

Step 2: Seed the corpus

Ask the student for their initial papers (PDFs, URLs, BibTeX). For each paper:

1. Acquire the PDF locally using the priority chain:

   • Student already has the file → copy to pdfs/YYYY_author_shorttitle.pdf
   • Open-access URL (Arxiv, university repo) → download via wget
   • Semantic Scholar API → query for open-access PDF URL
   • Prompt the student for upload or URL

2. Ingest into NotebookLM (async for large batches):

   source_add(notebook_id=<id>, source_type="file", file=<path>, wait=False)

3. Log in corpus_log.md with status "ingesting".

While papers load (3-5 min for 15+ papers), use the time productively — refine intent, discuss the topic, let the student start reading local PDFs if they want.

Step 3: Generate Pass 1 summaries

Once sources are ready, for each paper query NLM:

notebook_query(notebook_id=<id>,
    query="For [title] by [author]: provide CATEGORY (measurement/systems-building/theory/survey),
    PROBLEM (one sentence), CONTRIBUTION (exact quote of main claim), EVALUATION (system/dataset/testbed),
    KEY REFERENCES (3-5 most cited references), RELEVANCE to survey goals (high/medium/low)")

Write each response to a local file: papers/YYYY_author_shorttitle.md with a [Pass 1] header.

Step 4: Generate landscape map

Query NLM for a cross-corpus landscape:

notebook_query(notebook_id=<id>,
    query="Group all papers by PROBLEM addressed (3-5 major threads). For each thread, list papers
    and methodology. Identify chronological patterns. WYSIATI CHECK: What problem areas or
    methodologies are absent? What would a skeptical reviewer say is missing?")

Write to survey_triage.md.

Step 5: WYSIATI check and intent comparison

This is a System 2 checkpoint. Show the student their initial mental model (from survey_intent.md) alongside the landscape map:

"Here's what you thought the area looked like when you started. Here's what the corpus actually shows. What's different? What surprised you?"

Step 6: Prioritize reading depth

Help the student categorize each paper against their time budget:

• Pass 1 only — background knowledge
• Pass 2 recommended — relevant to core thread
• Pass 3 required — foundational, must deeply understand

Update survey_triage.md with the prioritized reading list.

Step 7: Surface expansion candidates

notebook_query(notebook_id=<id>,
    query="Which references appear in 3+ papers but are NOT sources in this notebook?
    For each, explain why it might be important to add.")

Append candidates to corpus_log.md with reason=FOUNDATIONAL. Present to student for Add/Bookmark/Skip decision (see /survey expand protocol).

Mode 3: Deepen — "What does this paper really say?"

Purpose: Structured Pass 2 and Pass 3 reading for individual papers. Forces System 2 engagement, captures insights locally, uses NLM for grounded extraction and calibration.

Step 1: Select a paper

Ask which paper from the triage reading list. Read their existing paper note if one exists.

Step 2: Pass 2 protocol

Run four NLM queries and write results to the local paper note:

a. Claim extraction:

notebook_query: "For [title]: Quote the 3 most important claims with section/page references."

b. Evidence audit:

notebook_query: "For [title]: For each claim, what evidence is provided? Rate as strong/moderate/weak."

c. Methodology probe:

notebook_query: "For [title]: Describe evaluation setup. What explicit and implicit assumptions?
What would break if workload/scale/topology changed?"

d. Dependency extraction:

notebook_query: "For [title]: What results from other papers does this depend on?
Which dependencies might not hold in other contexts?"

Write all responses to the paper's local note as a [Pass 2] section.

Step 3: Calibration check (System 2 prosthetic)

After the student has read the paper themselves, compare their understanding with NLM's grounded extraction:

1. Ask the student: "In one sentence, what is this paper's main contribution?"
2. Query NLM: "Quote the exact main contribution claim from the abstract or introduction."
3. Show both side by side in a [Calibration] section. Highlight the specificity gap.

Step 4: First-principles decomposition

notebook_query: "For [title], analyze along four dimensions:
    STATE: What state does the system manage?
    TIME: What timescales matter?
    COORDINATION: How do components coordinate?
    INTERFACE: What are the boundaries between components?
    Quote specific passages as evidence."

Write to the paper note as [First-Principles] section.

Step 5: Pass 3 — Virtual re-implementation (foundational papers only)

Ask the student:

• "If you had to build this system from scratch with the same goals, what would your design look like?"
• "What assumptions are never explicitly stated?"
• "Write three critical questions a skeptical PC member would ask."