CS Three-Pass Reading Method

Systematically read computer science papers using Keshav's three-pass approach adapted for CS structure.

Quick Start

1. First Pass (5-10 min): Get overview → references/first-pass.md
2. Second Pass (up to 1 hour): Grasp content → references/second-pass.md
3. Third Pass (1-5 hours): Deep understanding → references/third-pass.md
4. Reading Report: Integrate all passes → references/reading-report.md

Workflow

Phase 1: First Pass - Overview

Goal: Determine relevance and get bird's-eye view.

Read: Title, abstract, introduction, section headings, conclusions.

Answer the Five Cs:

1. Category: What type of CS paper? (algorithm/systems/empirical/survey)
2. Context: Related papers and theoretical foundations
3. Correctness: Are assumptions valid?
4. Contributions: Main contributions?
5. Clarity: Is paper well-written?

Decision: Proceed to second pass? → See references/first-pass.md

Phase 2: Second Pass - Content Grasp

Goal: Understand contribution without getting lost in details.

Focus: Figures, diagrams, tables, algorithms. Ignore proofs.

Tasks:

• Analyze methodology and innovation
• Evaluate experimental design
• Create structured summary
• Identify strengths and weaknesses

Decision: Proceed to third pass? → See references/second-pass.md

Phase 3: Third Pass - Deep Understanding

Goal: Virtually re-implement the paper.

Activities:

• Challenge every assumption
• Identify implicit assumptions and missing citations
• Assess experimental techniques
• Generate future work ideas

Output: Critical analysis with reviewer perspective → See references/third-pass.md

Phase 4: Reading Report Integration

Goal: Synthesize all passes into comprehensive analysis.

Activities:

• Integrate summaries from all three passes
• Synthesize strengths, weaknesses, and future directions
• Create actionable insights for research or implementation
• Generate comprehensive analysis with reviewer perspective

Output: Complete reading report → See references/reading-report.md

Computer Science Paper Structure

Adapt reading to typical CS sections:

• Abstract/Summary: Core contribution snapshot
• Introduction: Problem motivation and research gap
• Related Work: Positioning in field
• Method/Model: Technical approach
• Experiments: Evaluation methodology
• Results: Findings and analysis
• Conclusion: Summary and future work
• References: Key related work

Usage Patterns

For Literature Survey

1. First pass on multiple papers (5-10 each)
2. Second pass on relevant papers (1 hour each)
3. Third pass on key papers (2-5 hours each)
4. Create reading report for each key paper
5. Synthesize multiple reports into literature review

For Paper Review

1. First pass: Initial assessment
2. Second pass: Detailed evaluation
3. Third pass: Critical analysis for review
4. Create reading report with reviewer perspective
5. Use report as basis for conference/journal review

For Research Implementation

1. First pass: Scope assessment
2. Second pass: Technical understanding
3. Third pass: Implementation planning
4. Create reading report with implementation focus
5. Identify extension potential and collaboration opportunities

Tips

• Time management: Strict time limits per pass
• Note-taking: Use margins or separate document
• Reference tracking: Mark important unread papers
• Critical thinking: Don't assume authors are always right
• CS-specific: Pay attention to algorithms, complexity, systems design, experimental setup

When to Stop

• After first pass: Paper not relevant to your work
• After second pass: Paper is relevant but not central to your research
• After third pass: Full understanding achieved

Output Formats

Each pass has specific output templates in the reference files. Use them to structure your analysis and maintain consistency across papers.

Reading Report

After completing all three passes, use references/reading-report.md to create a comprehensive report that integrates all findings. The report includes:

• Paper metadata and executive summary
• Integration of all three pass analyses
• Critical synthesis of strengths and weaknesses
• Future directions and personal assessment
• Optional reviewer perspective