Triaging Literature with Three Pass Reading
Triaging Literature with Three-Pass Reading
When given a pile of papers (e.g., 50-100 search results), don't read linearly. Apply immediately:
- Pass 1 (5 min/paper): Read title, abstract, conclusion only. Ask: (a) What's the research question? (b) How does it relate to my work? (c) Does the conclusion look credible? Keep or discard. Expect to discard ~90%.
- Pass 2 (1 hr, only on survivors): Skim structure, write 3 specific questions before reading (e.g., "what dataset?", "how is the key variable measured?"), read to answer them, then close the paper and restate the core argument from memory.
- Pass 3 (deep critique, 1-2 papers only): Line-by-line scrutiny of methods, statistical assumptions, potential bias. Write down at least one point of disagreement.
Never read a batch of papers front-to-back. Filter first, invest time second.
Progress:
- Step 1: Collect all candidate papers into one list/folder
- Step 2: Run Pass 1 on every paper (5 min cap each) — sort into "discard" / "keep"
- Step 3: Run Pass 2 on the ~10% that survived Pass 1
- Step 4: Identify the 1-2 papers most central to your own research question
- Step 5: Run Pass 3 deep critique only on those core papers
- Step 6: For every paper that survives Pass 2 or Pass 3, fill out the structured note template (below) — never just copy quotes
- Step 7: File notes by category/index, not by long free-form summaries
Structured Note Template (use instead of copy-paste excerpts)
For each paper that passes Pass 1:
- Research question (1 sentence)
- Core finding (1 sentence)
- Method strengths (1-2 points)
- Method weaknesses (1-2 points)
- Relevance to my own research (mandatory — force an answer even if tenuous)
- One point of disagreement or skepticism (mandatory — force yourself to find one)
The last two fields are the most important. If you can't fill them in, you haven't actually understood the paper — go back and reread it.
Example 1: Input: 80 papers downloaded from a database search on "algorithmic bias in hiring" Output: After Pass 1 (5 min each, ~7 hours total), 8 papers survive. After Pass 2 (1 hr each, ~8 hours), 3 papers are clearly central. Pass 3 deep critique is done on those 3, each yielding a structured note with an explicit disagreement point (e.g., "sample size of 200 is too small to support the causal claim in Section 4").
Example 2: Input: A single paper that seems important but you can't recall anything about it two days after reading Output: Diagnostic — you likely read it front-to-back without a structured note. Reread using Pass 2 method: write 3 questions first, read to answer them, close the paper, restate the argument aloud, then fill out the 6-field template. This time the "relevance" and "disagreement" fields force retention.
- Set a hard timer for Pass 1 (5 min) and Pass 2 (1 hr) — the time cap is what forces triage discipline.
- Always write the "relevance to my own research" and "point of disagreement" fields first if stuck — they're diagnostic for whether you actually understood the paper.
- Organize notes by folder/category with an index, not by exhaustive per-paper essays.
- A paper is only "finished" when you can: (1) state its core argument in one sentence to someone else, (2) name one specific flaw or point of contention, (3) explain how it serves your own research.
- Let disagreements and open questions become seeds for your own research questions — don't just archive them.
- Reading front-to-back like a textbook. This is the single biggest time sink. Papers are not meant to be read linearly on first pass.
- Writing 500+ word notes per paper. High effort, low retention. Structured 6-field notes outperform long summaries.
- Copy-pasting "important" sentences. Excerpts without your own synthesis are forgotten within days. Force paraphrase and judgment instead.
- Being unable to name a single flaw in a paper. If you can't critique it, you haven't read it deeply enough — you've only skimmed it.
- Investing deep-read time (Pass 3) in papers that aren't central to your own research question. Reserve deep critical reading for the 1-2 papers that matter most.