How to Read Research Papers Faster: The Three-Pass Method

Most PhD students are never taught how to read a research paper. You get a reading list, a stack of PDFs and an advisor who says “get familiar with the literature.” So you start at the abstract, push through every paragraph in order, and finish three hours later with a sore neck and a vague memory. If you want to know how to read research papers faster without skimming your way into mistakes, the answer is to stop reading them front to back.

S. Keshav, a computer science professor at the University of Waterloo, says as much in his short guide “How to Read a Paper”: researchers spend a great deal of time reading papers, yet the skill is rarely taught, so beginning graduate students learn by trial and error and “waste much effort.” This article combines his three-pass method with guidance from university library guides at Michigan and Brown. Sources are linked at the end.

Why reading in order is slow

A paper is not a story. It is a package with a fixed structure: abstract, introduction, methods, results, discussion. Different parts answer different questions, and you rarely need all of them at the same depth. The Brown University library guide puts it plainly: reading a scholarly article is not like reading a novel or a news story, and you probably will not absorb it start to finish in one go. Their advice is to treat reading as inquiry, guided by a question from your class or your own research.

Keshav’s key idea fits that. Read the paper in up to three passes, not once from beginning to end. Each pass has a goal and builds on the last.

Pass 1: the five to ten minute scan

Keshav says the first pass should take about five to ten minutes and gives these steps:

1. Read the title, abstract and introduction carefully.
2. Read the section and sub-section headings, and ignore everything else.
3. Read the conclusions.
4. Glance at the references and mentally tick off the ones you already know.

At the end, you should be able to answer what he calls the five Cs:

• Category: what type of paper is this (a measurement study, an analysis, a prototype)?
• Context: which other papers is it related to?
• Correctness: do the assumptions look valid?
• Contributions: what are the main contributions?
• Clarity: is it well written?

Many papers should stop here. Keshav notes the first pass is enough for papers outside your area that might someday matter. If you can answer the five Cs, write them in a note and move on.

Pass 2: the one-hour read

In the second pass you read with more care but ignore details such as proofs. Keshav suggests noting key points or writing in the margins. Look closely at figures: are the axes labeled, and are results shown with error bars? He argues these small things separate rushed work from excellent work. Also mark relevant references you have not read.

He says this pass should take up to an hour, and afterward you should be able to summarize the paper’s main point and its supporting evidence to someone else. If you still do not understand it, he lists three options: set it aside, come back after reading background material, or go on to a third pass.

The Michigan library guide gives a similar section-by-section route for scientific articles: abstract, then conclusion, then introduction, then results, methods and discussion, and finally one more read in order. Its advice on results is useful: if charts are confusing, read the narrative first and then return to the charts. On the discussion section it says to form your own conclusions from the data, not just accept the authors’.

Pass 3: reconstruct the paper

You need the third pass only for papers central to your work, or ones you are reviewing. Keshav’s idea is to “virtually re-implement” the paper: make the same assumptions as the authors and try to re-create the work, then compare your version with theirs. That comparison shows the real innovations and hidden weaknesses. Challenge every assumption in every statement.

Keshav says this takes about four or five hours for beginners and about an hour for an experienced reader. Those are his figures from his own experience, not a measured result. At the end you should be able to rebuild the structure of the paper from memory and say where it is strong and weak.

Use the passes for a literature review

Keshav also applies the method to a literature survey in an unfamiliar field:

1. Use an academic search engine and well-chosen keywords to find three to five recent papers. Do one pass on each, then read their related-work sections. If you find a recent survey, read that.
2. Look in their bibliographies for shared citations and repeated author names. Those are likely the key papers and key researchers.
3. Check where those researchers published recently, and scan recent proceedings of those venues.
4. Do two passes on this set. If they all cite a key paper you missed, get it and iterate.

The Brown guide adds a related point: note authors and institutions you see repeatedly, and use the literature review section of a good article as a source of further reading.

A reading note that takes two minutes

Whatever method you use, keep a short note per paper. A format that follows from the five Cs (our adaptation, not Keshav’s template):

• Citation and where you found it.
• One sentence on what the paper claims.
• One sentence on the evidence.
• One assumption or limitation you noticed.
• How it connects to your question.
• Read to which pass: 1, 2 or 3.

Hypothetical example, not a real case: you have 40 papers on your list. You do pass one on all of them over two days, around ten minutes each. You drop 15, keep 25, and plan a second pass for 12 that are close to your question. Only three get the full third pass. The time you spend goes where it matters.

Common mistakes

• Reading the abstract and stopping. The Michigan guide says the abstract should never replace reading the whole article, because it is too brief to explain everything.
• Copying the conclusions. Check that the data supports them.
• Doing a third pass on everything. It costs hours. Spend it on the few papers that carry your argument.
• Not recording what you read. Without a note, the pass is wasted.

Make the reading actually happen

The method does not remove the need to sit down and do it. A fixed reading block, with your phone away, helps. If you work better with other people around, Refocus lets you study at the same time as others, which can make it easier to start a long reading session.

Summary

1. Do not read papers front to back. Use passes.
2. Pass one: five to ten minutes, answer the five Cs.
3. Pass two: up to an hour, grasp content and check figures.
4. Pass three: only for key papers, re-create the argument.
5. Keep a two-minute note on each paper.

Want company for your next reading block? Try Refocus: study alongside others, free to start. Start at refocus.co.in.

Sources: Keshav, How to Read a Paper, ACM SIGCOMM CCR 2007, University of Michigan Library, Reading Scholarly Articles, Brown University Library, How to Read a Scholarly Article. Keshav’s paper is from 2007 and is written for computer science, but the passes apply across fields.

Cover photo: Unsplash.