How to Read a Paper

Chapter 1 Episode 2

Knowing how to find papers is only half the skill. The other half, and often the more demanding one, is knowing how to read them, because reading 100 papers at the wrong depth teaches you less than reading 30 at the right one. A pattern I have seen in almost every junior researcher is that they sit down with a stack of papers, start reading the first one from beginning to end, and three hours later, they have finished a single paper and feel no closer to understanding the landscape. Both the exhaustion and the feeling that the reading will never end are real, because reading every paper at the same depth is the surest way to make a survey feel impossible.

If you followed the previous post, you now have a collection of papers sorted into three classes: highly relevant to your work, loosely relevant to your broader area, and outside your scope. The papers outside your scope are set aside. For the remaining two classes, the question is how to read each one in a way that builds genuine understanding without burying you in detail that does not address your question.

Reading Loosely Relevant Papers

For loosely relevant papers, you do not need to read every word. Your goal is to understand what each paper did and where it sits in the landscape, so you need enough to make an accurate classification and to extract the broad outlines. Read the introduction carefully, since the introduction lays out the motivation and the research questions regardless of what else the paper does, and if there is anything you can learn from or position your own work against, the introduction is where you will find it. Examine the diagrams and figures closely, since the authors put considerable effort into condensing their most important information into the visual presentation. Then read the high-level themes in the methodology, findings, and discussion sections. This content is enough to have a rough picture of what the paper set out to do, what it found, and what the authors themselves identify as limitations or open directions. You should be able to summarize the paper in a few sentences when you are done, and reading at this level is a genuine skill that keeps your survey moving without burying you in details that do not directly serve your question.

Reading Highly Relevant Papers

For highly relevant papers, you need to go considerably deeper, because these are the papers that occupy the closest territory to your own work. For instance, some pieces of the motivations may be adaptable to your work; you may have overlooked stakeholders they have included, or they may have already identified findings you believed were novel. Start the same way you would with a loosely relevant paper: read the introduction, examine the diagrams, and absorb the high-level themes of methodology, findings, and discussion. Then go back and engage with the substance.

In particular, you need to understand the rationale they built for the problem and how they arrived at their research questions. Pay attention to the key terms that define their scope, and ask whether those terms align with or diverge from yours. You should also examine the consistency among their motivation, research questions, methodology, and findings, since that coherence is what holds a strong paper together, and studying it in published work teaches you how to build the same coherence into your own. Read the findings closely enough to understand how they formulated their storylines, and note the classic or related papers they cite in their discussion, since those citations often point to foundational work you should know. Reading at this depth takes significantly more time, especially early on when you are still learning how to engage critically with a paper, but the investment compounds, and you will become faster and more confident with practice.

Note-Taking and Organizing

The volume of papers you read over time will make organizing them a really critical challenge. A reference manager like Zotero or Mendeley handles the bibliographic housekeeping, but the real skill is structured note-taking alongside it. For every paper you read beyond a quick classification pass, I would recommend recording four things in your own words: the problem they addressed, the method they used, the key finding, and the limitation or open question they left behind. When you restate their work yourself, you are forced to reinforce a genuine understanding of it in a way that simply copying an abstract does not. Record your relevance classification (highly relevant or loosely relevant) alongside the note, and explicitly explain how each paper relates to your emerging question, even when the connection is tentative. Without these notes, you will find yourself re-reading the same papers three months later because you cannot remember what they said or why you thought they mattered, and I have seen students lose entire weeks to that cycle.

The literature survey has a second, deeper use that goes beyond mapping the landscape, one that makes it the single most powerful method for teaching yourself the craft of research on your own terms. The next post is about exactly that.