This post is a slight edit of a "how to read a paper" pdf by S. Keshav: We removed the last remaining redundancy from the original pdf, so read carefully for full benefit.

It has come to our attention that many folks read few papers, even if deeply; It is our observation that those who've read many papers shallowly as well understand the manifold of a research field much better than those who only read papers deeply. Depth is important for the key papers, but throughput is critical for seeing the broader map and learning to make use of low-quality papers to build intuition. This seems especially true in pre-paradigmatic fields such as alignment. This PDF is a nice little intro to reading lots of papers fast, and I find myself frequently wanting to reference it.

Edit to add note (gearstoascension):

If you read few papers, and this post does not influence your likelihood to read more papers, I'd love to hear a step by step trace-through of what repels you from the actions that discover papers and why you don't read them - eg, are you busy? not expecting to understand? sure you already know better? etc. The goal here is to argue that those who are doing research should likely read dozens to hundreds of abstracts a month, first-pass tens of papers, and second-pass at least a few. If you aren't doing that, and this post gets you nodding but doesn't change your behavior, I'd be interested to know why.

See also Literature Review For Academic Outsiders: What, How, and Why.

Introduction: Up to three passes

Learning to efficiently read a paper is a critical but rarely taught skill. This two page pdf introduces a key idea that you should read the paper in up to three passes, instead of starting at the beginning and plowing your way to the end.

Each pass accomplishes specific goals and builds upon the previous pass: The first pass gives you a general idea about the paper. The second pass lets you grasp the paper’s content, but not its details. The third pass helps you understand the paper in depth.

(Zeroth pass: browse)

This isn't mentioned in the linked pdf, but it's also important to have high throughput on finding papers to read that have a shot at being both relevant and insightful, which can mean they're outside your explicit field. More in a later post or comments. Paper discovery is a separate topic from paper reading, but one key strategy is browsing the citation g...

Posts

Wikitag Contributions

Comments