How to read papers

When I started grad school lots of people had lot of advice for how to read academic papers. Recently, I found a blog entry that hits on the core of what everyone was telling me when it comes to reading academic papers. This is a very good general guide for how to read academic papers, but there are a couple other things I found very helpful when reading papers.

@ashedryden in my experience papers either take 10 minutes or 2 hours. anything else and you're not getting anything out of it @lindseybieda

— wilkie (@wilkieii) April 23, 2013
  1. Charts, graphs, and diagrams. These are generally floating around somewhere in the body of the paper, but they tend to give the most clear explanation of what is at work.
  2. If there is an experiment you will want to check what were the key variables in play, how they set a control, and what the sample sizes were, and what the p-value was.
  3. Pseudo-code, actual code (albeit rare), equations, and formulas are all also important things to be on the look out for.

Usually if I am interested enough and a paper is short enough I will read the entire thing and potentially follow up on a couple citations. Short enough for me is five pages, but I am generally not lucky enough to be reading a paper that short.

If you're looking to practice why not read up about seam carving, end-to-end principle, or microarchitectures. If you are looking for a good cover to cover read Growing a Language is fantastic.