How to read a paper

Since I’m writing my master’s thesis, I have to read some scientific papers. In my previous years it wasn’t necessary to read those, so I’m not used to it. That’s a pity because they contain bleeding edge knowledge compressed in a self-contained form of a few pages.

I started reading them like technical books which didn’t work because the contents of books are usually well edited (by people who are concerned about the language, not the content) and comparatively easy to read. Reading a paper is quite different, they are very compact and have nearly no redundancy which makes them hard to read.

I’ve collected a few tips on how to read papers:

  • Read only one paper at a time. Papers differ in their notation and sometimes in their definitions (even papers from the same authors). Switching between papers (and notations as well as definitions) confuses and doesn’t help you to get the knowledge you need.
  • Don’t start in the middle of the paper. As said before, papers are highly compressed and have low redundancy, if you start in the middle you’ll miss the explanations about their notation and the definitions. Scientists usually make their papers self-contained, so they include everything you’ll need to know to understand the paper, use the information!
  • Read the whole paper before you start using the knowledge unless you’re absolutely sure that you don’t need the remaining chapters. But read the conclusion at the end of the paper, usually it contains an informal summary and open issues which may be of interest.
  • It’s a good to do a simple real-world example while reading a paper. It’s a method to slow you down and force you to understand every sentence of the paper.
  • Sometimes it takes hours only for a few pages of a paper, that’s not unusual, even for those who worked for years in the field.
  • Don’t get frustrated if you don’t understand some paragraph, read the rest of the chapter, maybe there are further explanations on the topic, or just get some sleep and let your sub-consciousness work out the details.

Update: Thomas Krennwallner pointed me to a guide called Efficient Reading of Papers in Science and Technology, it was written in 1989, but updated in 2000. Thanks, dude!

Walking has a very good effect in that you’re in this state of relaxation, but at the same time you’re allowing the sub-conscious to work on you.
— Andrew Wiles