Some time ago there was a question on Lambda the Ultimate that asked how people read papers.

My problem is to quickly select the most promissing papers from the gazillions that are available. So, this is how I do it:

  1. Read the title

  2. Quickly skim the abstract - get an overview what it is all about

  3. Read the conclusion - find out what the paper really wants to tell you

  4. Really read the abstract - find out how the paper comes to the conclusion

  5. Read TOC - find out which chapters are worth reading

  6. Read the chapters that are worth reading.

After each step I decide if it is worth to spend more time on it or to move on to other papers.