15-440 - Distributed Systems

Home Syllabus Schedule Lectures Assignments Projects Exams Resources

Problem Solving Assignments

A note on how to read/critique a research articles

The papers/articles/chapters given for reading provide interesting approaches for designing distributed systems. We have chosen classical papers from each topic to make sure that you read a well-known arguments and approaches. Most of the arguments are time-tested and still hold while designing systems. Some may not applicable to present day distributed systems. While critiquing the papers, please keep in mind the current distributed systems that you see. What are your views on these guidelines to present day systems? Try to start with positive thinking while critiquing. It is generally helpful to first understand what and why the author is stating. Then apply your experience and common sense to see if the arguments are valid. Asking yourself the following questions may help while critiquing: Are the principles strong and can you see them being applied to present-day systems. Or has it changed over the years? Are there new principles/papers that are more relevant?

Before you start reading/critiquing: Read the following articles on: (1) how to read research papers, and (2) how to critique a paper. With respect to critiquing, keep the following points in your mind for the above papers (most of the points are taken from the article on "how to critique"):

• Do you find errors of fact and interpretation? (This is a good one! You won't believe how often authors misinterpret or misrepresent the work of others. You can check on this by looking up for yourself the references the author cites.)
• Is all of the discussion relevant?
• Have any ideas been overemphasized or underemphasized? Suggest specific revisions.
• Are the author's statements clear? Challenge ambiguous statements. Suggest by examples how clarity can be achieved, but do not merely substitute your style for the author's.
• What underlying assumptions does the author have?
• Has the author been objective in his or her discussion of the topic?
• If the paper is classical (and old!), is there any new paper that shines new light on the discussion?