How to Read a Paper

S. Keshav
David R. Cheriton School of Computer Science, University of Waterloo
Waterloo, ON, Canada
[email protected]

ABSTRACT 4. Glance over the references, mentally ticking off the

Researchers spend a great deal of time reading research pa- ones you’ve already read
pers. However, this skill is rarely taught, leading to much At the end of the first pass, you should be able to answer
wasted effort. This article outlines a practical and efficient the five Cs:
three-pass method for reading research papers. I also de-
scribe how to use this method to do a literature survey. 1. Category: What type of paper is this? A measure-
Categories and Subject Descriptors: A.1 [Introductory ment paper? An analysis of an existing system? A
and Survey] description of a research prototype?
General Terms: Documentation.
2. Context: Which other papers is it related to? Which
Keywords: Paper, Reading, Hints. theoretical bases were used to analyze the problem?

1. INTRODUCTION 3. Correctness: Do the assumptions appear to be valid?

Researchers must read papers for several reasons: to re- 4. Contributions: What are the paper’s main contribu-
view them for a conference or a class, to keep current in tions?
their field, or for a literature survey of a new field. A typi-
cal researcher will likely spend hundreds of hours every year 5. Clarity: Is the paper well written?
reading papers.
Learning to efficiently read a paper is a critical but rarely Using this information, you may choose not to read fur-
taught skill. Beginning graduate students, therefore, must ther. This could be because the paper doesn’t interest you,
learn on their own using trial and error. Students waste or you don’t know enough about the area to understand the
much effort in the process and are frequently driven to frus- paper, or that the authors make invalid assumptions. The
tration. first pass is adequate for papers that aren’t in your research
For many years I have used a simple approach to efficiently area, but may someday prove relevant.
read papers. This paper describes the ‘three-pass’ approach Incidentally, when you write a paper, you can expect most
and its use in doing a literature survey. reviewers (and readers) to make only one pass over it. Take
care to choose coherent section and sub-section titles and
2. THE THREE-PASS APPROACH to write concise and comprehensive abstracts. If a reviewer
cannot understand the gist after one pass, the paper will
The key idea is that you should read the paper in up to likely be rejected; if a reader cannot understand the high-
three passes, instead of starting at the beginning and plow- lights of the paper after five minutes, the paper will likely
ing your way to the end. Each pass accomplishes specific never be read.
goals and builds upon the previous pass: The f irst pass
gives you a general idea about the paper. The second pass 2.2 The second pass
lets you grasp the paper’s content, but not its details. The
In the second pass, read the paper with greater care, but
third pass helps you understand the paper in depth.
ignore details such as proofs. It helps to jot down the key
2.1 The first pass points, or to make comments in the margins, as you read.
The first pass is a quick scan to get a bird’s-eye view of
1. Look carefully at the figures, diagrams and other illus-
the paper. You can also decide whether you need to do any
trations in the paper. Pay special attention to graphs.
more passes. This pass should take about five to ten minutes
Are the axes properly labeled? Are results shown with
and consists of the following steps:
error bars, so that conclusions are statistically sig-
1. Carefully read the title, abstract, and introduction nificant? Common mistakes like these will separate
rushed, shoddy work from the truly excellent.
2. Read the section and sub-section headings, but ignore
everything else 2. Remember to mark relevant unread references for fur-
ther reading (this is a good way to learn more about
3. Read the conclusions the background of the paper).
 The second pass should take up to an hour. After this That will help you identify the top conferences in that field
pass, you should be able to grasp the content of the paper. because the best researchers usually publish in the top con-
You should be able to summarize the main thrust of the pa- ferences.
per, with supporting evidence, to someone else. This level of The third step is to go to the website for these top con-
detail is appropriate for a paper in which you are interested, ferences and look through their recent proceedings. A quick
but does not lie in your research speciality. scan will usually identify recent high-quality related work.
Sometimes you won’t understand a paper even at the end These papers, along with the ones you set aside earlier, con-
of the second pass. This may be because the subject matter stitute the first version of your survey. Make two passes
is new to you, with unfamiliar terminology and acronyms. through these papers. If they all cite a key paper that you
Or the authors may use a proof or experimental technique did not find earlier, obtain and read it, iterating as neces-
that you don’t understand, so that the bulk of the pa- sary.
per is incomprehensible. The paper may be poorly written
with unsubstantiated assertions and numerous forward ref- 4. EXPERIENCE
erences. Or it could just be that it’s late at night and you’re
tired. You can now choose to: (a) set the paper aside, hoping I’ve used this approach for the last 15 years to read con-
you don’t need to understand the material to be successful ference proceedings, write reviews, do background research,
in your career, (b) return to the paper later, perhaps after and to quickly review papers before a discussion. This dis-
reading background material or (c) persevere and go on to ciplined approach prevents me from drowning in the details
the third pass. before getting a bird’s-eye-view. It allows me to estimate the
amount of time required to review a set of papers. More-
2.3 The third pass over, I can adjust the depth of paper evaluation depending
To fully understand a paper, particularly if you are re- on my needs and how much time I have.
viewer, requires a third pass. The key to the third pass
is to attempt to virtually re-implement the paper: that is, 5. RELATED WORK
making the same assumptions as the authors, re-create the If you are reading a paper to do a review, you should also
work. By comparing this re-creation with the actual paper, read Timothy Roscoe’s paper on “Writing reviews for sys-
you can easily identify not only a paper’s innovations, but tems conferences” [2]. If you’re planning to write a technical
also its hidden failings and assumptions. paper, you should refer both to Henning Schulzrinne’s com-
This pass requires great attention to detail. You should prehensive web site [3] and George Whitesides’s excellent
identify and challenge every assumption in every statement. overview of the process [4]. Finally, Simon Peyton Jones
Moreover, you should think about how you yourself would has a website that covers the entire spectrum of research
present a particular idea. This comparison of the actual skills [1].
with the virtual lends a sharp insight into the proof and
presentation techniques in the paper and you can very likely
add this to your repertoire of tools. During this pass, you 6. A REQUEST
should also jot down ideas for future work. I would like to make this a living document, updating it
This pass can take about four or five hours for beginners, as I receive comments. Please take a moment to email me
and about an hour for an experienced reader. At the end any comments or suggestions for improvement. You can also
of this pass, you should be able to reconstruct the entire add comments at CCRo, the online edition of CCR [5].
structure of the paper from memory, as well as be able to
identify its strong and weak points. In particular, you should 7. ACKNOWLEDGMENTS
be able to pinpoint implicit assumptions, missing citations
The first version of this document was drafted by my stu-
to relevant work, and potential issues with experimental or
dents: Hossein Falaki, Earl Oliver, and Sumair Ur Rahman.
analytical techniques.
My thanks to them. I also benefited from Christophe Diot’s
perceptive comments and Nicole Keshav’s eagle-eyed copy-
3. DOING A LITERATURE SURVEY editing.
Paper reading skills are put to the test in doing a literature This work was supported by grants from the National
survey. This will require you to read tens of papers, perhaps Science and Engineering Council of Canada, the Canada
in an unfamiliar field. What papers should you read? Here Research Chair Program, Nortel Networks, Microsoft, Intel
is how you can use the three-pass approach to help. Corporation, and Sprint Corporation.
First, use an academic search engine such as Google Scholar
or CiteSeer and some well-chosen keywords to find three to
five recent papers in the area. Do one pass on each pa-
8. REFERENCES
per to get a sense of the work, then read their related work [1] S. Peyton Jones, “Research Skills,”
sections. You will find a thumbnail summary of the recent http://research.microsoft.com/ simonpj/Papers/giving-
work, and perhaps, if you are lucky, a pointer to a recent a-talk/giving-a-talk.htm.
survey paper. If you can find such a survey, you are done. [2] T. Roscoe, “Writing Reviews for Systems
Read the survey, congratulating yourself on your good luck. Conferences,”
Otherwise, in the second step, find shared citations and http://people.inf.ethz.ch/troscoe/pubs/review-
repeated author names in the bibliography. These are the writing.pdf.
key papers and researchers in that area. Download the key [3] H. Schulzrinne, “Writing Technical Articles,”
papers and set them aside. Then go to the websites of the http://www.cs.columbia.edu/∼hgs/etc/writing-
key researchers and see where they’ve published recently. style.html.
[4] G.M. Whitesides, “Whitesides’ Group: Writing a
Paper,”
http://www.che.iitm.ac.in/misc/dd/writepaper.pdf.
[5] ACM SIGCOMM Computer Communication Review
Online, http://www.sigcomm.org/ccr/drupal/.