Writing Tips

Andrzej Duda
Grenoble Institute of Technology
CNRS Grenoble Informatics Laboratory UMR 5217
Grenoble, France
Email: [Link]@[Link]

Abstract—When working on many papers, I realized that I G OOD : Assume that A is a group.
correct almost always the same mistakes. Here is a compilation 7) Simplify “which is”.
of examples illustrating various types of mistakes. I also gathered
a lot of useful tips on how to correct them. The report includes W RONG : The figure presents the time interval
the text by Mark Allman on the view of a referee on the process which is available for other nodes.
of writing papers. G OOD : The figure presents the time interval avail-
I. C OMMON M ISTAKES able for other nodes.
8) Avoid passive voice.
Here are the most common mistakes that you can easily
avoid. W RONG : The simulation results are presented in
Figure 1.
1) Nouns in a nominal group are singular except the last
G OOD : Figure 1 presents the simulation results.
one.
9) Check for missing articles, particularly if your native
W RONG : packets delivery ratio
tongue does not have them. Roughly, concepts and
G OOD : packet delivery ratio
classes of things do not, most everything else more
W RONG : The simulations results show...
specific does. Do not use articles in front of proper nouns
G OOD : The simulation results show...
and names.
2) Resist the temptation to use long strings of nouns as
G OOD : Routers route packets.
adjectives.
G OOD : The router architecture we consider uses
W RONG : Consider the packet switched data com- small rodents.
munication network protocol problem. G OOD : Internet Explorer is a popular web browser.
3) “Which” vs. “That”. G OOD : The current version number is 5.0.
W RONG : We present the structure of packets which G OOD : Bill Gates did not write Internet Explorer.
contain the neighbor addresses. 10) Do not use "etc." unless the remaining items are com-
G OOD : We present the structure of packets that pletely obvious.
contain the neighbor addresses. ACCEPTABLE: We shall number the phases 1, 3, 5,
More on this topic. 7, etc.
“That” is the defining or restrictive pronoun, U NACCEPTABLE: We measure performance factors
“which” the non-defining or nonrestrictive. such as volatility, scalability, etc.
The lawn mower that is broken is in the garage.
11) Avoid “etc.”; use “for example”, “such as”, “among
(Tells which one.)
others” or, better yet, try to give a complete list (unless
The lawn mower, which is broken, is in the garage.
citing, for example, a list of products known to be
(Adds a fact about the only mower in question.)
incomplete), even if abstract.
If you can substitute ‘that’ for ‘which’, do it.
12) Uncountable nouns are substances or concepts that we
4) Punctuation in phrases with “which” and “that”. cannot divide into separate elements (e.g., advice, infor-
G OOD : All the students that know when to use mation, news, work)
“which” and “that” will pass the quiz. W RONG : informations, interferences
G OOD : The exam, which took place at the begin- G OOD : information, interference
ning of class, was not difficult. W RONG : an information
5) Punctuation in enumerations. G OOD : a piece of information
G OOD : It had bells and lights. G OOD : the information
G OOD : It had bells, whistles, and lights. 13) Possessive ’s. Concerns persons not things.
6) Do not omit “that” when it helps the reader to parse the W RONG : packet’s size
sentence. G OOD : packet size
W RONG : Assume A is a group.
 2

14) Allow. Permit. Permit is more formal than allow. Enable. capitalized use double {{ }} in a bibtex item:
G OOD : They do not allow smoking here. @article{Jain2001,
Author = {R. Jain and A. Puri and R. Sengupta},
G OOD : They do not allow us to smoke here. Title = {{Geographical Routing Using Partial
G OOD : They do not permit smoking here. Information for Wireless Ad Hoc Networks}},
G OOD : They do not permit people to smoke here. Journal = {IEEE Personal Comm.},
Month = {February},
G OOD : It enables nodes to detect collisions. Year = {2001}
15) Bibliographic references are not nouns. The phrase }
needs to make sense if you delete them. W RONG : Classifying service flows in the encrypted
W RONG : The authors in [7] proposed the best skype traffic.
protocol for flooding. G OOD : Classifying Service Flows in the Encrypted
G OOD : The authors proposed the best protocol for Skype Traffic.
flooding [7]. 23) Adverbs before verbs.
W RONG : In [7], Jain et al. proposed the best W RONG : The protocol operates usually in indoor
protocol for flooding. environments.
G OOD : Jain et al. proposed the best protocol for G OOD : The protocol usually operates in indoor
flooding [7]. environments.
16) No articles before symbols. 24) Frequent repetition of a word like “this”, “they”, “just”,
W RONG : Based on the distance d, we can com- or “then”.
pute... Avoid nonreferential use of “this”, “that”, “these”, “it”,
G OOD : Based on distance d, we can compute... and so on. Requiring explicit identification of what
17) Symbols in different formulae must be separated by “this” refers to enforces clarity of writing. Here is a
words. typical example of nonreferential “this”:
W RONG : Consider Sq , q < p. W RONG : Our experiments test several different
G OOD : Consider Sq , where q < p. environments and the algorithm does well in some
18) Protocol abbreviations typically do not take an article, but not all of them. This is important because...
even if the expanded version does. 25) Use strong verbs instead of lots of nouns and simple
G OOD : The Transmission Control Protocol delivers terms rather than fancy-sounding ones.
a byte stream. W RONG : make assumption
G OOD : TCP delivers a byte stream. G OOD : assume
G OOD : The TCP design has been successful. W RONG : is a function of
19) Check that abbreviations are always explained before G OOD : depends on
use. W RONG : is an illustration
G OOD : illustrates, shows
W RONG : IEEE has developed a standard for LR-
W RONG : is a requirement
WPANs.
G OOD : requires, needs to
G OOD : IEEE has developed a standard for Low-
W RONG : utilizes
Rate Wireless Personal Area Networks (LR-
G OOD : uses
WPANs).
W RONG : had difference
20) Sections and figures. G OOD : differed
W RONG : In section 5, we show...
G OOD : In Section 5, we show... II. VARIOUS T IPS
W RONG : We can see in figure 5...
G OOD : We can see in Figure 5... Here is a compilation of several tips on writing style [1],
but [2]:
G OOD : We will explain in the section that... 1) Avoid words like “this” or “also” in consecutive sen-
G OOD : We can see in the figures that... tences.
21) The text should make sense if we read through it 2) Use “Eq. 7”, not “Equation (7)”, unless you need to fill
omitting the titles of subsections. empty pages.
3) Don’t start sentences with “That’s because”.
W RONG : 2. Contour Integration. This technique,
4) In formal writing, contractions like “don’t”, “doesn’t”,
invented by Cauchy, defines...
“won’t” or “it’s” are generally avoided.
G OOD : 2. Contour Integration. The technique of
5) Be careful not to confuse “its” with “it’s” (it is).
integrating along curves in the complex plane,
6) Avoid cliches like “Recent advances in ...”.
invented by Cauchy, defines...
7) Units are always in roman font, never italics or LaTeX
22) Capitalize titles (also in bibtex references). To keep titles math mode. Units are set off by one (thin) space from
 3

the number. In LaTeX, use ˜ to avoid splitting number What does the reader expect next and why?
and units across two lines. \; or \, produces a thin space. 21) Get the attention of your readers immediately. Snappy ti-
8) Use “kb/s” or “Mb/s”, not “kbps” or “Mbps”—the latter tles, arresting first sentences, and lucid initial paragraphs
are not scientific units. Be careful to distinguish “Mb” are all methods of doing this.
(Megabit) and “MB” (Megabytes), in particular “kb” 22) Don’t use the same notation for two different things.
(1,000 bits) and “KB” (1,024 bytes). Conversely, use consistent notation for the same thing
9) Avoid excessive use of “i.e.”. Vary your expression: when it appears in several places.
“such as”, “this means that”, “because”, .... “I.e.” is not
III. T ERMS AND P HRASES TO AVOID IN A F ORMAL
the universal conjunction!
D OCUMENT
10) Remember that “i.e.” and “e.g.” are always followed by
a comma. Douglas Comer gives a list of terms and phrases to avoid in
11) Do not use ampersands (&) or slash-abbreviations (such a PhD dissertation [3]. It contains many examples that help to
as s/w or h/w) in formal writing. formalize scientific discourse, however, the current usage may
12) “Respectively” is preceded by a comma, as in “The light evolve to something more relaxed, so use your judgement.
bulbs lasted 10 and 100 days, respectively.” • adverbs
13) “Therefore”, “however”, “hence”, and “thus” are usually Mostly, they are very often overly used. Use strong
followed by a comma, as in “Therefore, our idea should words instead. For example, one could say, “Writers
not be implemented.” abuse adverbs.”
14) Never use “related works” unless you are talking about • jokes or puns
works of art. It’s “related work”.
They have no place in a formal document.
15) Do not refer to colors in graphs. Most people will print
• “bad”, “good”, “nice”, “terrible”, “stupid”
the paper on a monochrome (black and white) printer
and will have no idea what you are talking about. Make A scientific dissertation does not make moral judge-
sure that graph lines are easily distinguishable when ments. Use “incorrect/correct” to refer to factual
printing on a monochrome printer. correctness or errors. Use precise words or phrases
16) Use hyphens for concatenated words: “end-to-end ar- to assess quality (e.g., “method A requires less
chitecture”, “real-time operating system” (but “the com- computation than method B”). In general, one should
puter may analyze the results in real time”), “per-flow avoid all qualitative judgements.
queueing”, “flow-enabled”, “back-to-back”, ... • “true”, “pure”,
17) Don’t overuse dashes for separation, as they interrupt In the sense of “good” (it is judgemental).
the flow of words. Dashes may be appropriate where • “perfect”
you want to contrast thoughts very strongly or the dash
Nothing is.
part is a surprise of some sort. Think of it as a very long
• “an ideal solution”
pause when speaking. In many cases, a comma-separated
phrase works better. If you do use a dash, make sure You’re judging again.
it’s not a hyphen (- in LaTeX), but an em-dash (- - - in • “today”, “modern times”
LaTeX). Today is tomorrow’s yesterday.
G OOD : Learn to use the em-dash—it is a good • “soon”
friend. How soon? Later tonight? Next decade?
18) “While” instead of “and”, “but”, “although”. In general • “we were surprised to learn...”
“while” should be used only in the strict sense of “during
Even if you were, so what?
the time time”; so search for “while” in your text to see
• “seems”, “seemingly”,
if the sentence is about time or could be replaced with
“although”. It doesn’t matter how something appears.
19) Provide sufficient information in the caption of a figure • “would seem to show”
so that the reader does not need to look for the descrip- All that matters are the facts.
tion in the text. • “in terms of”
G OOD : Figure 5. Outstanding data (cwnd) and Usually vague.
queue size at R1 and R2. Equal buffer size
• “based on”, “X-based”, “as the basis of”
(30,000 B) on both sides. The downlink buffer
remains empty for significant periods of time, Careful; can be vague.
whereas the uplink buffer never empties. • “different”

20) Motivate the reader for what follows. Perhaps the most Does not mean “various”; different than what?
important principle of good writing is to keep the reader • “in light of”
uppermost in mind: What does the reader know so far? Colloquial.
 4

• “lots of” writing an AI thesis, talk to someone else: AI people

Vague and colloquial. have their own system of rules.
• “kind of” • “...a famous researcher...”
Vague and colloquial. It does not matter who said it or who did it. In fact,
• “type of” such statements prejudice the reader.
Vague and colloquial. • Be careful when using “few, most, all, any, every”.
• “something like” A dissertation is precise. If a sentence says “Most
computer systems contain X”, you must be able to
Vague and colloquial.
defend it. Are you sure you really know the facts?
• “just about” How many computers were built and sold yesterday?
Vague and colloquial. • “must”, “always”
• “number of” Absolutely?
Vague; do you mean “some”, “many”, or “most”? A • “should”
quantative statement is preferable.
Who says so?
• “due to”
• “proof”, “prove”
Colloquial.
Would a mathematician agree that it is a proof?
• “probably”
• “show”
Only if you know the statistical probability (if you
Used in the sense of “prove”. To “show” something,
do, state it quantatively).
you need to provide a formal proof.
• “obviously, clearly”
• “can/may”
Be careful: obvious/clear to everyone?
Your mother probably told you the difference.
• “simple”
Can have a negative connotation, as in “simpleton”. IV. W RITING THE I NTRODUCTION TO A PAPER
• “along with” There are a lot of good tips for writing papers [4], [5],
Just use “with”. [6], [7], [8], [9], [10], [11], [12], [13]. As “Abstract and
• “actually, really” Introduction are the most important sections” [14], here are
just some clues about writing introductions.
Define terms precisely to eliminate the need to
Unless there is a good argument against it, the Introduction
clarify.
should consist of five paragraphs answering the following five
• “the fact that” questions [4]:
Makes it a meta-sentence; rephrase. 1) What is the problem?
• “this”, “that” 2) Why is it interesting and important?
As in “This causes concern.” Reason: “this” can refer 3) Why is it hard? (e.g., why do naive approaches fail?)
to the subject of the previous sentence, the entire 4) Why has not it been solved before? (Or, what’s wrong
previous sentence, the entire previous paragraph, with previous proposed solutions? How does mine dif-
the entire previous section, etc. More important, it fer?)
can be interpreted in the concrete sense or in the 5) What are the key components of my approach and
meta-sense. For example, in: “X does Y. This means results? Also include any specific limitations.
...” the reader can assume “this” refers to Y or to Then, have a final paragraph or subsection: “Summary of
the fact that X does it. Even when restricted (e.g., Contributions”. It should list the major contributions in bullet
“this computation...”), the phrase is weak and often form, mentioning in which sections they can be found. This
ambiguous. material doubles as an outline of the rest of the paper, saving
• “You will read about...” space and eliminating redundancy.
The second person has no place in a formal disser-
tation. V. F INAL T RUTHS
• “I will describe...” Here are some handy maxims (notice that each sentence
The first person has no place in a formal dissertation. below is “wrong” in the sense that it illustrates bad usage).
If self-reference is essential, phrase it as “Section 10 However, remember that rules are made to be broken if it
describes...” improves clarity.
• Watch out for prepositions that sentences end with.
• “Hopefully, the program...”
• When dangling, consider your participles.
Computer programs do not hope, not unless they
• About them sentence fragments.
implement AI systems. By the way, if you are
• Make each pronoun agree with their antecedent.
 5

• Don’t use commas, which aren’t necessary. was submitted (e.g., “The foo algorithm works better
• Try to never split infinitives. than the the bar algorithm.”). Again, if you do not
Put yourself in the reader’s place! Ask yourself what the take enough care in presenting your work, why
reader knows and expects to see next at some point in the should I be compelled to recommend accepting it?
text. Sloppy papers give the indication of sloppy research.
Do organize and do not distract. As well as reflecting poorly on your work, such pa-
Bad writing comes from bad thinking, and bad thinking pers are often very difficult to read and understand.
never produces good writing. Presumably, the point of writing a paper is to convey
Easy writing makes damned hard reading. new information to the world. Therefore, there is
Less is more. major value in clear writing. If a referee can’t get
Mais malheur à l’auteur qui veut toujours instruire ! Le the point of the paper, or has a very difficult time
secret d’ennuyer est celui de tout dire. getting the main idea, there is a high likelihood that
the paper will be rejected.
Voltaire, De la Nature de l’Homme.
C. I Read Submissions in My Lazy Boy
VI. M ARK A LLMAN ’ S P LEA
I print out all papers I review. I read them in
Here are the comments of Mark Allman quoted inline [15].
my lazy boy1 . I scribble notes on them. Do not
They are specific to the Networking community, but most of
assume that I am looking at color plots! Color is
the remarks also concern other domains.
certainly nice and helps out immensely in conference
I have just completed my first (maybe only!) stint on presentations. But, color plots come out of my black
the program committee for the ACM SIGCOMM an- and white printer very badly in most cases. Besides,
nual symposium. While I have refereed many papers most conferences/journals publish black and white
over the years I have never before read so many sub- papers. Make it easy for my weary eyes to tell the
missions in such a short span of time (I filed 20 re- difference between the lines on your plots. Do not
views within about a month and read/skimmed many plot on a grey background. Make the font on the
more papers). This experience was eye-opening in plots readable without a magnifying glass. Help me
that I now realize that the community generates a out! Please!
large amount of junk. And, I don’t mean “junk” Often the plots are key to understanding the paper.
in the sense of bad ideas. In general, each paper And, at the risk of being redundant: If I cannot
I reviewed for SIGCOMM this year had at least a understand the paper I will recommend rejecting it.
nugget of an interesting idea. I mean “junk” in the
“it took me twice as long to read this as it should D. Don’t Waste My Time I
have” sense. In other words, the paper was sloppy. I
had a very difficult time trying to figure out what the You do not need to give an in-depth tutorial of all
paper was proposing or what the experiments were background material. I cannot give you a good rule
really showing. The following is a short list of easy of thumb to help you know how much background
ways to improve your research papers. material to include. It depends on the conference
Most of the following comments are general and I to which you are submitting, the maturity of the
presume would apply to any situation in which one background work, etc. But, I can tell you that
is attempting to publish scholarly work. However, there are some topics (e.g., TCP congestion control)
the last comment is directed specifically at the that are well understood by people attending any
internetworking research community. networking conference. You do not need to describe
the algorithms in painful detail. A short paragraph
A. Fire Up the Spell Checker, Please! re-capping the main ideas and providing references
I cannot spell either. I have, however, mastered to the original works should suffice.
ispell. It seems to me that everyone has access to
a good spell checker these days. Not using the spell E. Don’t Waste My Time II
checker just indicates that either you are sloppy or You do not need to provide me with an example
you do not care. If you’re sloppy with your writing of every kind of plotting strategy you use. When
why should I believe you are not sloppy with your you first show a plot with notation that I might not
experiments/analysis as well? If you don’t care about understand, explain the notation. But, using half a
your paper, why should I? page to show me a plot that adds nothing to my
understanding of the topic is a less than compelling
B. Sloppy Papers Are Long Shots
use of real estate.
Proofread your paper. I see lots of mistakes that
indicate the paper was not looked over before it 1a comfortable coach
 6

F. Don’t Waste My Time III error prone, I believe. As a reader of a paper, I often
Roughly stick to the page limit and the requested just want the high-order bit without slogging into the
format. Don’t make me try to guess how long your theorems and proofs to get said bit.
paper really is because you 1.75-spaced the paper L. On References
rather than double-spacing it as requested in the call
for papers. One common mistake that authors make is failing to
(Note to program committee chairs and journal edi- include relevant references. It is better to err on the
tors: This reviewer’s opinion is that if a submission side of citing too much related work.
is not obviously in the ball-park of the page limit it Next, when previous work must be criticized, do so
should be immediately returned to the authors until in a constructive manner. In other words, do not say
such time that it is in the ball-park.) something like “Zevon’s [1] work is easily shown
to be less effective than the work presented in this
G. Answer All the Easy Questions paper”. Mr. Zevon might be reviewing your paper!
If you need a constant for the foo algorithm and you Something more along the lines of: “Zevon [1] first
define c = 17, I want to know why you chose 17. introduced the foo algorithm for decreasing the size
Also, some items in papers jump out at reviewers of routing tables. We show a more robust algorithm,
as odd. For instance, I was recently reviewing a bar, that goes a step further and reduces the size of
paper that simulated a network with a bottleneck routing tables by 50% when compared to the foo
bandwidth of “about T1 rate”. The actual rate of algorithm.”
the link turned out to be something like 1.3 Mbps. Finally, do not overstate what a paper you are citing
This begs the obvious question of why the authors says. It is often better to stick with the author’s
used 1.3 Mbps rather than using T1 rate (˜1.5 Mbps). conclusions and not come up with your own. For
Even if the explanation you end up with is rough, instance, many times I see simple simulation studies
it is better than no explanation at all in the vast cited as “proving” the algorithm in question is safe
majority of the cases. and works well for the entire Internet. If one goes
back to the original paper the conclusion is not
H. Make My Life Easy nearly as bold, often claiming that the simple simula-
It is not a super-big deal, but it is always nice to get tions are an indication that the algorithm is workable
a paper that is easy to read. This includes things like and that future work on testing the algorithm in the
using a very clear font that is of readable size and Internet is needed. Overstating previous work may
numbering the pages. make your argument stronger if you are not caught.
But, when I notice such references they cast your
I. A Little Restraint, Please paper in a not-so-wonderful light in my eyes.
Your paper does not solve the entire world network-
M. Read These Papers
ing problems. Really. Many papers rub reviewers the
wrong way by making bold claims about solving I have noticed that a number of other reviewers are
all sorts of problems without any sort of evidence including pointers to the papers on TCP [16] and on
(or only weak evidence). It is much better to show simulating the Internet [17] in their reviews. They do
some restraint and perspective when discussing your not necessarily provide hard-and-fast rules for how
results. It is alright to solve small problems, or to to conduct research. However, they do provide some
only show preliminary results that may indicate a collected wisdom that should at least be considered
solution. Just don’t over claim what you have shown. in your research efforts.
In addition, the note by Partridge provides specific
J. Use Units guidance on getting a paper accepted for presenta-
“The length of our data transfers is 1” means noth- tion at SIGCOMM [18].
ing. 1 what? 1 packet? 1 KB? 1 minute? Sometimes I don’t want to make it sound like I think you
it is obvious from the context—sometimes it isn’t. need to write perfect papers when submitting to a
If you use units you don’t have to worry about it. conference or journal. A mis-spelling here or there,
an extra word, an extra page, etc. are all OK. They
K. We’re All Not Mathematicians will not be cause for your paper to be rejected. But,
Theorems are quite a necessary part of many papers. it is surprising how often authors make many of the
However, it is not necessary to give only a theorem mistakes listed above in a single submission. The
with no summary or context. Summaries aid me as bottom line is that if it is difficult for me to make it
a reviewer to gain an intuitive understanding before through your paper (because the writing is poor or
trying to dig into the math to make sure what you I can’t read the plots or whatever) there is a higher
have it correct—thus making my task easier and less chance that I will err on the side of recommending
 7

that the paper be rejected. Furthermore, the above [4] J. Widom, “Tips for Writing Technical Papers,” [Link]
items are easy things to fix when compared with the edu/~widom/[Link], 2006.
[5] D. Lemire, “Write Good Papers,” [Link]
hard task of conducting your investigation. As I said rules-to-write-a-good-research-paper, 2011.
in the opening, I believe most of the papers I review [6] H. Schulzrinne, “Writing Technical Articles,” [Link]
have a decent idea – yet I recommend rejecting the edu/~hgs/etc/[Link], 2011.
[7] S. Bloomer and M. Haas, “How to Write a Scientific Paper,” http://
vast majority of the papers I review. I believe many [Link]/files/PressPDFs/[Link], 2004.
papers would have a much better chance at being [8] M. Morrison, “Tips on Scientific Writing,” [Link]
accepted if the authors simply take a little more time ~morrison/Teaching/[Link], 2004.
[9] M. Faloutsos, “The Structure of Paper/Report in Systems,” [Link]
in preparing their manuscript. [Link]/~michalis/TECHWRITING/[Link], 2011.
[10] ——, “Good Sites for Researchers and Students,” [Link]
VII. C ONCLUSIONS ~michalis/[Link], 2011.
After reading all this stuff, you may think that some remarks [11] A. Zale, “How to Write Good,” [Link]
Zalefrontpage/[Link], 2004.
are contradictory. Yes, maybe—use your common sense to [12] S. Jones, “How to write a good research paper,” http:
choose the right word, or better check on the Internet. //[Link]/en-us/um/people/simonpj/papers/giving-a-talk/
Good writing requires practice. A good starting point is to [Link], 2004.
[13] D. Patterson, “Dave Patterson’s Writing Advice,” [Link]
read well written papers, analyze, and use them as inspiration. [Link]/~pattrsn/talks/[Link], 2011.
[14] M. Faloutsos, “Writing a Paper,” [Link]
ACKNOWLEDGEMENTS TECHWRITING/doing-research/[Link], 2011.
The report contains many examples gathered from various [15] M. Allman, “A Referee’s Plea,” [Link]
2001.
sources on the Internet. Their authors are gratefully acknowl- [16] M. Allman and A. Falk, “On the Effective Evaluation of TCP,” SIG-
edged: Mark Allman, Douglas Comer, Donald Knuth, Daniel COMM Comput. Commun. Rev., vol. 29, no. 5, 1999, [Link]
Lemire, David Patterson, and Henning Schulzrinne. org/mallman/papers/[Link].
[17] S. Floyd and V. Paxson, “Difficulties in Simulating the Internet,”
R EFERENCES IEEE/ACM Trans. Netw., vol. 9, August 2001, [Link]
papers/[Link].
[1] H. Schulzrinne, “Common Bugs in Writing,” [Link] [18] C. Partridge, “How to Increase the Chances Your Paper Is Ac-
homes/dec/[Link], 2011. cepted at ACM SIGCOMM,” SIGCOMM Comput. Commun. Rev.,
[2] D. Knuth, T. Larrabee, and P. Roberts, “Mathematical Writing,” http: vol. 28, no. 3, 1998, [Link]
//[Link]/reviewing-papers/knuth_mathematical_writing.pdf, jul98/[Link].
1987.
[3] D. Comer, “How to Write a Dissertation or Bedtime Reading for People
Who Do Not Have Time to Sleep,” [Link]
[Link], 2011.