Publishing a Paper
Memo from a Chinese Economic Journal:
We have read your manuscript with boundless delight. If we were to
publish your paper, it would be impossible for us to publish any work of
And as it is unthinkable that in the next thousand years we shall
see its equal, we are, to our regret, compelled to return your divine composition,
and to beg you a thousand times to overlook our short sight and timidity.
When To Publish
A common progression:
- Give a talk about a neat idea
- Write up the talk into a short paper
- Give your friends/colleagues a draft to read
- Submit the draft to a conference [The paper will be 5-8 proceeding
pages, and will have obvious limitations. You will, however, get valuable
feedback and publicity.]
- Rework the draft, and do more research. Develop a journal paper.
- It is often not clear whether to publish soon, or wait for better results
and publish later.
- If you wait too long, you may wind up writing too long of a paper.
- If you try to publish too soon, you may get a reputation for shoddy
- It may be a bad idea to spend months revising the paper to be "perfect,"
because the referees may ask you to completely rewrite the paper.
What not to Publish
- fruitless speculations
- polemics and diatribes against a friend's error
- detailed working out of a known principle
Gauss discovered that a polygon with 65,537 sides are ruler-and-compass
constructible. Do not publish the details of the procedure -- it's been
- Do not publish a solution for a problem in 2 dimensions in 1992, 3
dimensions in 1993, and k dimensions in 199k.
- Do not publish your failures: I tried to prove so-and-so, and I couldn't;
here it is --- see?
Exception: Algorithm X solves cases A and B but fails on case
- A theory is worth studying if it has at least three distinct good hard
- If someone offered you $1000 for a paper you are about to submit, if
you instead tear it up, would you take the money? If so, do not submit
What to publish...
- Is it new?
- Is it true?
- Is it interesting?
Newness might mean that your paper contains:
- A new fact
- A new proof
- A new method (or algorithm)
Example of new fact:
Answering an open problem definitively (e.g., mechanically prove that
22**7 +1 can be factored).
Example of new proof:
Do in one paragraph what previously others took many laborious pages.
Example of new method:
Proof a well known fact with the same length proof as previous authors,
but use a new proof method that opens up a whole new approach to researchers.
Other constraints on what to publish...
You'd like to publish deep results, but it can be hard to judge
when something is deep:
- A complex proof does not imply a deep theorem
- A complex algorithm does not guarantee that the original problem was
- A short proof or simple algorithm does not mean the original theorem
or problem is trivial!
- Sometimes a "surprising" result can be a deep one -- even
if it is obvious after it is discovered.
What people publish is often guided by which topics are "hot,"
which is guided by the current
Halmos's informal survey:
- When asked what percentage of the math literature should have been
published, respondents answers ranged from 50% to only 2%.
- Less active, less motivated mathematicians reported about the great
pressure to publish, publish, publish.
- Give a talk without a proceedings paper
- Give a talk with a proceedings abstract
- Write an unrefereed proceedings paper
- Write a refereed proceedings paper
- Write a refereed Journal paper
- Invited talk
- Invited paper
Choosing a Journal
Questions to Ask...
- Which journal is appropriate?
- Compare the journal's objectives to your paper's content
- See which journals you cite in your references
- Ask the editor -- send an abstract or even the full paper
- Ask colleagues to read your paper and make suggestions
- What is the prestige and quality of the journal you select?
- High quality journals have lower acceptance rates, so publishing a
paper can be difficult
- Acceptance rates could be 50% to only 20%
- High quality journals are cited more often
- What is the circulation of the journal you select?
- If you publish in a new journal with small circulation, most libraries
might not even subscribe! So very few people will see the work.
- The first issue of most journals lists circulation figures
- Circulation may become irrelevant in electronic journals
- What is the audience of the journal you select?
Suppose you write a paper on a new parallel algorithm for searching.
Consider the prestige/circulation/audience of:
- Journal of the ACM (very formal, theory-oriented)
- IEEE Transactions on Parallel and Distributed Computing (readers
work in one discipline of CS/EE; formal research results, theory or experimentation)
- IEEE Computer (CS part of EE graduates, plus CS graduates; >20%
of paper must be tutorial)
- IEEE Spectrum (any EE graduate; no math allowed!)
More on choosing a journal...
- What is the delay from submission to publication?
- Some editors are better than others at nagging referees to finish their
reviews in 3-6 months.
- Some journals have a longer backlog for publication.
- Some journals publish the dates of submission and revision for each
paper, so you know the exact delay.
Note on Special Issues: Special issues have a lower delay, possible
more prestige (because they are selective), but are more competitive.
Submitting a Manuscript
Where to send the manuscript:
Look at the inside cover. Read the instructions for authors (usually
in one issue each year).
For some journals, submit to the editor; for others, and area editor.
Choose the area editor carefully (possibly consider sending e-mail if you
are unsure of whether you've chosen the correct area editor).
Be sure to get the address from the most recent issue! Editors
move, and new editors are appointed.
- the journal to which you are submitting the paper
- the title
- the address for correspondence -- just one author
- a note that your address might change (if this is a possibility)
- Double space the manuscript with wide margins!
- Submit single sided, stapled copies
- Consider putting a running head (with date) on each page
- Always put a date on the cover (helps identify different versions if
your manuscript is revised)
- Be sure that cross references are correct (especially with LaTeX)
Number of copies to send:
The instructions for authors will state the number of copies (4-10).
Keywords and Subject Classification:
Some journals require key words and subject classification. However,
it is a good habit to put these on all papers. These may be used by the
editor to assign the paper to an area editor, or to assign referees.
If your manuscript cites unpublished work:
You may consider enclosing copies of unpublished work that is critical
to the submitted manuscript, or copies of omitted proofs (which are likely
to be in technical reports).
Contact the editor if you do not receive an acknowledgement in a month.
The Refereeing Process
- If paper was submitted to editor-in-chief, editor assigns to area editor.
- Area editor assigns paper to two to five referees.
- Referees are asked to return paper in one month.
- Three to twelve months later, the area editor receives the reviews,
recommends action (accept, accept with minor changes, ask author to prepare
major revision, reject) to editor-in-chief.
- Editor-in-chief (but sometimes area editor) informs author.
- Author revises, if necessary. Returns manuscript (with original art
work) as instructed in step 5. Include letter stating how referee comments
were addressed. Possibly summarize differences between original and revised
- Editor may review paper on his/her own, or asks referees to review.
- Repeat steps 4 and 5. Possible repeat steps 6 and 7.
Submitting a Revised Manuscript
This may be the last manuscript version you submit. Be sure that it
is completely polished and error-free.
If you see a non-trivial error later (e.g., months later when you read
the page proofs), you may not be able to correct it -- so proofread
Send the editor:
- Original art in a separate envelope marked as such.
Consider printing the art larger than true size, so that the typesetter
can reduce it to a high quality figure.
- An original laser printer copy of your paper.
Why do this? The typesetter might treat a table or algorithm as artwork
and merely photograph it, rather than typesetting it!
- Memo detailing how you addressed the referees' comments.
The Role of Copy Editor
After your manuscript is accepted for publication, it goes to the copy
The copy editor will:
- do limited rewriting or reorganization of material to make the paper
- edit to correct grammar, syntax, and inconsistencies
- impose the journal style
The copy editor tries to minimize the number of changes to preserve
the author's style.
Typically you will receive the copy-marked manuscript when you are sent
proofs of the paper.
Original: "stable in a sense weaker than"
Copy-marked: "stable, and in a sense, weaker than"
Author's revision: "stable in a weaker sense than"
Checking the Proofs
Months after your manuscript is accepted, you'll receive page or galley
proofs showing your paper in typeset form.
Galley proofs: Sheets without page breaks
Page proofs: Sheets with page breaks (used with TeX)
You must return the proofs in 48 hours to 1 week, or the journal will
postpone publication of your paper.
Often the copy-marked manuscript is enclosed.
- compare line by line the marked manuscript to the proofs to find errors
- read the proofs by themselves "for meaning"
- check globally equation numbers to look for omissions
- check for consistent use of fonts and spacing for changes that the
copy editor made
- check the correctness of running heads
- check the correctness of the submission date, because it may help to
establish priority if similar work is published by other researchers
- check any tables, algorithms, or programs that were typeset
Greatest number of authors of a refereed paper:
P. Aarnio et al., Study of hadronic decays of the Z0 boson,
Phys. Lett. B, 240 (1990), pp. 271-282.
This paper has 547 authors from 29 institutions. The list of authors
and their addresses occupies three journal pages.
The shortest title:
Charles A. McCarthy, c_p, Israel J. Math, 5 (1967), pp. 249-271.