How does publishing a paper work?

1. Format. This depends on the journal. Any journal should provide LaTeX and Word templates for their preferred style. It may not be necessary to use them for the initial submission — check their author guidelines and do exactly what they say. It doesn't matter what you think of their style: their journal, their rules. When you want something from someone, do not give them reasons to say no.

2. Yes: you have to write the paper. Get other people to read it and tell you how bad it is. Give presentations of it and field hostile questions. It builds character.

3. Different institutions may have their own rules about author order. A frequent convention is that major contributors go first, the head of the research group goes last, and everyone else is in the middle. Alphabetical order for ties.

4. You can only submit a paper to one place at a time. Multiple submission is pretty much never acceptable. If detected it may be grounds for instant rejection. You can also only publish it once. A further paper must be a sufficient advance on previous work.

5. The journal editor sometimes makes the decision himself, without sending it out. When he turns it down, this is called a "desk rejection", because it never gets past the editor's desk. That aside, anonymous peer review is the way it's done almost everywhere, although it is lately a subject of some contention. Sometimes it's doubly anonymous: the reviewers don't know who the authors are, although realistically, it is often not difficult to guess.

6. I haven't encountered conference-journals, but there does exist the opposite: the authors of the best papers at a conference (as judged by the conference committee) may be invited to brush them up for publication (usually with peer review) in a special issue of a journal.

7. Mm, not getting into that one, beyond mentioning public-access journals like the PLoS family. Beware, however, of predatory journals and conferences that accept everything for a fee, even randomly generated nonsense, but which nobody reads. Subscription journals aren't the only rent-seekers around.

8. Arxiv: Yes. Also biorXiv for biological sciences. There is some small hurdle to pass, I forget exactly what. In physics, I've heard, everyone reads arXiv to keep up, and eventual journal publication is just a formal stamp of approval that the experts don't need to see because they're the ones who gave it that stamp. For junk and crackpottery that can't even get into arXiv, there is viXra, which is only technically publishing, i.e. anyone can read it but nobody will.

9. Sci-hub: Yes. Well, exactly who is committing an offence and in what jurisdiction I'll leave to the lawyers. BTW, it's worth just googling the title of a paper, because you can often find a copy put online by the author, maybe a pre-publication version. You can also ask the author directly for a copy. They will usually be happy to do so. Also, not all journals are paywalled, e.g. PLoS, and some journals publish a mixture of open and paywalled articles.

10. Value of publishing: In the general sense of making your work public, yes, or why are you doing it? (Well, maybe to make money from your discoveries, which you don't want to reveal until you've made a huge pile, but that's not the usual way things go.) Putting it in a journal or on arXiv gives it permanence and a standard way to refer to it which you won't get by putting it on your blog. As for prestige, except for academic promotion committees who just add up the impact factors of the journals you've published in, the prestige comes from the work, not the venue. The most that the venue can do for you is bring the work to more people's attention. After that, the venue no longer matters.

I'm currently in the process of trying to convert a preprint into a journal article (and another draft into a preprint), so this is very near-mode for me right now. Restricting my comments to points where I can add something over the other answers (or disagree with them):

• 1. I personally quite like 2-column PDFs. At the very least they are far preferable to 1-column PDFs. :-P

• 2. Yes, but a lot of it is pretty important work. I'm generally the plots guy in my collaborations, so a lot of the extra work is coming up with the best visualisations I can for the data, which is valuable. Though there is then a lot of extra extra work of making sure all the visualisations use consistent colour schemes / legends / layouts etc, which is slow and tedious.

• 3. This is extremely field specific. In mathematics authors generally go alphabetically. In biology the person who did most of the lab work generally goes first, the person who did most of the analysis (if there is one) generally goes second, the first author's boss goes last, and everyone else goes in the middle. Sometimes you have awkward things where the first two or three authors get marked as "co-first-authors", where they did roughly equal amounts but someone has to go first. And so forth. In many arts/humanities subjects almost all papers are single-author so they haven't really worked this out yet. For most other fields I'm not familiar with the conventions.

• 5. My limited prior experience of peer-review has been frustratingly slow but otherwise broadly positive. Our paper was definitely better after peer review than it was before, and I expect this to be generally true and good. Stephan Guyenet had some recent comments on this that got linked by Slate Star Codex.

• 6. As others here have pointed out, I think it's generally the other way around.

• 7. Contrary (or possibly just less diplomatically than?) to Richard_Kennaway, I think the situation here is exactly as terrible as you describe. I consider the major journal publishers to be parasites of the lowest order. But! This does not necessarily apply to the editors who work for those companies, many of whom do useful work.

• 8. How much preprints substitute for papers varies hugely by field. Physics is an outlier. In biology it's becoming increasingly common but is still far from universal (but at least most of the important journals accept preprints). In other fields it's much rarer, and in some fields the best journals won't take your paper if you preprinted it first (though I think/hope this is dying out?).

• 10. Is "publishing" in this point supposed to be distinct from preprinting / publishing not-in-a-journal? Assuming it is, "allows future research to frictionlessly cite your findings" is increasingly a non-issue (preprints have DOIs and most journals let you cite them, at least in my field/s). On the other hand, here are two other useful roles served by publishing in journals.

  • Peer-review is pretty good. You need some kind of peer review, broadly defined. I think there are probably vastly better ways of doing it than the current system, but the current system is much better than what most places outside of academia have.
  • When you're deep in the maw of Goodhart's Law it's easy to forget that the metrics everyone is now savagely gaming were originally good metrics. In the absence of another system (arXiv + karma?) for legibly aggregating expert opinion on the quality of academic work, a journal hierarchy does contain useful information. I have never (yet) published in Nature or Science, but my experience of personal encounters with those who have is that they are generally (certain sexy topics excluded) very impressive.

What Richard_Kennaway said... and I would add (from experience):

1. Journal often have guides for authors, e.g. introduction, method, results, discussion sections - I have always ignored them. And even when they rejected the manuscript, the editor/referee never mentioned that particular element.

2. Yes and no. The additional work is simply annoying: filling forms, writing cover letters, arguing with the referees, typesetting, reading proofs. Or simply putting into words and sentences the images and symbols in your head. It seems like a lot of work, because it is outside of science, but I would definitely say that most work goes into the research itself.

3. I only worked in smaller groups (4 or less), but never had this problem. Most of the time the order was alphabetical exactly to preempt this, and department heads etc. were never included. The awkward feeling was there, when someone had to say "Alphabetical?", even though that's what everybody wanted. In some exceptional cases (thesis), a particular person was first, and others alphabetical.

4. The usual practice is to try for the ,,best'' possible journal in the given context, and if the paper is rejected, go to the second best etc. So one at a time, but you can try as many as you want.

X. Re: arXiv and prestige. Publishing/posting on arXiv does not give the same legitimacy as journals. And I know of cases, where before the manuscript gets published, other groups analyse the preprint, further the topic, and manage to publish first. Often it's a race to get the result registered under your name with the journal stamp seeming more official. Some are reluctant to disclose their results for that reason. To make things even more interesting, some referees insist that the manuscript must have an arXiv version first, so that it can be,,discussed in the community'' before it can even be submitted to a journal.

Finally, depending on the country, published articles count in career advancement (habilitation, tenure) while arXiv preprints might not. This is also where point 3. gets modified: sometimes it is necessary to obtain statements from co-authors regarding who did what, and the political negotiations rear their ugly head regardless of the published name order.