Lately I have been thinking about what is the right way to cite other papers (i.e., what to put in the bibliography entries, in the section which is typically called "References"). As many papers are available online nowadays, it seems to me that bibliographic references should include, whenever possible, a link to the paper being cited: this is supported by the PDF format, and is easy to do (at least for works typeset in LaTeX: e.g., with \href{} or \url{}). Yet, strangely, I have never seen a paper whose bibliography entries consistently featured URLs or hyperlinks to the works being cited. Is it good practice to include hyperlinks in bibliographies? If yes, why is essentially no one doing it?

One possible explanation is laziness on the part of the authors, but, strangely, there are still many papers whose bibliography entries include information about volumes, series, and page numbers, even though in my field (computer science) I have never seen anyone look up a paper using this information: people just search for papers with Google using the title and authors every time (in other words, they work around the non-existence of a hyperlink which should be there). Maybe the explanation is that existing bibliography management tools make it easier to manage page information than URLs? But is this just inertia, or is there a good reason?

A side question is assuming you want to have hyperlinks in your bibliography, which hyperlinks should you use? I have a preference to link to open-access versions of the cited papers (i.e., which aren't paywalled, so everyone can follow the link), but this sometimes conflicts with other desiderata, such as having a stable URL (e.g., not a PDF on some author's personal web site), having a link which indicates the DOI of the paper, etc. Which criteria should one preferably follow?

Another side question in this case is how should the links be formatted? One possibility is to write out the URLs explicitly, but this ends up taking a lot of space, and may be illegible, especially for long URLs. The other option is to use \href{}, i.e., put a hyperlink on the paper title, but this is probably not very discoverable, and it means information is lost whenever someone prints the paper. Which one is preferable, or are there other formatting options?

Edit: For papers submitted to publishers, indeed, it probably suffices to follow the publisher style (although the question remains of why so little publishers seem to be doing it). My question was more about preprints and author-final versions, prepared by the authors to be published online (e.g., on arXiv) independently from any specific publisher.

Related questions (with URLs ;)):

  • I've seen quite a few recent papers which cite preprints using the relevant arXiv identification number. Using this means the information won't be lost if the paper is printed. However, when citing a published paper it seems that the full citation is expected. Perhaps this is field dependent (I'm in theoretical physics)? – astronat Dec 19 '16 at 17:38
  • 2
    For me it sounds like a cool idea. It could be the DOI as a link, which eventually could direct to the original paper regardless if its link has changed. But I think it is sometimes done from the publisher's side (as often there are links using Crossref) and don't rely on the authors to provide the respective information. Actually, I remember that during production, sometimes they cannot find some references, so they definitely (at least some of the publishers) look for the actual reference papers and probably at the same time collect the information to link to them. – BioGeo Dec 19 '16 at 17:51
  • @BioGeo: The problem of linking to DOIs is that they tend to redirect to paywalled versions of the articles on publisher webpages, so there are many readers who cannot follow such links. One possibility is to use doai.io or oadoi.org, but I don't know whether this is the best option (also they can't always find the preprints when they exist). – a3nm Dec 19 '16 at 23:17
  • @a3nm I suspect that any other link (directing to the "free" version of an article) than DOI would be illegal, so I'm not sure any journal would prefer the options you mention (which I admit I hadn't heard) over the "real" one. – BioGeo Dec 20 '16 at 8:58
  • @BioGeo: Free versions of an article are generally not illegal. Many publishers allow authors to put free preprints online on repositories like arXiv, institutional repositories, and personal home pages. (See e.g. academia.stackexchange.com/q/72809). In fact I wasn't thinking about putting links to free copies of articles without a clear legal status, e.g., Sci-Hub; I was mostly thinking about legal preprints on arXiv and elsewhere. – a3nm Dec 20 '16 at 9:24

Yet, strangely, I have never seen a paper whose bibliography entries consistently featured URLs or hyperlinks to the works being cited.

Actually, there are some publishers which consistently report URLs and hyperlinks in the bibliography entries. An example is IOP, and here is an example paper where hyperlinks are used consistently in the bibliography. Note that, in my experience with this publisher, it's the copy editor who provides most of the links (e.g., DOIs), not the author.

One possible explanation is laziness on the part of the authors

Not necessarily, it can also depend on the publisher's style. For instance, IEEE usually doesn't put hyperlinks to journal papers (for the journals I know of), and I suspect that they would be probably removed if put by an author.

I have never seen anyone look up a paper using this information

Many publishers have search bars where you can enter just journal, volume and first page of the article. I frequently lookup papers in this way, and I know other people who do it.

assuming you want to have hyperlinks in your bibliography, which hyperlinks should you use?

Assuming you want to have hyperlinks in your bibliography, check first the publisher's guidelines: they might not be allowed or the required type might not coincide with your favourite choice. If I were to choose, I would use DOIs for journal papers to have a permanent address to the, possibly, most up to date version of the paper.

how should the links be formatted?

This, again, is usually decided by the publisher. In the IOP example above, most of the links are formatted as coloured hyperreferences by publisher's choice: this is probably the least invasive way, but it might not be supported by other publishers.

|improve this answer|||||
  • Thanks for your answer! However, as I should probably have clarified, my question was about what should be done about the author when they have the choice (e.g., a preprint posted on arXiv or on a personal webpage), not about a camera-ready version submitted to a publisher (for which indeed the publisher style guideline probably applies). Thanks for not making this clear; I edited the question accordingly. – a3nm Dec 19 '16 at 23:13
  • @a3nm Tomorrow I'll add something to address the edited question. Note, however, that if you aren't submitting to a journal, you can even put multiple links, e.g., doi + arXiv link. – Massimo Ortolano Dec 19 '16 at 23:21
  • Yes, on arXiv I have complete freedom to link to what I like. I'm just wondering what would be the right links to add, and whether it's a good idea (and why aren't more people doing it). Thanks! – a3nm Dec 20 '16 at 0:40

Maybe I'm a bit late here, but I do think it is among the best practices there can be. I also can't understand why journals' names are abbreviated. IMO providing links is really important for many reasons:

  1. It provides quick access to the cited material (either text or anything else).

  2. If a link is available, checking that source is (normally literally) one click away.

  3. If it isn't, it can be a challenge to get it. If the link is a DOI or other persistent link, it ensures access even if the final link changes. If the link is not persistent, it provides some transparency: the material was online at URL "x" at day "y", so if it's gone it's not your fault but you stated every information you had, and your reader now has some more info to work with when trying to reach that same source (looking for cached/archived versions, maybe looking for other people referencing the same URL who might quote some bigger part or even have an offline copy, who knows...).

  4. A reason for deliberately using a not-DOI URL: "People click on URLs that they understand and view as authoritative", so depending on the domain or subdomain of your cited material's URL, it could be a (not so) subtle persuasive use of that information in one's favor (either a publisher's, a government's, an international body's or any other institution's domain).

And these are just the ones that immediately come to mind now. But I think all the reasons that justify all the efforts to create the idea of DOI as a persistent link and maintain a whole infrastructure around it apply here as reasons to use it.

Actually, the real question could be why not use them (DOIs first, but links in general), and there are as many reasons for not using it as there are for abbreviating journals' names these days.

As to formatting, reason 1 above suggests not using "doi:10.1234/abcd1234" as this is not clickable (unless you manually make each one clickable by formatting it as link etc.), but rather something (automatically) clickable like "https://doi.org/10.1234/abcd1234".

And providing the last date you had successful access to it is useful as well, for what I mentioned in reason 3 above. (Some styles suggest stating the date of your first access. I can't quite see why would this be more important.)

|improve this answer|||||
  • 2
    Journal's name are often abbreviated in my field because there is often a page limit on papers which includes the bibliography, so saving space in the bibliography is important... – a3nm Jan 8 '19 at 7:27
  • @a3nm if there's a page limit, then saving a few lines matter, sure. But in digital times like these, why are there such page limits? And even if there are good reasons for page limits (and I agree there might really be), why not set the limit considering fully readable references (i.e., why allow this kind of abbreviation that simply prevent immediate human readability of the references, and BTW humans are the targets here)? It shouldn't increase the total in more than 1 or 2 pages. – ASR Jan 9 '19 at 14:08
  • ASR: I was just describing the (broken) status quo. I completely agree that it makes no sense for page limits to apply to references, – a3nm Jan 9 '19 at 14:33

One possible explanation is laziness on the part of the authors

Another is laziness on the part of the publishers. I normally get my Bibtex entries from Mathscinet, and they include url and doi fields. Yet, in my experience, they rarely get displayed in the final version of my papers. Most journal styles simply ignore them.

|improve this answer|||||

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.