Any policies around Dead links?

Question

I have been reviewing a lot of older posts (~10 years old) and there are quite a few with links that are either dead or links that point to sites that no longer contain the relevant information that the link was originally referring to.

In one case, I found a comparable site that contained the result the dead link cited and replaced the link in an edit. But in a lot of cases where someone is not citing a specific result, this is sometimes not possible.

Should the 10-year-old question be edited to remove the dead link? Is there any policy on the process to take when deciding on what to do with a dead or stale link? Surprisingly, I couldn't find an existing discussion on this. There is a small sub-discussion on this topic in this question:

What to do with links to arXiv front end?

As an example of what I encountered, this posting:

edges minus vertices

has a link to a pdf document at:

http://www.mathunion.org/ICM/ICM1994.2/Main/icm1994.2.1375.1383.ocr.pdf

and while the link is not dead (it still goes to the ICM Math Union website) it does not find this pdf file it tried to cite. The question could potentially have this extra tidbit of information removed without affecting the message quality of the question.

Should there be a policy? Could I propose an edit that I feel makes the most sense? I wouldn't know if my decision is breaking any rules (written or unwritten) around editing.

One rule around editing old posts is, don't overdo it. Each edit bumps the old post to the front page, bumping some newer post off the page. Three or four such bumps a day is the maximum I'd recommend. — Gerry Myerson, Commented Sep 16, 2020 at 22:37
Sometimes dead links can be recovered via the Wayback Machine. For example, the Math Union pdf in your post can be found at web.archive.org/web/20170811094846/http://www.mathunion.org/ICM/…. — jeq, Commented Sep 17, 2020 at 1:40
... And then, once you know what the linked paper is, you can sometimes replace it with a current link: for example link.springer.com/chapter/10.1007/978-3-0348-9078-6_132 (which in this instance is behind a paywall). — jeq, Commented Sep 17, 2020 at 1:49
@jeq better is not a link.springer.com url, but a doi url, in this case doi.org/10.1007/978-3-0348-9078-6_132 (found in the "about this paper" section of the page). And there is still the IMU free copy of volume 2 of the 1994 proceedings: mathunion.org/fileadmin/ICM/Proceedings/ICM1994.2/… — David Roberts, Commented Sep 17, 2020 at 3:49
I have edited the question with a sample display of info and links one might want to include if editing to fix links — David Roberts, Commented Sep 17, 2020 at 3:58
I would guess that at least some of the posts linking to archive.org have been edited to restore dead links, especially if you look among recently active posts. Using SEDE you can find specific edits where such links were added. — Martin Sleziak, Commented Sep 17, 2020 at 5:25
The first comment mentioned bumping too many posts, which was previously discussed here: Do we have an unofficial quota on how many old questions one should bump for minor edits in a single day? Some recommendations on the way to link to papers was discussed, for example, here: Edits with links to material under restricted access. — Martin Sleziak, Commented Sep 17, 2020 at 5:34
Dead links are the reason for my inveterate hostility to "here" and "this paper" links. If you say what your link is supposed to point to, then your link is still somewhat useful even if the target goes away. (Even better is, as @DavidRoberts says, to combine this strategy with links to URLs specifically intended to stay alive as long as possible. I only learned to do that recently, and I still sometimes forget.) — LSpice, Commented Sep 17, 2020 at 11:28
@GerryMyerson Is this "bumping" feature of the software something we could make a feature request around? I imagine being able to make a "bumpless" edit which requires moderator approval. — Steven Gubkin, Commented Sep 18, 2020 at 15:46
@StevenGubkin There are already some feature requests related to this: Minor edits, subject to review on this meta. (See also some of the questions linked there.) Allow non-bumping minor edits, but review them on /review on Meta Stack Exchange (and the linked questions). — Martin Sleziak, Commented Sep 18, 2020 at 17:38
(For anyone who doesn't follow all of @MartinSleziak's references, this is an extremely oft-mentioned request, and has been for a long time, so it seems fairly likely that SE has no intention of implementing it.) — LSpice, Commented Sep 18, 2020 at 22:15
And it's not just a matter of SE being stubborn and resisting the will of the people – there are good and sufficient reasons for the bumping (as developed in those links). — Gerry Myerson, Commented Sep 18, 2020 at 22:58

YCor · Accepted Answer · 2020-09-17 06:17:30Z

17

I guess I usually act as follows:

a) If I know how to fix the link (typically, link points to an identifiable paper), I either edit accordingly, or add the link as a comment (I might even provide 2 links, e.g., link to journal, often behind paywall, ideally using DOI, along with arXiv link); also I might add the journal reference when it wasn't initially.

b) If I don't know how to fix the link, I'll post a comment such as "the link is broken", with the hope that OP, if still active on MO, might fix the link. If not, still one can hope some other user knows how to fix/replace the link. At worst, it's useful that the reader is warned that the link is broken.

Anticipating on this, it's useful to check links and fix them before they become broken, for instance adding information that make them identifiable/retrievable, adding reference info, DOI. Also in case of external link to a picture, to embed the picture.

answered Sep 17, 2020 at 6:17

YCor

63.9k28 silver badges39 bronze badges

14

$\begingroup$ Yes, even just adding text giving a journal reference is preferable to link text that just says "this", with an opaque journal/publisher-specific url filled with numbers and letters. $\endgroup$
– David Roberts Mod
Commented Sep 17, 2020 at 6:51
$\begingroup$ Just out of curiosity, what's your policy for when to edit vs. when to add a comment? I try to stick with a comment for questions that aren't already on the front page, but then that's only useful for people who think to check the comments on a years-old dead link. $\endgroup$
– LSpice
Commented Sep 24, 2020 at 0:07
1

$\begingroup$ @LSpice not sure of a very rational answer, but I'd take into account: whether there are other improvements to make to the post (e.g. formatting), whether the post is already bumped, and if not, whether I believe it's worth being bumped (i.e., of interest to readers), whether I already recently bumped posts; possibly also the fact adding a link in a comment will be more or less visible according to whether there are already many comments. (I don't think I often bumped a question just to update a link.) $\endgroup$
– YCor
Commented Sep 24, 2020 at 0:23

Add a comment |

Stack Exchange Network

Any policies around Dead links?

1 Answer 1

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged
discussion
editing
hyperlinks
.

Linked

Any policies around Dead links?

1 Answer 1

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged discussioneditinghyperlinks.

Linked

Related

Not the answer you're looking for? Browse other questions tagged
discussion
editing
hyperlinks
.