I've mixed feelings about this.

An external link (book, article, code) could solve completely the OP's problem. But considering it a valid way to answering may lead quickly to ridiculous situations (I spare here some gedanken examples born out of my particular sense of humor), not to speak of the problem with future dead links, etc.

On the other hand, some topics are too cumbersome and/or have canonical and good answers already published and re-writing/condensing/analyzing them could be really time demanding.

So: when is appropriate flagging such answers as "not an answer" so they reincarnate as comments and when should we let them be in this site?

Please Note: There is a clear exemption not to be considered here when the OP explicitly asks for references.


When is it okay to answer by simply posting links to code/packages? (Almost dup)

Is it okay to answer a Stack Overflow question with a link?

2 Answers 2


Mathematica.Stackexchange has tagged along since its inception without having this problem. It only cropped up today because of a misinterpretation where the answerer thought the OP was asking "how can I implement algorithm xx" while the OP was really asking for a code-review and performance tuning of his implementation.

There is no doubt in my mind that that answer should really be a comment. Since it is the only example I can think of I say let's wait a while and see if any more questions get answered with a link only. Then we can determine if there are any legitimate use cases for link-only answers and decide whether we should have a policy on it or not.

Personally I will not post link-only answers. I would prefer to use a comment for such a minor contribution. I will also not vote for link-only answers but I will not downvote them either, for now.

  • Thanks for your input. In order to clarify, I'm not tying this question with any particular answer/poster. Historically I haven't seen too many of these answers, but we are luckily welcoming new users that are not aware of that "tradition", or even think otherwise. I'm not sure about what is the more convenient behavior in that cases (Flagging? Downvoting? Ignoring? Politely asking for enhancements? ...) Oct 1, 2014 at 16:07
  • @belisarius Absolutely; some may think I was rude to tie it to a specific question even, that should not reflect on you. Your question is general, my interpretation of it is not. I hope everyone understands that I only tied it to the specific question, the only one that I am aware of, to be able to make my point. Not to call anyone out.
    – C. E.
    Oct 1, 2014 at 17:09
  • @belisarius My opinions: Flagging) In order to get and explicit or implicit opinion from the moderator team. Downvoting) Never. Not constructive enough. Ignoring) Yes. But not always, it depends. Comments) Yes. But don't argue against a response you don't like. Comments aren't meant for that. Meta) This is where discussion belongs. I may think this particular discussion is premature, but it is still where it belongs.
    – C. E.
    Oct 1, 2014 at 17:12
  • Mostly agree except with Flagging) ... mod team. We shouldn't rely on them . Nah, joking. Seriously: the mod team helps a lot in keeping the site clean, but the policies are up to the community. I flagged a few answers because of this issue, but my feeling is that I'm burdening the mods with taking a decision that should be discussed previously here to support their actions. Oct 1, 2014 at 17:22
  • @belisarius Actually it doesn't reflect what I do in reality either, so scratch that.
    – C. E.
    Oct 1, 2014 at 17:50

Well, the future dead links problem is certainly something to consider, but otherwise, if the OP asks for a solution to the problem, and that solution has been published elsewhere, and takes three pages of code (which the coder should be credited with - a significant issue), I really don't see another solution (and since such a link answers the question, it really is an answer, not a comment). Notice that this is distinct from similar situation on, say mathoverflow, where you can say: the general idea is xxx, go read the paper for details: with code (especially implementing some well-known algorithm, like Baum-Welch) the details are the point.

I certainly agree that this can be abused, but so can anything else (notice also that on the math sites it seems to be more acceptable to refer to journal papers, which in many ways is more fraught, since these are frequently behind paywalls, and this is frequently solved by posting links to "pirated" versions). On this site it is somewhat less likely that the reference is to a journal article, which may be viewed as a good OR a bad thing.

You must log in to answer this question.

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