IMHO a one line answer that is not backed up by references and doesn't show significant research efforts doesn't fit to P.SE. An example (no offense to Thorbjørn who is actually one of the best contributors): How would you know if you've written readable and easily maintainable code?. Peer code reviews is indeed one of the best ways but this answer doesn't explain why, doesn't give any references, etc. It seems like the contributor didn't have the time to work on the answer. In that case I think that a comment would be more appropriate. Am I right or I'm missing something here?

share
4  
It's not an issue. :P –  Michael K Apr 17 '12 at 19:19
add comment

7 Answers 7

Writing a novel when a sentence will do is a bad answer.

share
add comment

I think that you are partially correct.

A one line answer without any demonstration of experiences or citations of references isn't an example of a great answer. I wouldn't go so far as to say it doesn't fit the site, but it would be something to look at. A short answer that doesn't share experiences or references could be a bad answer, or it could be indicative of a bad question that doesn't elicit the need to share experiences or references.

However, no answer should be posted as a comment, even if it is short enough. Comments are for asking clarification or providing additional information, not providing answers.

share
add comment

This MSO discussion sums it up well:

Of course, obviously the better answers will generally be longer than one line. To bulk up the content, you might consider doing things like:

  • linking to the appropriate documentation
  • including a code sample
  • suggesting a superior alternative approach
  • cautioning against some things that people in the same situation commonly do wrong
  • adding superfluous bulleted lists

There are many good answers that can be written in one sentence. However, the best answers explain the reasoning behind that one line. They explain the thought process that is required to arrive at the correct decision. They may include a personal experience that supports the conclusion reached. I've seen some good answers that take the approach the OP originally thought of and show how it leads to failure, and then present a better solution.

The best defense against harmful one-liners, as with many things on SE, is your voting powers. Don't be afraid to downvote the one-liners!

share
    
One-line answers should not be comments. If one was flagged, I'd be more likely to leave a comment asking the poster to elaborate and then give it some time, then delete the "answer" if it was not expanded upon within a day or so. –  Thomas Owens Apr 17 '12 at 19:31
    
@ThomasOwens You're right, I didn't think through that completely. One line answers might still be answers - just inferior ones compared to those that are fleshed out. –  Michael K Apr 17 '12 at 19:34
    
I absolutely agree with what you have now - use your powers of downvoting incomplete or sub-par answers, leave a comment asking for more information or elaboration, and flag so that moderators can remove the low-quality posts that aren't improved upon. –  Thomas Owens Apr 17 '12 at 19:36
add comment

Judge an answer by it's quality and correctness, not it's length.

Downvote it if the quality is below an acceptable standard, or the answer is incorrect, but don't discriminate against answers just because of their length.

Also don't forget, you can always leave a comment requesting the user expand their one-line answer into a more detailed response that covers the hows and the whys of the situation. I find many users are quite willing to do this if asked, and it will probably lead to them writing better quality answers in the future.

share
1  
I mentioned "a one line answer that is not backed up by references and doesn't show significant research". So it's not just the length. Also, it is unlikely that you can fit in one line references, personal experiences, etc. –  faif Apr 17 '12 at 20:42
    
@faif My answer would be the same for one-liners that don't contain references or show significant research. Judge the answer by it's quality and correctness, not by unrelated properties such as length, research shown, or references given. Sure these attributes might indicate a lower-quality answer, but not all the time. –  Rachel Apr 18 '12 at 13:28
add comment

I'd argue that an one-line-answer with many upvotes is an indication of a bad question. An answer that doesn't need citations, explanations or references to be deemed correct is most likely either an obvious answer ("yes, a compiler compiles") or a highly subjective/religious answer ("Linux is better"). Good questions do not provoke such answers, at least not with that many upvotes.

share
add comment

Note: I only found this discussion accidentally.

I find it interesting that you consider my answer to be an example of "...a one line answer that is not backed up by references and doesn't show significant research efforts..", and I believe this is not so.

Perhaps the misunderstanding is that you would like explicit mentioning of all the references and all the research effort in the answer.

In this particular case it is simply not necessary. The laconic style is intentional since lots of wording will just muddle the point being made, namely that you as the author of the code cannot tell, and the computer being a victim of the halting problem cannot either. You MUST ask another human (or yourself when you have forgotten sufficiently of what you knew when you wrote it). I believe the number of upvotes reflect that other programmers have enough experience to recognize this too.

So, in this case a one-liner is the most precise way of mentoring the asker.

share
1  
upvotes aren't a reliable indication for answer quality don't you think? For your case, it seems to be more compelling to point out that answer gained only 2 downvotes - with 2K+ views at generally downvote-happy site like ours this sends pretty strong signal that vast majority of readers were able to infer needed context without explicit pointers –  gnat Apr 18 '12 at 12:59
    
@thorborn I'm not saying that your answer is bad (I up-voted it too). I just don't consider it complete. I think that both sides can do better; You as the contributor can provide more information, and people who like the answer but don't consider it complete should ask for extensions (which is what I'm planning to do from now on). –  faif Apr 18 '12 at 17:22
    
@gnat The question is: Should we expect from the readers to infer the needed context? –  faif Apr 18 '12 at 17:27
    
@faif expanded my answer with clarification on that –  gnat Apr 18 '12 at 18:02
    
@gnat you may have a point about the downvotes being a more reliable indicator than the number of upvotes. –  user1249 Apr 18 '12 at 23:08
    
@faif a complete answer to this particular question would most likely be very, very long. –  user1249 Apr 18 '12 at 23:11
    
Thorbjørn I noticed that I am displayed as a "CW-author" for your answer. Given its revision history, this just doesn't feel right. I think if you add then remove <!-- some substantial amount of invisible text -->, this might bring not particularly smart CW heuristics back to normal. I would appreciate if you test that workaround –  gnat Apr 28 '12 at 20:51
    
@gnat, I'd rather that you bring it up with a moderator. –  user1249 Apr 28 '12 at 21:18
    
Thorbjørn after discussion with @ChrisF I brought this question at MSO Meta –  gnat Apr 29 '12 at 12:20
    
Thorbjørn per MSO question (now closed as duplicate), it looks like inserting <!-- some substantial amount of invisible text with more line breaks than were in my edits --> would be considered a reasonable workaround for a known limitation of CW-attribution algorithm –  gnat Apr 29 '12 at 14:11
    
@gnat And all this for you trying to "improve" on my answer... –  user1249 Apr 29 '12 at 14:19
    
this is not surprising - yet another point that "expanding high quality one-liner could do more harm than good" –  gnat Apr 29 '12 at 14:38
1  
@gnat I tested your workaround. The authorship was changed. –  user1249 May 9 '12 at 0:17
    
great to hear that! I felt really uncomfortable seeing my name under the post where my contribution was zero :) –  gnat May 9 '12 at 4:50
add comment

Your example looks like an awesome answer to me.

- How would you know if you've written readable and easily maintainable code?
- Your peer tells you after reviewing the code.

Not only does it answer the question, it also introduces important notions of peers, code review and feedback, ties them together and presents in a way that is easy to understand and use.

Answer style is not quite typical, in the sense that most other good answers I've seen at Programmers tend to be more verbose. But since stylistic consistency is not the only criteria I use, this doesn't worry me.

For the sake of completeness need to note that this particular one-liner is pretty exceptional; vast majority of other one-liners I deal with at P.SE are not like that - easy targets for downvote, flag or "expanding edit".


The question is: Should we expect from the readers to infer the needed context?

Good question. My take is

  1. in more typical cases it's better to just err on safe side and provide context explicitly
    Mainly because expanding average one-liner does no harm.
  2. in this particular case - yes, it looks reasonable to expect that from a reader
    Because expanding high quality one-liner could do more harm than good.

To avoid misunderstanding, I would like to make it perfectly clear that I support general guidelines you referred to. I believe that one-liners are typically not a good fit for Programmers. I also think that providing explicit answer context is as a rule beneficial for the answer quality.

The fact that I may occasionally encounter particular exceptional case (like it was with your example answer) won't stop me from following these guidelines.

share
add comment

You must log in to answer this question.

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