148

Over my years on StackExchange I've come to view answering SE questions as its own, highly specialized writing subgenre, with its own demands, and its own ideal format. By trial and error, and observation of highly upvoted answers, and of answers that I personally find useful, I've created my own standardized format for answering StackExchange questions --one that I have found productive across all the SE sites I participate upon. In addition to reliably garnering me votes, mastering this technique has also taught me useful things about writing in general, particularly persuasive writing.

I will provide my own answer below, but I'm also interested in hearing from other StackExchange members about how they write StackExchange answers, as well as if and how they vary their format for different sites.

4
  • 2
    Comments are not for extended discussion; this conversation, including discussion about site scope, has been moved to chat. Feb 26, 2018 at 16:20
  • 3
    Purely for the sake of symmetry: How to ask good questions: catb.org/~esr/faqs/smart-questions.html I wish more people would read this before posting questions anywhere. I have seen (and tried to answer) so many "It doesn't work" questions which don't give any details and don't even mention the environment/context.
    – Joe
    Jul 25, 2018 at 5:43
  • As there are lot of answers here, I am writing it here to let you know that the very first golden rule is mentioned here Oct 1, 2019 at 17:08
  • I just made a major edit to my answer here. Specifically in the context of edits
    – Honey
    Feb 25, 2020 at 14:45

20 Answers 20

93

Structure your answer properly

This is something that is relevant across all sites. You should be used to markdown and know at least the basics:

  • Using headings
  • Paragraphs and soft linebreaks
  • lists
    1. numbered
    2. and unnumbered

Put your most relevant points in the start and make them bold so that people will know at a first glance whether they share your opinion or not. Don't post a wall of text without any highlights and breaks. If you prefer not to use bold text and headings that is okay on most sites as long as you use paragraphs.

Know what you are talking about

If you don't know what you are talking about people will see through your lies quite fast. There are many experts and using false data or drawing false conclusions will be detected quite fast. This is especially true on very specialised sites, such as for example StackOverflow, but you will also find experts about any topic that you are talking about on sites like WorldBuilding.

Get to the point fast

Most sites prefer to get to the point as fast as possible and not beat around the bush. There are differences though. For example on WorldBuilding many people love to read stories. If you phrase your answer in a in-universe or in-character style you can basically talk as long as you want. On technical sites like StackOverflow people don't have the time to read through long texts. Writing is probably somewhere in between as people spend a bit more time on reading in general and normally don't have a problem that needs to be fixed now.

Adhere to traditional styles

On RPG you can find that many people reference guides and rules in a specific style, by mentioning the book name and page number and then using the citation markdown with added emphasis on the names of an ability/ rule/ ... and the important words on which they focus later.

Know the most important problems

On RPG, especially when talking about D&D, there is always the option to say "the DM decides". It's a non-sensical answer if you have spent time on the site as it's always applicable - the gamemaster can always change anything. That is not the kind of answers people are looking for. They want rules. Rulings can be made by anyone. Answers only referencing the option to make a ruling as the DM will get downvoted. But answers who explain why they would rule a certain way can get highly upvoted, as that is important information.

Adhere to standards about your tags

On WorldBuilding you should keep in mind that the square-cube law is not your friend when designing big creatures. The bigger, the more muscles and more dense bones are required, making the weight a bigger problem, which is why animals are capped at certain points in size. You know this rough rule if you spend some time looking through creature-design questions. It's an easy way to answer, but most of the time people expect more, as it's basically the same as "The DM can decide" on RPG. You have to discuss the ramifications of this usual problem. But it's nearly always a part that needs to be discussed, so you should spend some time explaining it.

Know the resources

SFF has a lot of story-identification questions and they have a long list of questions that you should try to answer as best as you can if you want to ask such a question. That also means that these are important things to keep in mind when answering. Furthermore you have to discuss every matching and mis-matching point in your answer. So knowing the standard things that are discussed makes it easier to compile a standard answer for story-identification questions.

Use pictures where appropriate

On WorldBuilding people love pictures. Especially sketches. If you have useful pictures you can easily get a highly voted answer. It's not always applicable, but on a theoretical and creative site like WorldBuilding it's far easier to use sketches to your advantage to illustrate your ideas, than on a technical site like StackOverflow where you need actual code or a site like Writing where you need citations or citation markdown for example phrases.

Know the site - and go through a trial and error phase with each one

Every site, and often every tag, has individual properties that you have to get used to. It takes some time and you will make mistakes. As long as you format your answer properly and know what you are talking about you should be fine on most sites, but getting to know the site and tags is important for high quality answers that are well-received across the wider community.

Adhere to the criteria and put them at the end together with other resources

As of writing this I have:

18
  • 65
    You might add something like, "Know when not to answer". Sometimes the most difficult response I have to a question is leaving alone for someone who I know would answer better than I can. Feb 20, 2018 at 16:44
  • 35
    @ToddWilcox Another good one is "post your answer as an answer, instead of as multiple comments" :D Feb 20, 2018 at 17:16
  • 5
    @ToddWilcox I can respect that. But the collective entirety of your comments seem to suggest a unified approach to SE answers that addresses the original question and doesn't duplicate other answers. Given that there do seem to be some principles you've identified that are valid across sites, you might rethink your opposition to the question. Feb 20, 2018 at 17:27
  • 5
    I'm not really sure that worldbuilding is representative of the overall SE network.
    – enderland
    Feb 20, 2018 at 17:47
  • 5
    Meh, this answer is to long. :P
    – DonQuiKong
    Feb 20, 2018 at 18:58
47

I like Secespitus' answer, and I also like Sphennings' point about actually answering the question. But I didn't see an answer which combined those two things, and addressed everything I've found important. So I'll write that answer.

Answer the Question

Not to steal any thunder from sphennings, but I believe this is probably the most important criteria for providing answers. It seems like an obvious rule, but I can fairly consistently find answers which do not answer the question, and instead explain how the OP should in fact be asking something else, or go in a different direction. While potentially quite helpful, these answers also assume the OP hasn't considered all the options. This is a problem when what they talk about is in fact something the OP is already aware of, and knows won't work.

A better practice would be to ask in a comment if the OP has considered X, Y, or Z. If not, and if X, Y, or Z would solve the problem, then post it as an answer. As leftaroundabout pointed out in his comment below, a likely better practice would be to open your answer by actually answering the question, and then explain why you think that answer isn't what the OP is really after. If they've considered that option, then you've already answered the question above. And if not, then your answer will likely help them, not to mention anyone else in the future who might have the same question.

Clarity

Next to answering the question, it is imperative that the OP can read the answer. This obviously starts with basic writing skills and a knowledge of English grammar, but extends to the use of headers and things like a tl;dr section.

Headers are great for breaking up long answers, as they make it a lot easier to read, and also easier to go back and check a particular section. Secespitus had a great point about getting to the answer quickly, and if that's not an option, a tl;dr at the top is a great addition. Instead of a tl;dr, I will sometimes simply supply a 'short answer', often only one word long, and then label the rest of my answer as my 'long answer'.

Be detailed

As Mark pointed out, answers coming in late on a hot question are usually buried under all the early answers and not seen as much. This certainly doesn't happen 100% of the time, as Todd Wilcox pointed out, but it does happen, especially on hot questions where there are lots of answers, and where those answers have at least 3 or so upvotes. Those two things tend to keep your answer at the bottom of the pile.

However, that doesn't mean that you should rush to create your answer. We're after quality here, so that's what you should deliver. I'll often be the first to spot a question, but because I take a long time to write my answer, I've seen as many as three answers come in while I'm writing. This has happened on this question, in fact.

The truth is that a lot of those quick answers are pretty short. They aren't wrong, but they also aren't as detailed as a longer quality answer. They don't take the time to fully explain a concept, and possibly even illustrate it. You can sacrifice being first for having better quality. It consistently gets my answer accepted.


These are the main criteria I adhere to. There are some obvious ones, like know what you are talking about, but I figure that those go without saying.

4
  • 1
    Thank you very much for addressing the "not an answer" responses. I absolutely hate those. The right thing to do is to answer the question they meant to ask and explain why it had to be answered that way, not just say, "You asked the wrong question. Do it better!"
    – Andrew
    Feb 21, 2018 at 16:57
  • 3
    I would like to add that when you do answer the question (and this goes especially for the more technical sites), please answer the question conceptually/abstractly first. I hate when I'm looking for a solution to my problem and I come across an overly specific question & answer combo that's not quite my problem but related. If the answerer would answer the question generically first so that others could learn the nature of their problem better, rather than having to sit there and dig through some complicated, specific answer to figure this out for themselves, it would help so much.
    – Andrew
    Feb 21, 2018 at 16:59
  • 3
    I agree that answer the question is very important, but I'm not sure about “better practice would be to ask in a comment if the OP has considered X, Y, or Z. If not, and if X, Y, or Z would solve the problem”. This can easily lead to the question becoming quite different, which is bad if the question was answerable in the original form. Better: start the answer with answering the question, literally, then add an explanation why it was the wrong question to ask (even if the OP had already considered alternatives, somebody else stubling on the question might not). Feb 25, 2018 at 21:11
  • @leftaroundabout Good point on that. I've suggested more or less the same myself at times - it must have slipped my mind when writing the answer. Feb 25, 2018 at 23:58
40

I try to always answer in 3 paragraphs whenever possible. Less is often too little for a substantive answer, and more becomes less and less likely for people to read. The first paragraph should always be the most direct answer to the main question in the original post, as asked, with a minimum of editorializing. It should generally cite a reputable source, with a link, but also recapitulate the information that is linked. Sometimes I will boldface a key idea, no more than one per paragraph at the most. The most important thing is to remember that this is audience-oriented, persuasive/informative writing, it must be simultaneously substantive, useful and compelling.

My second paragraph is a place to expand on my answer, or to answer the question that I really think should have been asked, instead of the one that actually was asked. Even when doing so, however, I think it's important to treat every question as legitimate, as reasonable and as asked in good faith (otherwise you shouldn't be answering it). It's also important to NOT make the querent feel stupid or ignorant for not having your level of knowledge. Quite often it's not the content that makes for a highly upvoted answer, it's the structure and the tone.

Does it even make sense to ask for "the" way to write SE answers, since they are all so different? I do think there are some generally useful structures that aren't content-area specific. For example, if I want to criticize the question itself or offer original or controversial ideas, I reserve those for the last paragraph. I've found that the very same things that enrage people, or cause them to dismiss an answer when contained in the first paragraph, are often accepted without comment in the last paragraph, especially if they are clearly labeled as "original" or "editorial." It's worth noting that the exact same conceptual content can garner very different reactions, depending on how it is presented, and in what order. Finally, I nearly always revise frequently, and to do my best to accept all critiques in the comments non-defensively, and to respond to them productively (generally through revisions).

8
  • 22
    (counts 1, 2, 3 ) Upvote for following your own suggestions precisely.
    – Criggie
    Feb 20, 2018 at 23:26
  • 8
    Including the bit about revising frequently Feb 21, 2018 at 1:03
  • 6
    @Wildcard Thanks for giving me a critical comment to respond to :) Feb 21, 2018 at 3:34
  • 4
    This is pretty much my answer. However, I also like the actual tl;dr answer to be quite visible near the top, in boldface if nessecary. Basic structure should be "Answer. Reasoning behind answer. Answer." Much like the old "Tell em what you're gonna tell em, tell em, then tell em what you told em." advice for presentations.
    – T.E.D.
    Feb 21, 2018 at 22:02
  • 3
    Very disappointed that you didn't take the opportunity to troll your own answer, by providing a very short 4th paragraph.
    – jpaugh
    Feb 23, 2018 at 22:51
31

Rule 0: Answer the question

If you're not answering the question then you're wasting your own time and others. Everything you do in an answer should be in support of this goal.

Most answers tend to fall into three categories Yes, because..., No, because..., and Let me explain this concept to you... Start with a short answer to the summarized version of the question.

Rule 1: Sell your answer

Now that you've "answered the question" you need to write to convince people that your answer is correct. Add more information, and explore nuances brought up in the full question's text. All the extra detail should be in support of thesis of the answer.

Answers should be formatted in a way that is simple to read. Don't have a wall of unformatted text. Similarly don't distract from the text by overformatting it. Avoid pictures if at all possible. They look nice but they break up the text and make it harder for people to understand your point.

Rule 2: Have fun

You aren't paid to do this. At the end of the day you're competing for arbitrary internet points. If you find yourself arguing in the comments or getting frustrated, step back, take a deep breath, and go and do something fun. If you're not enjoying yourself it's going to show in your answers.

If you can add a bit of levity, do so, but only if it doesn't detract from the readability of the answer. It's very frustrating reading through walls of text that ultimately have nothing to do with the question being asked.


Some sites allow challenges to the premise of the question in which case your answer will be of the form Don't do x, because... The same basic rules apply but it selling the answer becomes even more critical since you need to defend why you are breaking breaking Rule 0.

Frame challenges are hard to do correctly. At a minimum they need to explain why the expected answers to the question won't meet the actual needs of the OP. Ideally they should also address the actual needs of the OP.

5
  • 3
    As an addition to Rule 0, on some sites it is acceptable to answer "How do I do X" questions as "You should not do X because...". This most certainly isn't "answering the question", or at least the question being asked, but will still usually provide good advice. Feb 20, 2018 at 21:02
  • 5
    @CaptainMan I'd argue that that's still answering the question. Frame challenges are harder to do well than answering well asked questions. They're also inconsistently received by the community. They're either incredibly popular or highly contentious.
    – sphennings
    Feb 20, 2018 at 21:19
  • 4
    @CaptainMan concur - "addressing the underlying problem" is sometimes better than answering the stated question because its misleading due to faulty understanding. IE "My room light lamps keep blowing so how do I change the lightswitch?" the leap from the cause to the question-statement is missing.
    – Criggie
    Feb 20, 2018 at 23:25
  • Good answer, but I disagree about having to answer the question or not answer at all. I have "answered" a number of questions which have had no answer for many days with information that may help the OP find an answer even if I don't know the actual answer. IMHO, a bit of guidance and support is way better than silence. I do this when I have more to say than what would fit into a comment or my response requires structured text which comments do not support.
    – Joe
    Jul 25, 2018 at 4:06
  • I also think it's quite valid to post an "answer" which adds nuance or covers additional concerns not covered by the other answers - again when the response won't fit into a comment because of size or format.
    – Joe
    Jul 25, 2018 at 4:10
24

That is an interesting question and and interesting observation. Is the Stack Exchange answer a distinct genre? Or perhaps more broadly is the QA site answer a distinct genre. If it is, I think it is an example of a more pervasive genre that was created by the Web, which we might call persistent conversation.

Persistent conversation is a type of communication that is not as formal, not as researched, verified, and worked, as you would find in a formal publication such as a journal or a book. It is a contribution to a forum where other voices are being heard and it is subject to commentary and editing, and where it will remain on permanent public display. Therefore a great deal more thought and discretion goes into the composition than you would give to a chance utterance in private conversation. It is this status as new genre that leads to so many debates about the appropriate level of formality, correctness, and editing that should go into composing any Web posting.

This certainly affects how people write SE answers. Yes, they put some thought into what they say and pay some attention to how they say it (some more than others, clearly). But they publish with far less care and polish than they would if they were writing a book or a journal article. I think there are a couple of contributing factors to this:

  1. You know that if you miss something you will have the chance to edit it and if you get something wrong or miss something, someone will correct you and you will be able to fix the problem.

  2. To the extent that you are interested in gaining reputation, speed matters. We would like to think that a great answer posted a week after the question was asked would slowly get voted to the top, but we all know it won't. First to answer is a big advantage in the reputation sweepstakes, and this creates a clear incentive to get an answer out quickly.

That said, I think I have a distinct approach to answering questions (on this site, anyway) and that is to tie the answer back to a general principle. You should do X because of principle Y which is based on human characteristic Z.

One of the consequences of this approach is that I often end up stating the general principle with very little caveat or refinement. This sometimes leads to disagreements that would take far more room than this format permits to fully explore and resolve.

On the other hand, I personally feel that merely making a suggestion without tying it back to a principle is of limited value. Such suggestions have no provenance. They don't really meet the SE criteria for providing objective researched answers.

But those criteria, which were developed for answering programming questions, don't really fit writing questions. Programming is highly analytic, breaking problems down into smaller and smaller pieces. Writing is synthetic, bringing all the pieces together to work as a whole. This naturally means that any writing question has very deep and complex roots. To deal with them in this format requires some means of simplifying the complex set of principles and practices that would inform a full answer.

10
  • 9
    "We would like to think that a great answer posted a week after the question was asked would slowly get voted to the top, but we all know it won't." Actually I have had this happen to me several times. I've also had the opposite happen - where my answer was first and took an early lead but then was surpassed by a much later answer. Also I've had my answer unaccepted in lieu of another answer after several months. Feb 20, 2018 at 16:06
  • 2
    @ToddWilcox I'm glad to hear it. I suspect it may happen more on stacks with more concrete questions and answers that are perhaps more searchable than questions and answers here. On this stack, it seems that almost all the activity is on the leading edge. I will pick up a point of two from older questions now and then, but is it a mere dribble.
    – user16226
    Feb 20, 2018 at 16:09
  • 3
    Interestingly enough, what's probably my most upvoted answer across the network is a relatively trivial answer to a simple question, which keeps attracting the occasional upvote, despite the fact that I posted it a little over five years ago. Other answers that I have put very much more thought and time into linger in obscurity with barely a vote either way.
    – user
    Feb 20, 2018 at 16:19
  • 8
    @MichaelKjörling One big factor in number of upvotes on content is the popularity of the Stack itself. My highest upvoted answer is on Worldbuilding and I am sure it's because Worldbuilding attracts a lot of traffic. You don't have to know anything about any particular topic like Java, networks, or Law, HNQs from WB are often tantalizingly worded, and I expect most SE users love to brainstorm and fantasize with their spare mental clock cycles, so even a mediocre WorldBuilding answer will blow away the best possible Music Practice & Theory answer (to say nothing of Sound Design (beta)). Feb 20, 2018 at 17:15
  • 3
    @KRyan I wonder if it has to do with the consistency of questions. There is always more than one way to ask a question, even on the most concrete topic. But on SO, for instance, most of the questions use the same CS jargon and so if you are searching Google for that question, your terms are likely to match and old questions should actually come up more frequently than new ones. But I don't think that happens nearly as much on Writers because the way people phrase their questions is much more diverse. It would be interesting to compare traffic sources for different stacks.
    – user16226
    Feb 21, 2018 at 20:16
18

I was asked to provide this answer by the original poster. It was originally a comment.

  1. As I understand the SE model, questions are to be focused and there will be 'a best answer' among the responses. Viewers can vote and the asker chooses the answer.

That's fine, but given the diverse nature of human cultures, languages (and their barriers), human brains, even one's own thoughts varying between different times of day ... the effect of persuasion on belief, and so on ... as a model this is less than perfect. Again, it is fine, but I would personally caution against any tendency to think there is "an answer" to a question. There are many that are valid.

Communication (and perhaps writing SE in particular, as visitors are interested precisely in the written form of communication) is fluid by nature, it is not math. There is no proof.

So. I generally disagree that there is 'one' correct answer to a question (or a best approach to answering a question). I dislike the tendency towards lowest common denominators, which internet 'groups' (and any human grouping) tend to suffer from at some point.

As a result, I sometimes provide answers to fill a perceived void among existing answers (or, as a nuance of this, to respond to part of the question tangential to its main thrust.) I'm not often looking to win any argument, (although I am gunning for some new badges at the moment) and I benefit from reading some other contributions.

I firmly believe that an array of answers that provokes thought is the best outcome. To that end, I'm not certain of the goal of this particular question... It seems designed to curtail certain types of answers.

  1. Sometimes my answers are more selfish, in that I'd like to just work through what I think about a topic. Of these, I post a subset if they seem to have any use.

  2. In terms of style, formatting and links are good. I'll add in a link here about the possible value of considering diverse opinions.


(Incidental post script thought - I have noticed that when an answer is chosen, it often garners more up votes, and at a faster rate than previously. Maybe people feel good about upvoting the winner? Similarly, out of curiosity (and because I am gunning for new badges!) I edited one of my old questions, which brought it back into circulation, and got a few more up votes on the question. Sadly, I need 100 up votes to get that gold badge. These thoughts are somewhat tongue - in - cheek regarding functions on SE.)

11
  • 4
    To be honest, I'm not really seeking a definitive answer, I'm looking to create a forum for useful knowledge-sharing. Shh! Don't tell anyone :) Feb 20, 2018 at 16:57
  • 5
    Too late, @ChrisSunami, the software has already flagged your comment for automated peer review by the matrix. (Did I make scrambled eggs out of too many metaphors, now?)
    – user
    Feb 20, 2018 at 18:31
  • 1
    Referencing points is a distraction in an answer. It implies answerer is only doing it for the reward rather than wanting to help by sharing knowledge. +1 for showing that.
    – Criggie
    Feb 20, 2018 at 23:27
  • 2
    The chosen answer is shown first, so it is most likely to be read. After that probability of being read goes with number of votes the answer already has because that is how answers are ordered. Length and structure might also have an impact, people avoid reading walls of text. People tend to only up or down vote answers they have actually read, which is probably good, so the odds of answer being read match the odds of getting votes pretty well, I'd expect. Stack Exchange probably has analytics on this? Feb 21, 2018 at 1:15
  • 2
    @KRyan True, but the idea that questions have one best answer is itself usually wrong. Even in "hard" sites like physics or math there are usually several alternative ways of explaining the correct solution. You can't really expect a single answer to cover all of them and while you can work out which works best for the person asking the question, guessing which explanation works best for the random guy googling the answer two years from now is... difficult. So you want multiple correct but different answers. Feb 21, 2018 at 20:35
17

Be brief, clear and correct.

(Note how my top answer to this question overtook the more complete accepted answer).

5
  • 1
    Right. I have no patience to read lengthy questions or answers. Writers on this forum seem to have infinite time to polish the turd and assume their audience does too. Seems unique to this forum.
    – BSalita
    Feb 21, 2018 at 10:06
  • 2
    Note that a useful consequence of this is that if you insist on writing a novel, you can use headings in such a way that only reading the headings is also an answer! Feb 21, 2018 at 14:45
  • Sometimes short answers are called for. Other times, long answers are called for. I strongly doubt that I would have gotten 100 points on this answer if it had been a one-liner.
    – celtschk
    May 16, 2018 at 4:39
  • 1
    @celtschk yes some answers must be long, of necessity. But it still should be as short as possible while still answering the question clearly and correctly (within reason).
    – MGOwen
    May 17, 2018 at 2:16
  • @MGOwen: What you should avoid is not being long, it's being long-winded. You don't need to be brief, you need to be concise. And BTW, I think your linked answer overtook the accepted one not because of its briefness, but because it actually provided the code to do it.
    – celtschk
    May 23, 2018 at 10:33
16

Many answers have covered various points, but I've yet to see a detailed answer for writing a scientific answer (i.e. on Math, CS, Physics, etc.) on SE that I like. (Disclaimer: I'm only experienced in math and CS, but I think most of this is useful for the other (emperical) sciences). I'm assuming we're not talking about CSTheory or Mathoverflow as those concern expert to expert communication, which is a different beast entirely. I'll provide a list of tips. Some of these tips might even be useful for technology or non-technical answers.

Know your audience

Most people asking questions are completely uninitiated or at least a novice in your field. If they would be experts, it is likely they are able to research the question on their own, or have other resources to ask (e.g. colleagues). From this, we get to the following:

Always offer (at least) an intuitive explanation

Ideally, your answer should provide insights into the question matter to the asker. As we have established that the asker is often a novice, a long, technical, mathematical proof is probably less than helpful. However, sometimes, technical remarks are necessary. These have to be accompanied by high-level ideas or intuition. (e.g. don't merely show the complicated proof step, explain why you're performing this step) The precise level of intuition or non-technical explanation of the technical ideas involved is up to personal preference, but a good answer should provide at least some.

Jargon

A certain amount of jargon is unavoidable, but it should be used only when it makes the answer clearer. When in doubt whether a term is known by people able to ask the question, provide a definition/explanation. For instance, when answering a question about graph theory, you may assume 'edge' and 'vertex' are known terms. (But you might have to define them/refer to definitions in case graph theory is only a part of the answer, not of the question).

Be aware that there is a cultural component to jargon. For instance, I've long been in the dark what an 'Abelian group' was, even after taking a (elementary) course in algebra. When I finally decided to look it up, I was quite surprised that it was 'merely' a commutative group. ('Why didn't these people just say so?', is what I thought. They probably thought that was exactly what they were doing.)

Mathematical Notation

A very specific kind of jargon, considering the Latin alphabet insufficient for its purpose. It is a very powerful tool in communicating precise arguments and definitions. However, like most powerful tools, using them badly leads to terrible results. As a general rule, prefer formal English to mathematical notation if it does not make the answer less readable. Compare "For all state machines x,y such that ..., we have that x simulates y" with "\forall_{x,y}\in \mathcal{S}: P(...) => (X Sim Y)".

State your business, at the start

Some questions are such that there exist clear objective answers, but nevertheless can have different, useful approaches. It is a good idea to state at the top of your post what exactly you're going to do and how it differs from answers already given (if any).

Use images, if you think they help

This title should speak for itself. An image, however, shouldn't. Always explain precisely what your image depicts (it is possible to do this partially inside the image, but only partially). While an image may clarify your argument, it is not a replacement for it. Do not use too many images, though, this makes your post harder to read eventually.

Be clear about what you know, and what you don't

Sometimes, you will only be able to answer a part of the question. First of all, you should indicate this (see 'State your business'). Second, it is useful to give a reason why you don't know something or why you cannot explain here. A particular case is a question that is simply too complicated to answer directly. It might be the case that an external resource (e.g. some published paper) does provide (the whole) story. In that case, refer and link to the resource, but do at least try to make a brief summary of the relevant portions of the resource your answer needs.

Think further than the questioner

We've already established that you likely know more about the subject that the questioner. This means that it is likely that you could have asked an even better question! Don't be afraid to add some more comments that don't directly answer the question or answer questions that easily arise from your question. It is a good idea to visually separate this from the rest of the question, with a horizontal line, for example.

'Credentials'

My reputation on my 'main' site is there for all to see, but I'd prefer you take my words for what they are, instead of based on who I am.

That said, I think an example of a good technical answer given by myself is this: reservoir sampling algorithm probability

1
  • I think almost all of this applies to less technical fields as well.
    – Joe
    Jul 25, 2018 at 5:12
16

In addition to the many good suggestions here:

Pay particular attention to the level of sophistication the OP shows in the question. Remember that you're writing primarily for the OP (also for the community and future visitors, but that's secondary). For example, on Math.SE an answer that suits a pro may be well beyond an undergraduate, or a high school student. There I try to teach as well as answer.

I wonder what the balance is between usefulness to OP vs. to the community. I suspect it varies from site to site. On math.se many of the questions are quite particular. On tex.se they seem to be more general purpose. There's a nice small research project waiting to be done, looking at the statistics on questions, votes and views.

7
  • 3
    This is an interesting question --I always view myself as writing primarily for the general audience, and secondarily for the OP. Feb 20, 2018 at 21:03
  • 1
    Yep. Except for stuff extremely specific to the questioner's situation, the number of people who find their way to the question by Google is generally going to be more than one. So the general audience is almost always more important.
    – MGOwen
    Feb 21, 2018 at 4:43
  • @ChrisSunami See my edit. Feb 21, 2018 at 13:11
  • @MGOwen See my edit. Feb 21, 2018 at 13:11
  • On StackOverflow.com, if an answer would pretty much only be useful to the OP, I don't usually consider it worth answering. I might post a comment to point the OP in the right direction, but only questions with future value deserve answers, as a general rule. Not posting an answer makes it more likely for the question to be deleted in the future instead of cluttering up SO search results with yet another crappy debugging question forever. Feb 22, 2018 at 21:56
15

If I answer late, I only answer if I can think of something not already said. With an exception for things said, but I think poorly, or weighted with unnecessary baggage. I +1 anything said that I would have also said, unless that answer includes something I definitely disagree with.

For normally answered questions, recently posted, I have no definite structure. I am a discovery writer, for starters, and my approach to teaching has been successful and loosely structured: I figure out what the student is thinking, where they are going wrong, and address that issue.

I don't think it makes a difference what the subject is: Student's asking questions, especially things I'd expect them to know, have some fundamental misunderstanding that needs to be corrected. Sometimes that is shallow, sometimes it is deep.

So my objective is always to see if I can find what is probably their fundamental misunderstanding about the mechanics of writing, or science, or culture. I take my best guess at what are they missing about writing something entertaining, why they had to ask the question, and I address that.

I only tie it back to some kind of fundamental if I think they have a fundamental misunderstanding. If they just aren't clear about something fairly simple, I'll just clear that up.

So my structure depends on that. Headings don't hurt, but might be silly in a few paragraphs. Citation links don't hurt, but I may not have any, or cannot quickly find what I am looking for.

Quotes from existing publications might help, but preferably only if I can cut and paste them from somewhere.

Original examples might help, if I can think of something fairly short.

I am a discovery writer, I do not write from an outline or template. I do revise, I will often write a fairly lengthy answer and in the process realize what the headline answer should be, and go back and stuff it in, and perhaps rewrite and rearrange to make that the central idea of my response.

1
  • (See, this is why i think you and i don't have a problem learning names.)
    – SFWriter
    Feb 20, 2018 at 20:28
14

It has been discussed repeatedly on Meta.SE that the first few answers will almost always receive the most votes, e.g. here. For two reasons:

  1. People read from the top. If the top answers are good, they vote for them. If later answers (lower on the page) are better than the upper answers, they don't retract their votes for the higher answers. But they only vote for lower answers, if those are better than the higher answers. This leads to the effect that mediocre but higher answers keep their rep, while only exceedingly outstanding good answers lower on the page receive any rep at all.

  2. Most people visit a question only once. Questions, therefore, receive most of their traffic (and votes) while they rank high on the front page. Answers given after a question has moved down to the bottom of the front page are not seen by as many people and cannot garner as many votes as those posted immediately.

For that reason:

Be the first to answer.


Many visitors on many SE sites have no clue. That is why they are there! They want answers to their ignorance. For example, many members on Writing.SE have never successfully published a novel in their lives. They are either attempting to write a novel, or they aren't even writing novels at all. And yet they all answer questions about novel writing! And vote on answers to those question.

Given their ignorance, how should they know which answers are correct? Well, the truth is, they don't. Because they don't know (yet) what works and what doesn't in writing novels. So they do what people always do when they have no clue: they look for clues.

One such clue is "expertise". Psychological experiments have consistently and repeatedly shown that people who have no clue about something tend to believe the arguments put forth by those who appear to be experts. The actor in the white doctor's coat is very successful in selling snake oil to the gullible house wife.

On SE sites, "expertise" is signaled by "reputation". We all, whether we are conscious of it or not, are impressed by users with high rep. And if we have no clue about something, what a high rep user puts forth appears more likely to us than what a low rep user posts. So we vote for the answer given by the high rep user.

For that reason:

Have a lot of rep.


People love it when they are proven to be right. Therefore, they vote for the answer that coincides with their own opinion, no matter how unfounded that opinion is.

That is how Trump's campaign worked, and that's how SE works, too, at least on "soft" sites like Writing.SE, where you cannot easily see if a solution runs or not.

For that reason:

Post what everyone believes.

19
  • 10
    Within nine minutes of this being posted, the user's account was deleted, or so it would seem to me. Curious.
    – F1Krazy
    Feb 20, 2018 at 20:10
  • 11
    While this is true to a large extent, I think the complete repudiation of the SE model is a big over-cynical. After all, I don’t think Stack Exchange has been as successful as it has been—as go-to as it is in many fields—because the model doesn’t work. I have seen many late, contrary answers go on to considerable success—it does happen. And if it doesn’t happen often, well, there usually is a reason a lot of people are thinking one way. A system in which outliers succeed often suggests a system that isn’t working very well.
    – KRyan
    Feb 21, 2018 at 19:51
  • 15
    I downvoted this answer because we don't want cynical answers designed to do nothing more than get upvotes. We want the best answers possible from people who actually know. I hope and believe that the asker of this question was looking for how to write quality answers, not how to write answers that get votes. Feb 21, 2018 at 20:11
  • 7
    @ToddWilcox Frankly, SE is an attempt to intentionally gamify answering questions. People attempting to game the system are literally doing what is expected of them. Gamification only works if the ways to succeed at the game involve performing the desired behavior to the desire level of quality—that is, that maximizing points means maximizing the quality you’re offering. If that isn’t the case, and people can game the system in abusive ways, that user may be a mild problem but the systemic problem is far more severe. Supposedly, the system avoids that. If you think it doesn’t—meta.
    – KRyan
    Feb 21, 2018 at 20:22
  • 5
10

Answer the question in a way that will provide the reader (askee or searcher) with enough information to have actionable next steps. Do it in the smallest number of valuable words that you can. If you must expound, at least organize your thoughts.

Aside, the number next to your answer is a popularity contest for people who have registered. The real metric is unobservable: did this help someone? You may never know.

Allowances: sometimes the real question is not asked; an actionable next step may be to follow up with further research; complex topics may require concise complex answers; people tend to like good answers or those that affirm their views.

Addendum: Read and understand the question, first. We all have preconceived notions we want to talk about. A disproportionate number of answers are simply attempts at proving how smart one is or proselytizing one's personal philosophy. A text which which does not answer the asked question (or the mappable root of that question) is just masterbatory verbiage.

10

IF YOU MAKE AN EDIT THEN WRITE A NICE EDIT SUMMARY!

I just went to see the edit history of my own post and I was ashamed that I haven't done much to add a nice edit history. This is more important if your answers are long and require updating through out the years.

Bad Example:

enter image description here

Good Example:

enter image description here


In addition to all great answers. Put yourself into the OP's place and instead of just writing series of statements, write sub-questions you think comes to the mind of the average reader of that question. Understand the root of what they don't understand and then create related sub-questions and answer them.

Instead of saying "A class is different from a method", try writing a question with "What is the difference between classes and methods?" This verbiage grabs the attention of the reader.

I have somewhat done that in my answer here: https://stackoverflow.com/questions/42275060/what-is-difference-between-remote-notification-and-silent-notification-in-ios/42302369#42302369

In that one answer itself I've literally wrote and then answered the following sub-questions:

  • Do I need to request access once for local Notifications and once for remote notifications?
  • Do I need to enable something to receive Silent notifications?
  • Why am I saying all this?
  • Do I need to enable something to receive remote notifications?
  • Curious...Can you tell me what should my payload look like?
  • OK Got it. What is content-available?
  • I know you told me that I can only download something into my app when I'm using silent notifications, but is there way that I can also wake my app up in the background AND download something for remote notifications?
  • I'm confused about remote notifications...I thought whenever I get a notification...my app becomes active in the background and downloads something...can you explain?

Additionally I've used the word CAVEAT to cover some edge cases.

My point is really try to get in the readers head and think of what meta subjects could come to their mind.

(I do believe the question was a bit broad, but nonetheless the logic still applies)

Some more obvious suggestions:

Use <sub></sub> for adding subscripts. I sometimes use it to add the reference of a quote.

Don't add super big size images. You can change its size with a single click. See here.

For titles that you want to be made bold, just use = or -. It's just easier to edit vs. wrapping it inside **

My title with =

My title with -

My title with**

**My title with** (and a simple space messes up the formatting. This line should be bold but since there is a space before **, it's not getting rendered properly! **

Separate between subjects use horizontal line like below,


9

Bold and/or italicize the important points

Don't worry about organizing your answer in a logical format where the first sentence/paragraph has the main point, the supporting details are in the middle, and the concluding idea is at the end. Instead, write in a meandering fashion, and bold your most important points so that they are noticed, which is necessary because otherwise they would be in the middle of a paragraph where no one would see them. Formatting draws peoples' attention, and if you use a lot of formatting, your writing will get a lot of attention! This type of writing does mean that people will jump from main idea to main idea without focusing on the points in between, but that doesn't matter does it?

It's even better if you bold or emphasize the parts of your answer that aren't actually the important parts of the answer, as one answer to this very question demonstrates:

I don't expect this answer to garner many votes, but it may serve as food for thought, at least for one or two people, an outcome that is valuable to me as an educator.

(emphasis not added)

Break your flow with blockquotes

Let's face it, your writing probably isn't great. However, you can find some good writing on the topic, and through the liberal use of blockquotes get the reader to focus on the good writing instead. It's best if you find a source that is considered reputable, such as Wikipedia or another Stack Exchange answer. There's some more good advice in this article on the topic:

Quotations must be verifiably attributed, and the wording of the quoted text should be faithfully reproduced. This is referred to as the principle of minimal change. Where there is good reason to change the wording, enclose changes within square brackets (for example, [her father] replacing him, where the context identifying "him" is not included in the quotation: "Ocyrhoe told [her father] his fate").

The quotes can be long or short, but it isn't necessary to talk about the quotes after you quote them: just quote someone else making a point, and then move on.

Use ellipses to indicate omissions from quoted text. Legitimate omissions include extraneous, irrelevant, or parenthetical words, and unintelligible speech (umm, and hmm). Do not omit text where doing so would remove important context or alter the meaning of the text.

By no means should you summarize the conclusions of another source in your own words, which would take up less space and less time. Instead, use blockquotes liberally throughout your answer to provide a distraction from your own writing.

Refer to common knowledge, and don't worry about whether the common knowledge is actually common or correct.

Common knowledge is things that everyone knows, so there isn't any reason to verify it. As one user insightfully notes:

That is how Trump's campaign worked, and that's how SE works, too, at least on "soft" sites like Writing.SE, where you cannot easily see if a code runs or not.

The only thing you need to know to write answers is common knowledge and logic. With these two tools you can provide insight into any topic.

Things about Stack Exchange are more interesting than things about the actual topic of the site

Why talk about actual problems writers face when you can write answers about Stack Exchange? People love talking about theirselves, so on a Stack Exchange site, talking about Stack Exchange users is the golden ticket. If you can find a way to connect whatever topic you're writing about to Stack Exchange, you're set!

4
  • 1
    I really wish the time I spent on Stack Exchange made me a better writer. Feb 21, 2018 at 1:05
  • 6
    LOL, a well-played parody. I wonder which high-ranking user this sock-puppet account is disguising :D In all seriousness, however, I'd say participating in SE has definitely improved my own writing skills --more from the questions I've answered than from the ones I've asked. Feb 21, 2018 at 3:37
  • 1
    >Break your flow with blockquotes Ah yes, I hate that I can't quote without destroying my paragraphs. Feb 21, 2018 at 15:41
  • 1
    I do bold use a lot of bold in my answers, but the goal is to make it skimmable so people can see the key point of a paragraph. I don't expect most readers to read every word of a long answer, especially when they don't yet know whether their problem matches the question I'm answering. Sometimes it's just the first sentence of a paragraph on a slightly different topic, instead of a section header. But I definitely try to organize my answers into a logical order. e.g. Why is this C++ code faster than my hand-written assembly... has all that. Feb 22, 2018 at 22:08
8

Answer the question.

Provide examples (e.g. English SE) or code (e.g. StackOverflow), or diagrams / wireframes (e.g. UX StackExchange).

Use the formatting tools provided, especially code snippets (so the code can be run).

Provide a link to back up any assertions or claims that you make.

Explain the reasoning behind your answer. ("Show your workings", as all those Maths teachers used to tell us!)

One of my most upvoted answers is this one, which follows the above rules.

4
  • 5
    I doubt the reason you receive that many upvotes on that particular question is mainly due to the answer quality. In fact, the absolute number of upvotes on a (reasonable) answer has more relation to the popularity of the question. The relative number in relation to other answers is dependent on (writing) quality, but usually saying what people want to hear and being particularly sensible (even particularly insensible seems to work at times!) is a more powerful predictive factor. Feb 24, 2018 at 11:30
  • yes, that could also be the case. And I didn't actually cover everything I could have covered in that answer. But it more or less followed my suggested outline. Feb 25, 2018 at 11:55
  • Oh, sure, giving an example answer that adheres to the rules/tips you propose is a good idea, I'm just saying that the votes on it don't matter as much as you seem to imply. Feb 25, 2018 at 12:03
  • 1
    Ah, gotcha. Yes, on reflection, I have written better answers elsewhere that got less upvotes :) Feb 26, 2018 at 9:43
7

Some of the advice given in answers here is very sketchy. People keep saying that they try to make their answers as brief and as short as possible. However, the Stack Exchange guidelines in the help center are pretty clear about the place of brevity in a Stack Exchange answer:

Any answer that gets the asker going in the right direction is helpful, but do try to mention any limitations, assumptions or simplifications in your answer. Brevity is acceptable, but fuller explanations are better.

Writing purposefully short answers is the opposite of what people should be doing, particularly on subjective sites where, again, the official guidelines advise that answers to subjective questions should be "long, not short."

7

Get to the point first. On most SE websites, the main flow of traffic is people coming from Google/DDG for quick answers. If there is code which solves the problem, make sure it's correctly formatted (e.g. using code formatting for code) because many people skip straight to it.

Details are appreciated, but only after making your point and giving (or proposing) the solution. It's an explanation for your point, not a story leading up to your point.

While writing 200 answers on various SE sites, I've also noticed that writing summarizing answers can be very helpful: when there are some answers which each mention part of the answer, or which are all inferior in some situations, it is appreciated to write up a single good answer. Usually those questions are old and upvotes will only trickle in over time, but it does seem to be appreciated by future visitors.

The writing here is often very impersonal: no "hi user249" or "thank you" needed; it's encyclopedic rather than a chat. Upvotes are how you say "thank you", because an upvote also implies a recommendation for others to read the answer (or question). But it's also not formal to the point of a scientific paper: you can say "I think" and "I did" rather than "one might assume" or "we performed".

1
  • 1
    'People come here mainly from Google for quick answers (on most SE websites at least). ' While it purpose of most technology SE's is to provide 'quick and easy' answers to problems, on most science SE's, a more detailed and clear explanation could be better. In other words, brevity must sometimes be sacrificed for clarity. Nevertheless, you should first state what you're answer is explaining, of course. Feb 24, 2018 at 11:34
5

Oh, lets see:

  • I insert my personality as much as possible,
  • drop a slew of totally-made-up but ever-so-apt puns,
  • stick to what I've experienced rather than what I've read and
  • try not to get too hung up about whether people like what I've written.

Also: substance over style every time. Although, to be honest, I have insanely good style too.

Plus, I'm amazingly humble considering how humblingly amazing I am.

Oh, and I don't vary my format for different sites.

3
  • This does not provide an answer to the question. To critique or request clarification from an author, leave a comment below their post. - From Review May 18, 2018 at 11:25
  • 8
    @Galastel - I think the main issue is just that it's not clear whether this is meant to be satirical or not. If you just focus on the actual content, it's entirely possible this is meant as a legitimate answer, it's just written in a style that makes it sound like a parody. May 18, 2018 at 11:48
  • @Chris states that he is 'interested in hearing from other StackExchange members about how they write StackExchange answers' This is precisely how I write StackExchange answers. If this answer is deleted then I will suspect there to be something very hinky going on here.
    – robertcday
    May 18, 2018 at 12:11
5

"Ask a stupid question, get a stupid answer" does not apply on Stack Exchange.

On multiple occasions, I've seen people post "answers" that are either sarcasm/jokes, personal attacks, or just plain don't answer the question, and then try and justify it by saying something like, "Well, this is a stupid question and doesn't deserve a serious response."

If you feel a question is stupid because it's based on a false premise, you can write a "frame challenge" answer explaining, as objectively as possible, why the premise is wrong. If, for any other reason, you feel a question does not deserve to be answered, simply downvote/close-vote/flag as necessary, and move on. At most, leave a comment explaining, in the most neutral way you can, what you believe the problem with the question is.

Posting a stupid answer to a "stupid" question achieves nothing but fleeting catharsis. It will inevitably get deleted, either along with the question, or from the VLQ/NAA queues, or possibly even by a moderator if what you say violates the Code of Conduct. It's a waste of your time, and a waste of everyone else's time.

1
  • 3
    I'd add the fact that you can sometime write a great answer to a bad question... Oct 11, 2019 at 17:40
1

How do you write a Stack Exchange answer?

I answer your title.

Because my goal is to teach you how to fish. Once you know how to answer SE questions well enough, you'll never have to ask another one.

A good answerer knows how to distill an entire page of text into a single googleable sentence that culminates an empirical result * (y'all are killing me with these clickbait titles on the HNQ...). If it cannot be condensed, it's off topic, or it's in the upper 1% of actually hard questions that need an explanation why.

(*) understand the innate underlying question in the walloftext you're reading and find an answer in an accredited text in a Google search you can cite for it.


If I have a question in my head, usually after I'm 110% done writing the SE question, I have my own answer. It proceeds as thus: write catchy title; fill in the body. As I write; come to understand a different point of view that might make it easier to answer (edit: that part at the end of any random question where they always say, so my question is thus: ). Realize the body has a superior, condensed title that would better serve posterity (edit: this is when you're supposed to replace your catchy title, so that you realize it's a googleable question). Cut and paste title into Google. Profit (discard draft).

8
  • If you've followed all the rules to ask a SE question (to make it ontopic, and not POB like this question is), the only thing that should be left to do is explain why, or goggle the title.
    – Mazura
    Jan 6, 2019 at 22:46
  • 1
    This does not provide an answer to the question. To critique or request clarification from an author, leave a comment below their post. - From Review Jan 7, 2019 at 0:35
  • @SaraCosta - I'll admit the last paragraph is entirely unnecessary. But what's listed in bold is how I answer SE questions... I find the innate underlying question that an accredited google search can be cited for.
    – Mazura
    Jan 7, 2019 at 0:39
  • How to answer an SE question? Using the same process that keeps me from having to ask them (that it taught me).
    – Mazura
    Jan 7, 2019 at 0:44
  • I'm sorry, but I feel the opposite. If you were to expand the last paragraph and clarify the process, I think the answer would be much improved. It is actually the bold sentence I've got an issue with. 'Distill an entire page of text': what text are we even talking about? Jan 7, 2019 at 1:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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