The answer to this question (since deleted) gives 4 suggestions as a single answer: (click for full-sized version)

enter image description here

I think suggestion4 is great and suggestion1 is crap, but I can't express it by voting.
For the user, there is no way to quickly see which suggestion is the most upvoted. They will think it is suggestion1, because answerer happened to mention it first.

I believe each suggestion should be posted as a separate answer. This way they can be voted separately, and the best suggestion climbs to the top, which is what is the most useful for readers. If not, nobody knows whether the answer was upvoted for app1 or app2 or app3.


For the voting system to mean anything, we need to employ the system of using one recommendation per answer. Allowing multiple recommendations in a single answer introduces ambiguity into the voting since it's not obvious which recommendation is being voted for.

This use of the Stack Exchange engine is similar to the way the gathering of questions for the Q&A sessions occur during moderator elections, in addition to past attempts to use the Stack Exchange engine for recommendations.

Take this example:

A post contains 3 recommendations. One of these is actually really good but the other two are average. Because of the really good recommendation, this post becomes highly upvoted.

At some point in the future, a random passerby (our target audience) turns up and sees the highly voted recommendation but the first link (the good recommendation) no longer works. We've just recommended substandard software to somebody because of the way we chose to allow multiple recommendations in a single answer.

The proposal:

The way Stack Exchange works, in general, is multiple answers, with the best answer voted to the top. This is how we should use the Stack Exchange engine - one recommendation per answer with the best recommendation voted to the top.

Not only does this provide clarity to the people already using the site but it provides the most value to future visitors to the site.

  • 9
    We aren't voting on products, we're voting on answers. A good answer isn't solely defined as an answer that mentions a good product. So the voting argument is invalid. This is still a site for questions and answers, not a site for polls. History has shown that the Stack Exchange platform sucks at polls — for polls to work, ratings would have to take into account when an answer was posted, otherwise the first post wins. – Gilles Feb 11 '14 at 14:36
  • 8
    We are voting on solutions, an answer that provides multiple solutions suffers from the problems outlined in my post - which solution is being voted on? How do future users find this information out if you're talking about multiple different things in your post? – Flyk Feb 11 '14 at 14:38
  • 1
    Flyk, I just need a clarification because I am sure I understand your example. Links get broken equally in ORPA (One Recommendation Per Answer) and SRPA (Several), right? What makes ORPA more resilient to links that get broken? – Nicolas Raoul Feb 11 '14 at 14:38
  • 2
    Links isn't the key point - it was merely an example. If an answer contains multiple different <things> in it - what does the vote stand for if you up or down vote it? This is a site for recommendations, each answer should contain only one recommendation so that the votes can be used to determine which recommendation posted is the best of the bunch according to our community. Votes on answers are not useful to anybody if it isn't clear what is being voted for. – Flyk Feb 11 '14 at 14:39
  • 2
    @Flyk You aren't voting for products individually, but the solution as a whole. An answer that recommends several products should still be a single solution: “In circumstances A and B, use product X <blah blah>. In circumstances C, product Y is better because …”. – Gilles Feb 11 '14 at 14:40
  • 6
    "Circumstances" should never be a situation that arises - we've already outlined that questions be specific and outline a list of requirements so something that says "I might want to do this" or "I might want to do that" wouldn't meet the question requirements? – Flyk Feb 11 '14 at 14:43
  • "Circumstances" is inevitable! But I think we should embrace it, or at least take it in stride. Good answer can mention multiple recommendations – and they can be voted on as a good answer. If someone disagrees with some portion of an answer, they're free to create a new answer and highlight their disagreement! They're also free to create a new answer for each recommendation if they're so inclined. Truly, Voting Can Save Us. It's the non-voted-upon which is usually the dregs of SE sites. – Kenny Evitt Feb 12 '14 at 17:44
  • 3
    Do we really want people posting answers to voice their disagreement? – Flyk Feb 12 '14 at 19:51
  • Instead of disagreement, the answer should just be reposted with the parts agreed with? Although we perhaps descend into a situation where we get all permutations posted, and thus voting on each solution individually would have been the right answer? – Rikki Nov 30 '14 at 9:46
  • I disagree, for similar reasons to @Gilles. It would be like we would only accept one-solution answers in SO, answers containing alternative approaches should be split up. I don't think SO would be so successful without the ability to provide "bigger picture", comparing different approaches. – gaborous Mar 15 '16 at 21:44
  • "For the voting system to mean anything" - Whether voting means anything or not is a minor consideration IMO. If there's a narrative to an answer discussing two potential recommendations, the question of whether or not this can be easily voted on should not, again IMO, be reason enough to split someone's answer by force. – einpoklum Mar 21 '16 at 19:38

Nothing special.

One of the traps that this site can fall into is to become a poll site. Polls on Stack Exchange do not work. The voting reflects who posted first, not which solution is best.

A good answer says: “This method has upsides and downsides, that method has upsides and downsides. You should use method 1 under circumstances C and D, method 2 under circumstances E and F.” If you split answers, you lose the possibility of comparing them.

An answer with three suggestions is three times as good as an answer with a single suggestion. If you feel that an upvote isn't enough, you can award a bounty.

Regarding https://softwarerecs.stackexchange.com/a/15: you can express your opinion of the answer by voting. Is the answer as a whole good or bad? If you think suggestion 4 is bad, is that because you agree with the reserves in the answer (“The interface isn't friendly or as intuitive”, …)? Do your objections apply to the scenarios that the answer recommends that tool for?

Posting separate tools as separate answers would solve nothing and would only make the answer collection messier for no benefit at all.

  • 2
    This. People should not take one obviously fantastic answer and dilute it into three. Not only does this get too far into polls, ideal questions have at most one page of answers. – Tim Post Feb 5 '14 at 6:43
  • 9
    What if you totally agree with one of the suggestions, but think any of the other suggestions is rubbish? – Bernhard Feb 5 '14 at 7:19
  • 3
    @Bernhard then you add a comment, or upvote comment that already said it. Comment with most votes (after some threshold) is shown first. – Olli Feb 5 '14 at 8:54
  • 1
    Leave a comment, other than that nothing special, yes. – mirabilos Feb 5 '14 at 16:07
  • If one wants to compare answers, the individual answers can do so just fine. – Ira Baxter Feb 9 '14 at 13:53
  • @Bernhard, I know it's feels kinda wrong to do so, but it's actually fine to basically copy someone else's answer and edit it to match your own beliefs (or opinions). Like some of the suggestions but not others in an answer? Add your own answer, include the suggestions you like, and maybe explain why you don't like other suggestions. If you do this, it's nice to link to the original answer. – Kenny Evitt Feb 12 '14 at 17:19
  • @KennyEvitt You may copy someone else's answer, but you must provide attribution, including linking to the original. However, if your answer only differs in minor points, it may be deleted as a near-duplicate answer. – Gilles Feb 12 '14 at 18:02
  • 1
    @Gilles, I'm not sure what you mean by "must", but I agree that it should provide attribution and a link, regardless of the similarity of the actual text. ["Must ..." seems to indicate that it's not possible otherwise, which seems obviously not true, or that failing to do so is a moral failure. You, personally, and others on this site, are definitely scaring me off contributing. I'd recommend that everyone encourage others to improve their contributions instead of admonishing them from failing to meet some unknown quality threshold initially.] – Kenny Evitt Feb 12 '14 at 18:09
  • 1
    @KennyEvitt “Must” here is a legal requirement, due to the license that Stack Exchange content is published under. See the very last line of every SE page. This is a requirement if you reuse content that was originally contributed by someone else. There is no such requirement if you are not reusing content, for example if you express the same ideas in your own words. – Gilles Feb 12 '14 at 18:25
  • @KennyEvitt I am not trying to scare you off by contributing. But please listen when people tell you what's been tried before and failed — we don't need to repeat our past failures, we should learn from them instead. I definitely encourage people to improve their contributions, that's why I strive to provide constructive feedback and guidance. – Gilles Feb 12 '14 at 18:26

I'm agreeing with Flyk here, primarily in the light of how we seem to be asking questions at the moment. A good chunk of the better questions have lists of requirements to be met, and the answers tend to be in the form of which requirements have been met, and personal experience.

It seems to make more sense, to me, to have one quality answer per recommendation (and of course, if someone can post more answers, and keep them detailed, organised and specific). There is clearly room for abuse - maybe someone trying to be clever and posting 5-6 one liners. In this situation though, they would/should need to be distinct enough to split up rather than post as one answer

I'd also add that we're voting on the quality of a solution, and unlike many of the other sites, like SU or SF, a solution here is not a process using a piece of software but the software itself. As such, keeping them seperate lets good (or at least popular) solutions rise to the top. I'm not a firm believer in the FGTW theory, simply cause I've subverted it with quality answers far too often.

This is an example where I chose a single answer about two different software packages, since they were not terribly distinctive in terms of the features asked, while here I recommended two different ways to perform a task, distinctive enough that they needed their own answers

We have, and need to make it hard for questions to have listy answers. Its hard for someone to say "Oh, this tool is what I'd wholeheartedly recommend and this". I personally find that a quality, multi-answer question is likely to be rare.

On the other hand, there's always exceptions if its a class of software, with broadly similar characteristics - (this seems to be a nice edge case) having them as one answer makes sense. I wouldn't encourage multiple answers explicitly, but I'd say its a better way to handle knowing more than one great tool for a specific task.

  • 1
    “a solution here is not a process using a piece of software but the software itself” No! A solution here is a process using a piece of software. An answer that lists great software that is inappropriate for the task is not a good answer. This is a very good illustration. – Gilles Feb 22 '14 at 2:55
  • 1
    An answer that lists great software that is inappropriate for the task is not a good answer, but if you post it in the same answer as a piece of software that is suited for the task, the upvotes against the answer make the inappropriate software look like it's suitable for the task... hence - one recommendation per answer. – Flyk Feb 22 '14 at 2:57
  • I've clarified, with examples where I had a multi-recommendation and a multipost answer. – Journeyman Geek Feb 22 '14 at 2:59
  • I'm going to have to agree with Gilles on this one, the essential unit that we should be recommending is a solution that uses one or more pieces of software and the quality we are voting on is the aptness of the solution not a specific app. If you go the later route voting degrades to software popularity polls and answers Q's became "prepare a feature matrix of options for me". If our questions are user stories that have problems to solve (they should be) our answers must me thought of as solutions using various software as appropriate to meet the use case. – Caleb Feb 22 '14 at 7:51
  • Here is an example in which I recommend a file format as the primary unit of the answer, then add various clients that use that format as suggestions, but the solution is still intact and could be the accepted answer if the OP switched out one or more of the clients. Same with my answer to the speedometer question. – Caleb Feb 22 '14 at 7:54

I think it depends – if the answer is really just a list of different suggestions, than it's fine to add a comment suggesting that the suggestions (ha) be broken into separate answers – or just do it yourself.

But if, as described by Gilles in their answer to this meta question, the answer includes multiple suggestions and describes under what circumstances the individual suggestions would be more or less appropriate, i.e. if the answer actually includes some context about the specific suggestions, leave it alone and enjoy a great answer!

I think it's perfectly fine if this site is basically or almost entirely a "poll site" because 'popularity' is often a pretty good proxy for good recommendations for software.

  • It's been showed repeatedly on Stack Exchange that polls don't work in this format. You don't end up with votes showing what is most popular (which would already not be very good, since we're primarily going for most suitable, and popularity is at most a side concern). You end up with votes showing who answered first, which is completely useless. That's why answers should be solutions (which may involve alternatives), not items. – Gilles Feb 5 '14 at 17:42
  • 5
    I think the evidence is much more mixed than you claim. Sure, there's positive feedback because early up-voted answers for questions with lots of answers (i.e. more answers than fit in a few 'screens' via scrolling) are more likely to be seen, but diligent users of SE sites are aware of this and compensate. I don't think it's wrong for a question to have several good (or great) answers. I don't think the ideal should be One Answer To Rule Them All for every question, so splitting answers can, in some circumstances, help users provide feedback with more granularity. – Kenny Evitt Feb 5 '14 at 18:06
  • As evidence I can offer the list questions on Science Fiction & Fantasy and Stack Overflow (mostly deleted or historically locked by now; scifi.stackexchange.com/q/1 was one of the best of its kind of SFF). No, users don't compensate. – Gilles Feb 5 '14 at 18:21
  • 4
    I'm definitely in the (small?) set of SE-site-users that don't like what I (we?) perceive as over-zealous moderation. I mean sure, there are lots of deleted and locked questions that blatantly or brazenly violate the very clear conventions carefully laid out in the FAQs, but I know I read thru hundreds of answers on questions like that in the early days of various sites. In the scifi example question to which you linked, all of the answers fit on one page! How do you even know whether votes reflect popularity or 'suitability' or any criteria at all? – Kenny Evitt Feb 5 '14 at 18:28

I find it totally fine if an answer contains more than 1 recommendation.

What if you used 2 different software solutions in the past, both with their pros and cons and both fulfill the needs defined in the question?

A fine answer mentions the strength of both solutions and leaves the rest to the OP which one fits better.

And it would be way easier to compare 2 recommendations in a single answer.

  • 2
    -1, because the votecount will be biased – Bernhard Feb 5 '14 at 7:17
  • 1
    @Bernhard No, that is wrong. The voting on one-item-per-answer is completely meaningless. It only says who posted first. – Gilles Feb 5 '14 at 10:16
  • @Gilles That is not a problem specific to this SE-site. – Bernhard Feb 5 '14 at 10:25
  • 1
    @Bernhard Indeed, which is why this topic has come up before and one-item-per-answer is generally discouraged. – Gilles Feb 5 '14 at 10:35
  • 1
    @Gilles, voting on one-item-per-answer questions isn't "completely meaningless" and it doesn't only indicate the order in which answers were posted. I have in fact an 'existence proof', in the form of myself, because I often read all of the answers to questions and upvote all of the ones I think are good. Is there not some way we could encourage that among more users? – Kenny Evitt Feb 12 '14 at 17:22
  • 2
    @KennyEvitt I have several years of experience, and I assure you that the order in which answers are posted is the dominant factor. So you're one of the 10% who read all of the answers? Good for you — but we need a model that accounts for the other 90%. In addition, what if a newer answer comes after you've read the thread? Poll-like questions don't lead to upvotes for the answer being good, but for the item being good, and that removes a large part of the correlation between score and answer quality. – Gilles Feb 12 '14 at 18:03
  • @Gilles, I know we disagree about this, but I think that 'list questions' are an acceptable, and fairly graceful, 'failure mode' for this site particularly among the SE family. "The item being good" is, or I thought it was, the primary point of this site, i.e. (good) software recommendations, so it doesn't seem bad if that's the lowest-acceptable-level of answer. – Kenny Evitt Feb 12 '14 at 18:12
  • 2
    @KennyEvitt List questions are not acceptable here; it's because recommendation questions on other sites degenerated into list questions that they came to forbid such questions. Avoiding the traps of list questions is one of our major challenges. – Gilles Feb 12 '14 at 18:28

Proposed policy:

  1. Leave the first suggestion in place, and split the following suggestion into several additional answers.
  2. Comment on the answer: "Welcome to Software Recommendations and thank you for your answer! We prefer answers that contain a single suggestion, so next time please you have several suggestions, please post one separate answer for each suggestion. Thank you!".
  • But we do not prefer answers that contain a single suggestion. On the contrary, multiple suggestions should be posted in the same answer, with some indication of when to choose which. – Gilles Feb 5 '14 at 10:15
  • 2
    @Gilles: [citation needed] – Nicolas Raoul Feb 5 '14 at 10:28
  • 9
    These metas have no power here. SR already breaks many rules of other sites. Just because on other sites you have heard One idea per answer is BAD because it looks like a "POLL"! does not mean that the same is true here. – Nicolas Raoul Feb 5 '14 at 10:47
  • I have not just heard it, I've observed it repeatedly. The rules of other sites were established for a reason. I was, in particular, involved in Science Fiction & Fantasy since the private beta and for two years as a moderator. “List questions” were a big deal there, and their main problem was that they always got one item per answer. One item per answer systematically turned out not to work, because there was a plethora of answers and the voting told you essentially who came first, and to a lesser extent what is more popular, not when a suggestion was suitable. – Gilles Feb 5 '14 at 11:07
  • 1
    Let me guess, you also observed that shopping questions don't work? – Nicolas Raoul Feb 5 '14 at 11:59
  • Please define “shopping question”, it's used with many different meanings. In the very broad meaning used by some SU mods, there are questions that would work (basically, they call any question that would be on-topic here a shopping question) and questions that wouldn't (like “which motherboard should I buy, needs to be top-of-the-line, cheap and on sale”). – Gilles Feb 5 '14 at 12:03

You must log in to answer this question.

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