15

I understand that Stack Exchange is not built to support open-ended questions and discussions. And having read some of the arguments I can understand why it's problematic. I understand that at one point a failed attempt was made to support them. No problem. The infrastructure here does not support this kind of discussion.

I agree that there is a fundamental problem with the world having an open-ended discussion. I thought this summed it up quite well:

In such an environment, your subjective conversation will probably go like this: first person asks a (subjective) question, second answers it (so far so good huh?), twenty more add answers from different perspectives, fifty more attempt all imaginable kinds of jokes, hundred more add all imaginable kinds of side notes... and so on and so on, over and over again, until your brain explodes.

Human Interaction Management (HIM) teaches that systems should be built to account for the way people actually work. It eschews the idea of having the system attempt to force users to abide by the design. Rather it suggests that as users evolve their use cases the system should be adapted.

Consider that spreadsheets were originally build for what-if number crunching and that people actually used them to create schedules, todo lists, information matrices and so on. So the developers rather than complaining that those uses weren't intended instead chose to accommodate the emerging use cases and as a result spreadsheets apps became more useful tools.

A lot questions asked here end up in hands getting slapped. It's inevitable. I occasionally get my own hand slapped (by way of down votes) even after doing my best to form a proper question. It's annoying. Sometimes I'm left to wonder what exactly was wrong with it.

It feels that there is a bit of a group moderation mentality. Once the first person down votes a question the next person is slightly more likely to concur. That builds momentum and one wonders if it's just a matter of the wrong person getting to it first since there are often follow up replies that seem to indicate some interest in it. When this happens I'm disgruntled for having wasted my time — and sometimes I had invested a lot of thought. In the end to avoid the negativity and potential damage to reputation, I just delete the question.

I understand that Stack Exchange is designed for objective questions. I then considered whether there might be some other high-quality site that permitted a more subjective discussion but after some thought I realized that if the infrastructure could be built into the Stack Exchange network that would be ideal. I notice very often that questions are migrated to more appropriate areas. I also notice that people are continuing to ask broken questions. So in the spirit of HIM I figure accommodating the reality of what's happening makes sense. The solution involves building a different infrastructure for managing subjective exchanges.

The question Why functional languages? somehow skated by the subjective police. There are lots of subjective reasons that people like/prefer functional languages. This question resulted in a firehose of replies. It looked somewhat like the quote I mentioned: different perspectives and side notes were offered and the end result is the size of a chapter in a book. I'm not saying there isn't a lot of good information there, but it's anything but organized or answered as one cannot definitively answer a subjective question.

So when I talk about a different infrastructure for managing these kinds of discussions a few things come to mind.

  1. Each answer should simply aim to offer a single idea (one bullet point).
  2. Side notes and caveats that expound upon any one idea should be attached to it somehow and eventually someone of proper reputation could better integrate those caveats into the individual idea (and the attachment could be deleted).
  3. Sometimes one person has a better knack for explaining an idea than another person. It's thus possible that some bullet points will be duplicates of others. Duplicates could be grouped together so that eventually by way of votes the better ones are left and the less popular ones are collapsed and hidden. (A user could expand an idea if he wanted to see how others articulated it and maybe upvote another version, which might eventually become the main version.)

In any case, the idea is that the infrastructure is similar to a wiki in that by way of votes, aggregation and editing, the result is tight and well organized with the most popular ideas floating to the top. In this way, subjective questions would have no correct answer, but popular and well-organized thoughts on the topic.

I haven't worked out the details. All I'm attempting to say is that with the right infrastructure, subjective questions do have a useful place in the exchange of ideas. In a way, such a site would be similar to Wikipedia; but it's content would be organized by popular vote.

I know this is no little task, but is there value in doing so? I think there is. Nothing is going to prevent people from asking ill-formed questions. Wouldn't it be nice to somehow accommodate these? In this way both objective and subjective questions would have their own place. And people that like Stack Exchange the way it is would actually benefit. Lots of flagged-and-closed questions remain on this site. Such questions could be migrated to the sister site so that only the purely objective ones remained here.

Ultimately what you have is a one-stop shop for asking questions. You don't have stop and wonder if your question is well formed or worry that your hand will be slapped.

Google has become so synonymous with "search" that it's name is used as a verb. It does so by continuing to expand its definition of search. As such there's going to come a time when regardless of what you're searching for you can find it by googling it.

In the same vein, why shouldn't Stack Exchange become synonymous with Q&A. Why stop to wonder if the Q is broken?

  • 3
    By all means, try it. Look up mind mapping software. Give shirky.com/writings/group_enemy.html a very close read. – user213963 Jan 21 '14 at 4:38
  • 1
    Regarding the feature-request tag, I don't really think this is cooked enough to submit - You've got a reasonably well thought out discussion , which is the precursor to a feature request. I'm going to leave the discussion tag on this for now. – Tim Post Jan 21 '14 at 5:03
  • 2
  • 5
    «Ultimately what you have is a one-stop shop for asking questions.» Speaking personally, I don't want a one-stop shop. I want specialization. I want a butcher, a baker, and a greengrocer, all of whom are the best I can find for their respective services. – jscs Jan 21 '14 at 6:48
  • 1
    @JoshCaswell You'd also want to be able to ask your butcher what kind of round makes the best burger ;) (Devil's advocate, and all). – Tim Post Jan 21 '14 at 8:50
  • Related: Good Subjective, Bad Subjective. – jmac Jan 21 '14 at 10:26
  • I do indeed, @TimPost, but at the risk of taking the analogy too seriously, let me argue that this is not at all an open-ended question. It's got a concrete, expertise-generated answer. At worst, it's "Use a if you want X, b if you want Y. Don't use c because Z." which would fit in pretty well on SE. Asking your butcher what you should serve for your son's graduation party, on the other hand... – jscs Jan 21 '14 at 19:43
  • 1
    Sorry, vote to close was too quick, this question actually asks for an alternative to the SE format for subjective stuff while the other one asks to actually use it – Tobias Kienzler Jan 22 '14 at 10:47
  • @JoshCaswell Keep in mind that this was migrated from Programmers (Meta), where it's commonplace to close the equivalent of Tim's example as "primarily opinion-based" because anyone can chime in - even if a good, solid answer does exist. – Izkata Jan 24 '14 at 2:00
14

I suspect that most of us wish this was possible in an implementation that would not detract from the core purpose of our sites. We're all about building and maintaining an archive of facts, as we know them, in the form of questions and answers. To this end, the small amount of discussion that we permit is generally scoped to the content of a single post - be it a question or answer.

But that's just it, discussion on our sites is a means to an end, and it's that end that we care the most about. You want to know something, you type your question into a search engine, you see one of our sites in the results and you know that you'll see the best answer as voted by the community right on top - no need to go through 14 pages of stuff that simply doesn't help you in order to arrive at something that does. That's a big part of our brand and what people have come to expect.

I sometimes wonder if we're stifling some of our academic and research level sites by not providing some means of preserved and protracted discussion that is well suited for collaborators in different parts of the world. Chat is great for real time interaction, but breaks when users are on opposite sides of the world. Additionally, reading chat transcripts tends to be extremely painful, at least for me. But I digress.

What breaks here is varying degrees of open when you talk about open-ended. We went out on a limb and tried to define good subjective vs. bad subjective with a reasonable amount of success. We do have sites that feel more discussion oriented than others because the basis for most answers on these sites is personal experience. Parenting, politics and pets to name a few, and waiting in the wings is relationships and dating. Other than demonstrating that interesting sites often begin with the letter P, it's pretty clear that we are interested in seeing how far we can stretch the engine before it begins getting in your way instead of helping you.

While the engine is definitely not designed to handle subjective topics, it does a remarkably good job at doing so. It's easy to see why, the system encourages people to participate only when they feel that they have something valuable and meaningful to contribute. Our worst open ended questions are better than most long winded threads that you'd find on typical forums or newsgroups.

Where this always breaks is where people used the wrong tool for the job - you didn't want to write a question, you wanted to start a poll. Perhaps you didn't really have a question, but a series of assertions you wanted to open for comment. That's where we start to break down badly and things become a sea of noise or just mindless fun or snark.

We've brain stormed about ways to make something work as another 'place' aside from the core Q&A. Ideas ranged all over the place, nothing concrete has come of it, but we did find out that we're all really good at arguing. This comes back to most of us wish we could find a way to make it work, therefore there must be some way to do it. There may be.

There's also this proposal which we're going to allow to progress (with some ground rules that I'll be posting this week) that will probably test our format to the limits, we're going to learn quite a bit if it does well.

We want our focus to remain on Q&A. We want the quality to be better than you can find anywhere else. We never want someone needing to go through six pages of useless chatter in order to find the gem that they wanted, and we have to stay true to that. Still, we are working to identify ways in which some degree of productive discussion can take place, but we're very much at the drawing board.

I don't think this is something that can just be designed, this is something that's going to need to be discovered a little at a time. But, don't think that we're not open to it - it's just ... difficult to pull off correctly in a way that doesn't detract from the one thing we decided to do very well.

  • Thanks for the insight. I appreciate all the effort behind SE; it's amazing and incredibly useful. I only bother to bug a bit, because the problem of sifting through noise in forums is one I experienced and thought about since long before SE. I spent some time thinking about how to aggregate discussions into useful content so that every user visiting a given page need not also sift. I kinda look at this as one of those problems still waiting to be solved. I agree the solution would need an iterative process but I think a good deal of design could be done up front. – Mario Jan 21 '14 at 9:12
  • @Mario It is a problem waiting to be solved. We've got great Q&A software, and some really good forum software has since emerged. Where it gets hard is how and where to marry the two, and to what extent. Sometimes I wish questions or answers on SO had associated discussion threads, similar to how proposals have discussion threads on Area 51. Then I realize just what that would entail and .. well, not wish so much for it. I truly believe there is a sweet spot to be found, and that we have yet to find it. I just don't know what it is. – Tim Post Jan 21 '14 at 9:24
  • I think the best course is to reframe the problem(s). I dunno that this is spot for brainstorming, but if you guys attempt to start a dialog about this in a better place, I'd be happy to participate. As I mentioned, this is a problem that I have thought about for some time. – Mario Jan 21 '14 at 9:47
  • Breaking off comment discussions into individual chat rooms would be a good model, were it not for the clumsiness of that interaction (you have to wait for the "would you like to move this discussion to chat" link to appear, it doesn't "move" the existing comments but only copies them, and they have poor visibility). Normal chat rooms only have a problem with scaling because they have multiple real-time conversations in them, but the break-off chat rooms have a single subject, and most folks don't expect them to work in real-time. – user102937 Jan 22 '14 at 1:56
  • To address the problem of popular replies getting too much attention; what about supporting two columns? The left column would sort by popularity and the right column in reverse chronological order. This would allow newer answers to get some consideration and potential upvotes. – Mario Jan 22 '14 at 2:18
  • @RobertHarvey I think you nailed it perfectly with the term 'clumsiness' - almost every idea I've ever come up with to help facilitate more discussion but sort of off to the side has been just that once I finally mocked them up - awkward, difficult and clumsy. I feel like it could be done, I'm just at a loss to wrap any kind of UI around the idea. – Tim Post Jan 22 '14 at 7:25
  • 1
    Learning a new piece of technology can be difficult. In the past, it was common to have to purchase a set of books authored by some at the forefront of the technology in order to be able to produce anything. One great aspect of SE is that it allows this process to be streamlined. I love your point that there is "no need to go through 14 pages of stuff that simply doesn't help you" because at times that is an understatement. I have gone through hundreds of pages where perhaps 10% of it was actually relevant to me since the technology was so new and the topic so broad at that point. – Travis J Jan 22 '14 at 19:57
11

Quoting relevant pieces from the OP's question, emphasis mine:

All I'm attempting to say is that with the right infrastructure, subjective questions do have a useful place in the exchange of ideas. In a way, such a site would be similar to Wikipedia; but it's content would be organized by popular vote.

I know this is no little task, but is there value in doing so? I think there is. Nothing is going to prevent people from asking ill-formed questions. Wouldn't it be nice to somehow accommodate these in an intelligent manner? In this way both objective and subjective questions would have their own place. And people that like Stack Exchange the way it is would actually benefit. Lots of flagged-and-closed questions remain on this site. Such questions could be migrated to the sister site so that only the purely objective ones remained here.


Responding to individual points:

[Stack Exchange subjective] content would be organized by popular vote.

I think Grace Note did a good job explaining the issues with handling subjective questions on Stack Exchange in this Arqade Meta post; see the discussion about Repositories.

The problem with voting for answers to subjective questions is that the votes indicate popularity instead of suitability... that's why these discussions are subjective. For instance, suppose I ask this question: "What graphical design program should I use?"...

Do I have to mention all the possible answers and religious battles about Gimp vs Adobe Illustrator vs Inkscape vs Visio... Of course, the problem is a vague question, and voting for answers to vague questions tells me very little about solving my problem... What do the votes really mean other than "Other people like this software for what they think the (extremely vague) question meant".


I understand that Stack Exchange is not built to support open-ended questions and discussions...

Actually, that's only partially true. Stack Exchange's main site doesn't support discussions, but we do have chat.

I know this is no little task, but is there value in doing so? I think there is. Nothing is going to prevent people from asking ill-formed questions. Wouldn't it be nice to somehow accommodate these in an intelligent manner?

We have a pretty successful platform for well-defined questions; Stack Exchange has built a monumental reputation in the span of a few years (currently ranked #3 in Reference by Alexa) because we do focused Q&A so well. I'm not sure it's worth diluting our success to target something (subjective discussion) that's covered well by reddit, and yahoo answers.

We are known for Q&A... if someone isn't sure what their question is, or it's too subjective for the main site, they can still ask a lot of subject matter experts in Chat assuming the have 20 points on the site.

It's not a perfect solution, and it doesn't have answer votes; however, I don't really think voting helps subjective questions. Subjective questions need discussion... chat is the best we have.

  • 1
    There are in fact some sites whose entire topic strongly inclines towards subjective questions. The Workplace is a good example, and they have an ongoing struggle with answer quality because of it. (And whenever I pop over there, it drives me nuts.) – jscs Jan 21 '14 at 6:53
  • Please. By no means to I wish to down vote the SE network. This is the best thing since sliced bread in the Q&A world. I'm simply exploring the idea of evolving it. You made some great points I hadn't considered. I apologize if we users keep beating a dead horse. – Mario Jan 21 '14 at 9:03
  • 2
    @Mario It's not really a dead horse, it's something I think many people wish could work, if we could just find a way to keep more of what we like about the idea, without the cost of what really goes wrong with it on our platform. – Tim Post Jan 21 '14 at 9:06
  • @TimPost I think the essential ingredient involves some kind of merging/content aggregation. Consider how often duplicate questions are asked. I'll give more thought into how to best articulate how this might be done. – Mario Jan 21 '14 at 9:43
  • 3
    @Josh, we also have some tremendous answers from very knowledgeable people. Over 70k people have viewed and been helped by our most popular question which, while subjective, has a very clear and well-referenced answer that solves the problem people are finding the question trying to solve. Read: don't sell us short because some answers are poor -- help us downvote! – jmac Jan 21 '14 at 10:29
  • 1
    Sorry, @jmac, I shouldn't have included the parenthetical remark. The Workplace was just the nearest example to mind because I had been reading those "Hot" questions you linked. I understand that Skeptics had problems along similar lines when it started, but is now doing very well. I hope that the Workplace manages the same. – jscs Jan 21 '14 at 19:43
  • 1
    @Josh -- no harm no foul! Mind pointing me to any related discussions on Skeptics? I tend to agree that we have a lot of work to do on community moderation on TWP as we grow to make sure quality stays high, but I would hope that we prune the worst and visiting shouldn't make you cringe most days (we have been bombarded due to the hot network questions list the past week or so which could explain why you have a negative impression as of late). – jmac Jan 21 '14 at 23:10
2

Take three questions, from the domain of simple arithmetic:

  • what is 2 + 2?
  • what prime numbers exist?
  • who is the world's greatest living mathematician?

Only the first can be handled by current Stack Overflow engine and policies. The third is clearly out of scope and unwanted; the issue is the second.

Counting things as purely _objective _ or subjective misses this key nuance. Some questions have no right or wrong answers. Some questions have one correct answer. But many have multiple correct answers, while still having such a thing as a wrong answer.

What graphic drawing tools exist that are suitable for my task is no more or less a subjective question than how do I search for a substring in Javascript. The fact that it has wrong answers means a system capable of suppressing them should be able to do better than a simple messageboard.

  • I actually find the third question to be the interesting one. Yes, it's subjective, but if I were a math enthusiast I would be interested in the supporting ideas. I'm not thinking that the question would actually receive a definitive answer (e.g. no answer would be checked as correct), just that by popular vote people's thoughts on the subject would be prioritized. The trick is to offer a system that helps separate the signal from the noise by way of aggregation and curation. Everything you mention ("Counting things [...] a wrong answer") is what makes a question interesting. – Mario Jan 23 '14 at 14:28
0

It's generally understood that subjective questions lead more to discussions than to answers. I alluded in the OP that I wouldn't expect subjective questions to have one reply marked as the answer. I think that's the nature of subjective questions: they're more or less opinionated points and counterpoints.

Before Stack Overflow my fallback for Q&A was forums/newsgroups. When you ask a question it leads to more of a back-and-forth exchange. The end result is that when a person searching for answers finds that thread, he has to follow it from top to bottom in order to make sense of the exchange and disseminate the information, the good from the bad. I always felt this was a dreadful problem.

For a moment consider a thread a data structure that supports reads and writes. Initially, a thread is write heavy but in the long run the writes peter out and the primary use becomes reads. That being the case, threads should be optimized for that but they're not. Each person who discovers the thread has to traverse the thread to attempt to disseminate the information it contains. Essentially, the person reading the thread runs a mental map/reduce over the replies in order to formulate a summarized understanding of what was said. The worst thing about long-lived discussions is that all reads pose this same wasteful overhead. The longer the thread the more costly the overhead.

Consider another problem presented by open-ended discussions: the reader who arrives late. When someone arrives to a discussion that has existed for several days there may already be a mountain of discourse. The typical tack is to read the opening post and perhaps scan some of the early replies, but more likely the visitor will pay closer attention to the recent discourse. This results in information that is less useful because fewer readers are willing to incur the overhead -- e.g. TLDR.

The core problem here is the signal to noise ratio. Readers have to separate the noise from the signal and attempt to make conclusions. Multiply this by the potential number of readers who try to make sense of a discussion (map/reduce it) and it's easy to see how much effort is wasted. The solution I propose is one of curation and aggregation.

Imagine if the format for a subjective exchange periodically consolidated the discourse. That is, take the points and counterpoints and organize them; weed out the cruft. (Consider how adept writers rewrite and prune their prose.) Essentially, distill the discourse to its bare essence to reduce the overhead associated with reads. And in doing so attempt to keep each individual idea isolated just as answers on SO are isolated entities. By keeping thoughts isolated it allows them to be dealt with individually (up/down votes, etc.); this is imperative as it's impossible to properly deal with compound items.

Who would do all this? Primarily the user who initiated the discourse; however, others of sufficient privilege could participate if they wanted. If I cared enough to ask a question, I would be willing to do as much.

Essentially, the idea is that each question is its own open source project and the author is the project maintainer. Curating is handling pull requests. In the end, while effort would be put on the maintainers, the content would be better optimized for reads.

Let me elaborate a little on the infrastructure:

  1. Each thought is expressed in an isolated node.
  2. Others may express similar thoughts. Nodes that are duplicates of others would simply be pointed to the original node and marked as duplicate. Superior expressions of an idea by way of votes are displayed while duplicates are collapsed and hidden.
  3. Nodes could be supported/opposed by others nodes. Doing so would be a matter of pointing one node at another an marking that relationship as pro or con.

While this is perhaps not fully baked, it feels sufficient as a starting point for subjective discourse. It's not that complicated. It supports nodes (isolated replies) exactly as SO does. The part that varies is that nodes can be tagged as duplicate, pro and con and directed toward other nodes. In the same manner, when a thread itself is marked as a duplicate of another the two can be merged (SO supports merging I believe).

In wiki style, the entire contents of the exchange can be preserved; however, most of it will be curated out of the public eye as a more consolidated summary of the discourse will be immediately visible.

Over time as ideas develop and things change one might notice different ideas floating up while others drop down. What I'm saying is that this model would essentially support an axis of change over time. To better do so, up/down votes could support a rate of decay. That is, newer up/down votes would carry slightly more weight. There are of course other things that might go into weighting replies. Figuring this out won't be easy.

One last thought on improving a subjective infrastructure. It might be nice if replies had their own subject line (e.g. a summarizing title). In this way, it would make it possible to collapse a discussion to its essence and to explore (expand) only the interesting nodes.

The main idea I'm making is that the format should be focused on optimizing discussions for reads even if that demands a bit of work. Online content has a long shelf life. It's going to be read far more than its going to be updated. Because of the dynamic nature of open-ended discourse, I see no better alternative for supporting it than some method of curation and aggregation. Shift the map/reduce from the reader to the system infrastructure.

You must log in to answer this question.

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