Are we primarily helping the asker or building a repository (or sticking to the middle ground)?

Stack Overflow has a whole lot of questions asked (and answered) in such a way to be hopelessly too specific to the asker’s problem or just another copy of an endlessly duplicated question, meaning they’re not helpful to anyone else. There are also more than enough people perfectly happy to provide a direct answer to such questions instead of trying to provide a more general answer or close or edit the questions. This is perhaps my biggest problem with the site and it’s (from my point of view) causing a lot of conflict and frustration all around.

Of course helping askers and building a repository aren’t mutually exclusive. There are many cases where actions benefit both (which is great), but there are also plenty of cases where you need to choose one at the cost of the other. And these choices define whether you’re more welcoming or higher quality.

This is addressed here and summarised in this older post (and some of the other answers there).

Would there be any change in terms of this (e.g. attempting to minimise the number of such questions) on this site?

Improved tools was mentioned in the post above, but it might be more important to consider the incentives for answering and moderation (i.e. closing and review), and perhaps also how easy it is to publicly post a question in the first place (e.g. send a question to review before it goes public). There has many other suggestions on Meta SO over the years, but I can’t recall that many of them right now.

1 Like

Basically, in my view, shift the focus of the site from “get your question answered here” to “come add to the repository of knowledge, with the idea that an entry you add is intended for people in the future, not primarily yourself”.

6 Likes

The flip side is, maybe we could allow certain things not allowed in SE. Specifically “shopping questions”. I don’t mean the “Where can I buy ‘x’?” questions - for a lot of reasons those should generally be off-topic. I mean questions like the one I recently asked about a diagramming tool for PostgreSQL. It got rejected as a “shopping question”. But while “best tool for xyz” does change over time as new products appear and old ones disappear, and there is an “opinion” factor involved (but opinions come into play as to “best way to code xyz” too), I think that these types of “shopping” questions should be allowed, particularly if the end result is a “Community Wiki” style comprehensive answer.

1 Like

I think this is going to vary by site. Stack Overflow is reportedly swimming in homework questions and could benefit from more canonical questions. On Mi Yodeya there are many specific, non-duplicate, overlapping questions about areas of study, and we welcome those questions because the education factor itself, and not just “my code is broken”, is important to us. On Writing we get a lot of repeated questions that we’d like to drive toward a common target (canonical question would be great), but we also get a lot of the other sort. On Worldbuilding, we’re all over the map here.

One size does not fit all.

Each community needs to work out where it wants to be on this spectrum. We need to build a system that supports a range of goals.

13 Likes

There’s widespread agreement that a large number of questions on SO are bad, but there’s a lot of disagreement as to which questions are the bad ones. The dominant position on MSO is that good questions involve effort and need to show non-working code. Like Academic-Quantity, I find this frustrating because when I’m looking for an answer on SO, I could care less about someone else’s working code. The questions I find helpful questions are mostly “how do I do this thing?”, with zero or little code in the question, no need for effort in the question other than looking it up in the manual, and answers that help me because they show how to do this thing that I also wanted to do.

I don’t completely oppose debugging questions, because they are useful, but I wish that they were edited to highlight the specific issue. This is something he asker can’t do — if they knew where the problem was, they’d usually be able to find a solution. Unfortunately, SO has a widely enforced rule that you do not edit code in a question (even fixing obvious errors like a stray or missing character at the end of a copy-paste often gets rolled back!).

Going beyond SO, the whole reason I’m on Stack Exchange and Codidact is to build a repository of answers. I want to get answers for myself, and I pay it forward by answering other people’s questions. I’m not here to help specific people, I’m here to help myself and to help people in general. So for me, as long as the site maintains a welcoming atmosphere, the priority is on the community, not on individual members.

I’m aware that we aren’t all on the same page here and we need to compromise. But I do hope to push the compromise further towards a knowledge repository and less towards one-on-one help. A forum or discussion list without a concept of duplicates is perfectly fine for one-on-one help; it’s not what we’re here for.

Stack Exchange does not ban recommendation questions (“shopping questions”). Some individual Stack Exchange sites do, most prominently Stack Overflow and Super User. Others don’t, for example Unix & Linux has no particular problem with questions of the form “best tool for xyz”.

Software Recommendations was created because SO and SU don’t accept recommendation questions and weren’t willing to start. I was a moderator for the first year of the site. What I learned (or rather confirmed, none of it really came as a surprise) is that recommendation questions do work in the Q&A format, but they require strong moderation. SR enforces question and answer quality guidelines, not just through voting but also by closing questions and deleting answers that don’t meet the guidelines regardless of their score.

Indeed. That’s a big part of why I’m opposed to a Quora-like model where any topic goes. Different topics, and different communities, have different views on what makes a good question or a good answer.

12 Likes

I’ve been thinking about this, too. It’s fine starting overlapping with the more “fringe” SE sites, which StackOverflowTheCompany will probably be fine dropping. But when the time comes for us to overlap with SO, the majority of users are going to be those who are already fed up with SO’s restrictions. Just take a look at all the “SO is bad” articles - the vast majority are coming from the “too mean to newbies” perspective, which is really close to “too mean to people who haven’t learned how to ask questions yet”. Creating a more restrictive site isn’t going to attract users.

I’m not opposed to a Q&A site, but I’d love to see some kind of a forum-ish nature as well. So “newbie” questions enter a different flow: they cannot be answered/voted on for [some time period], during which time we have incentives for regulars to help the op improve the question. So a more of a back-and-forth forum-style question-improvement and duplicate-suggesting phase, after which the (updated) question may “graduate” into a real question.

The nice thing about forums is that topics fade. When a question is poorly asked on a forum, there’s some back-and-forth to improve the question; if the op is unresponsive, the question fades to the background over time. I’d love to capture that (IMO more user-friendly) aspect of forums with our site, while keeping it a Q&A site at its core.

2 Likes

A few other sites have implemented sandboxes using their metas, so people can get help framing a question before asking it for real. Some have asked to have a sandbox be a first-class thing that newcomers are directed to, rather than a hack on meta. And SO has triage, which (IIUC) is limited exposure during a new question’s first minutes or hours.

It seems prudent to plan for something like this being needed eventually, as a per-site configuration (and possibly per-tag on the site).

6 Likes

The only site where I’ve seen Sandbox get serious usage is Code Golf, though I freely admit that I have only sampled a small fraction of the SE sites and only heavily participated in a few. Code Golf is a very different type of site where every little detail of the question matters a lot and where the question isn’t “how do I fix this” but rather a challenge to the rest of the community.

That is simply not the case with most other sites - either the details hardly matter at all or they can be easily edited in by others making reasonable assumptions or they are basically “clean it up, make the code clear, DON’T SHOUT, etc.” In all sites I’ve seen except Code Golf, trying to get a new user, who is often anxious to get their first question answered quickly to use a Sandbox, wait for comments, revise, etc. before getting a chance to really ask their question, is not practical.

3 Likes

I’m largely “fed up” with the lack of restrictions and the lack of a focus on building a repository of questions and answers (that is: accepting questions that don’t seem in line with that goal and not trying to edit or answer them in a way to maximise future value). I can only speak for myself, but I think a lot of experienced users might have similar thoughts, or at least related thoughts (especially given posts on Meta).

I’d argue in most cases where people see SO as unwelcoming it’s due to a mismatch of expectation vs reality. If someone knowingly jumps into a jellyfish tank without taking required precautions, it wouldn’t be reasonable to expect to not get stung nor to be angry at the tank owner or the jellyfish if it happens (not to mention they probably won’t jump in in the first place if they know there’s a good chance of getting stung). If that person jumps into a jellyfish tank thinking they’re jumping into a swimming pool instead without being told what to expect, what to do or what not to do, it’s a different story.

Take a look at what’s shown to new users when asking a question or what’s written in the help center. It seems kind of easy-going about what is expected of users (by seeming more like general guidelines than rules), but this doesn’t match reality.

Something else to think about is whether we want to welcome as many people as possible (at the cost of quality) or whether we’d rather keep things smaller (at the cost of losing out on some good questions). Of course this is relatively speaking: “smaller” can still mean 10s or 100s of thousands of active users and millions of visitors.

I like the idea of a pre-post sandbox / review, but I don’t think it would scale well if you welcome as many people as possible. SO already has a problem where a non-negligible portion of questions get no response whatsoever. Making it more effort to help someone and cutting down those who can respond to only those interested in taking the time and with the patience to help users improve their question will leave a lot more questions without a response.

6 Likes

SO/SE is a large scale format and like quora and reddit. It thrives because it nearly excepts any question. Not that much of the questions get deleted

(although it is increasing and nowadays it is roughly 35% on StackOverflow and on other sites it may have gone up to 50% of the questions being deleted. But note that a large part of that is automatic deletion after one year without positive score and answer)

StackExchange is not a concise database of ‘frequently asked questions’ but it is instead a database of ‘every ever asked questions’.

  • This is on the one hand what makes it currently explode and it indeed has a downside. It has become difficult to manage.

    Now the dumpster is full and several people have started throwing their dirty questions next to the dumpster (and nobody cleaning it up). We have the copy-behaviour effect that more and more people are dumping next to the dumpster and the pile is growing.

  • But on the other hand, not very unimportant, it makes the website like a big strong magnet for search engines. All those little questions together will give you an answer to any possible question. If you reduce the size of questions then you get more something restricted like a manual or another redacted/reduced form of help/information. I believe that this is a strong power of SE. It is only not able anymore to manage the questions (there should be more use of AI and possibly less fear to remove more questions).

I believe that it would be good to make use of all the input of questions. This is what makes the Q&A format so unique and stronger* than an ordinary wiki-encyclopedia format. The information is based on pragmatic content/questions that people actually have, rather than questions that experts come up with**.

But, to improve if from SE it would need better ways to clean up or organise the rubbish (personally I would get rid of gamification and create an atmosphere that invites more editing of questions and answers. Improving, ordering and categorizing the questions should be considered as valuable, or even more valuable, as answering the questions).

For smaller communities it can be possible to have more strict rules about which questions are allowed or not. But then one might be more talking about an organized database that will grow towards a few 10k questions and not about a project like SO that has more than 20 million questions.


*We could see the Q&A, the way SE/SO does it, like training artificial intelligence. You need a lot of data and input to create a nice product. It is the filtering which eventually will make you not see this mess of data. However, you need a format that allows the input of a lot of questions, no matter what quality. Then the AI, in this case experts on SE/SO, will turn it into information.

**This makes me think of all the times that I hated answers on forums that do not solve the problem - in a pragmatic way - but give rather obnoxious indirect answers like use tool X which you can find under Y, while the questioner (me) has no idea where Y is and how to install X

4 Likes

If you have many questions that have little future value so answerers can only really address the asker’s immediate problem, askers and answerers alike get used to answers just addressing the asker’s immediate problem instead of being more general. This also means questions are phrased in a way to encourage such answers and potential editors just tolerate it.

This comes at the cost of future value in cases where we want a canonical answer to something, which applies to almost any question we want to have show up on a search engine or use as a duplicate target.

And it’s really repetitive and boring to deal with the endless stream of duplicates, debug-my-code questions and other “rubbish” as an answerer, editor or closer.

Yes, it might be harder or take longer to reach critical mass by being a bit more exclusive, but I’d argue we could be much better off for it, if it’s implemented well and it works.

2 Likes

Worldbuilding uses its sandbox heavily, and I forget where we got the idea but it wasn’t Code Golf. I think there are three or four sites out there doing it. That’s not a lot by sheer number, but I think it’s fair to say that some topic areas and/or sets of site rules call for it more than others.

4 Likes

Strongly disagree.

“Get your questions answered here” is why those with problems come. “Come add to the repository of knowledge” isn’t going to attract many people.

I might write a post or two occasionally about something new and cool I’ve done or researched, but that would be nowhere near the volume of writing answers to (good) questions. Then who would be around to read the former anyway? So why bother at all?

We need people asking questions to drive the activity. But - and this is the important part - we don’t want just any questions, we want GOOD questions.

Three good questions is better than five good questions intermixed with five bad ones. The goal must be quality, not quantity. In my view, this the one thing above all that SE got wrong, especially over time. Some bean counter looks at total visits and ads clicked and revenue generated. More traffic is therefore better, so don’t ever make anyone feel unwelcome. That might work in the short term, but in the long run the site quality declines and the system collapses.

I think that the repository of knowledge will take care of itself as long as the site is known to be a good place to get questions answered.

But, to keep the experts around to answer the questions, the place must be kept clean. Poorly researched, poorly written, off topic, lazy, or otherwise bad questions must be ruthlessly and expediently dealt with. No, we don’t want to help the poor lazy ingrate make his question better. That noises up the site, dissipates volunteer energy, and burns out the experts. The poor lazy ingrate needs to be thrown out, and preferably kicked in the butt on the way out.

One of the frustrations that caused me to leave EE.SE was that as the place grew and got more questions, we weren’t allowed to properly deal with the bad ones.

9 Likes

If someone knowingly jumps into a jellyfish tank without taking required precautions, it wouldn’t be reasonable to expect to not get stung nor to be angry at the tank owner or the jellyfish if it happens (not to mention they probably won’t jump in in the first place if they know there’s a good chance of getting stung). If that person jumps into a jellyfish tank thinking they’re jumping into a swimming pool instead without being told what to expect, what to do or what not to do, it’s a different story.

Yes, but these are people who are learning how to swim. The site promises to teach them how to swim, and then throws them into a jellyfish tank where they get stung, and then they’re blamed for it.

And these are the people who will be most likely to adopt an alternative platform. The majority of SO contributors will not be joining us - at least not right away. SO has many faults, but they do a decent job of keeping the best contributors (they’re good enough), and most of our adopters - probably for multiple years - will be those cast off of SO. I think our only real chance of survival is to recognize the user base we have been given, and structure our Q&A accordingly.

4 Likes

I don’t think this is a good approach or way to present the site and community we’re trying to build.

There is, clearly, a line that needs to be drawn on acceptable standards. But the words used here belie contempt, and contempt should not be foremost in what we are trying to achieve.

There are those who fire-and-forget questions that are beyond hope of repair. Moderate those, deal with those effectively. There are others that may be poor quality for other reasons: lack of familiarity with site’s topic, lack of familiarity with asking smart questions (no-one is born having read ESR’s essay on that), lack of familiarity with jargon, lack of familiarity with the language used on the site, lack of familiarity with the software (“I hit ‘post’ instead of ‘preview’!”). These are off the top of my head, doubtless there are others.

I get that cleaning up mess is a crappy job – that’s the clean version – and I get that it takes energy that feels better spent elsewhere. But today’s newbies are tomorrow’s experts. And people remember.

I think there is a way to help the asker, encourage people to ask well and maintain standards, without resorting to completely alienating those who fall short of the mark.

There are ways of ensuring quality without calling people lazy ingrates.


Separately, your point about eyeballs and revenue driving considerations at SE Inc may well be close to the mark!

On the other hand, consider that one of the things that got SE so twitchy was a perception of “it’s not worth asking a question on SO, it’ll be downvoted, closed and deleted in minutes”, which has persisted even after they’ve made a big push to try to change that culture.

If SO and SE are seen as the sites hostile to newcomers, we don’t want to start out by saying “we are even more hostile”.


What in particular were you not allowed to do to bad questions?

7 Likes

I do think it’s worth pushing very hard on something SE recently started trying to do: designing the UI and UX itself to be friendly and helpful in clearly setting expectations and explaining difficulties to new askers (and answerers). Since we aren’t yet known for anything, we can reasonably initiate a site culture and software experience that maintains quality standards with much less noise from frustrated expert moderators, shifting the load of explanations more to the software, and building a wider reputation from the beginning that matches or improves on what SE is currently trying to pivot to.

So: more focus on prompt feedback about specific quality issues triggered by expert voting, less focus on general downvotes and comment explanations. Quicker, more distinct warnings that experts are about to freeze answering until the question is fixed.

7 Likes

I think it would be useful to support both “My specific problem” questions and “The best way to do X” broader documentation style community wiki posts with lots of discussion and commenting space available for clarifications and variations depending on parameters, so that multiple situations could all be addressed on one page, unlike we’re used to dedicating a page to every different “specific problem” question.

It’s worth noting that I have not participated in Documentation project frequently enough to get a good sense of it before it got closed, so I can’t even comment on what it did right or wrong, I just feel like it was a good idea, but people may not have approached it properly because of too ingrained prior SO experience (wild guess).

Also, not only programming sites could use a sort of “Documentation” approach, but every other topic too. I just don’t know how it should look like.

Either way, this would be better thought of after the main work on MVP is done, I guess.

4 Likes

Helping more than the OP is the goal, but failing to solve the OP’s problem in the process of writing a canonical doesn’t help either.

Couple of the pitfalls I have seen for the canonical questions

  • A canonical question is asked in the title but the body is far to specific and people answer that instead leaving the people arriving from Google high and dry.
  • A general, canonical question is asked but people reply with far too narrow answer. An example would be asking what broad features one should look for in a product, only to have people answer with specific recommendations.
  • Saying that if the OP already knows the answer or isn’t experiencing that exact problem they shouldn’t ask the question, basically self answers and hypothetical questions are heavily discouraged.
  • Closing canonical questions as too broad which defeats the purpose.

The other thing is that really narrow questions such as identify this object still have a purpose in my estimation because they can be fun and drive community involvement even if the long term SEO value isn’t very high.

5 Likes

I think it would be very useful (or at least interesting to see it in the wild) for “My specific problem” question askers to be able to link to a canonical “guide” and explain how that doesn’t help you solve your specific problem, so that everyone has something to work with, and OP has proof of prior effort.

Also, I don’t think we should concern ourselves with search engine optimization of content at all. If anything, googles of the world should become better at giving us the answer to our questions. Maybe one day we’ll be discussing how to build a better voice-activated AI search assistant, I wouldn’t be surprised.

1 Like

I couldn’t agree more. Once you generalize an answer enough, you get a manual or a textbook. The success of SO was in large part the focus on practical answerable questions. I.e., a real problem users run into in the practice of software development. Do this well, and it also becomes a repository of knowledge.

1 Like