What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 130 Stack Exchange communities.

Stack Overflow's rules have grown very complex and often counter-intuitive over the years. That doesn't mean they don't exist for a reason, but grokking SO's culture can be really tough even for good-faith newcomers.

Many people learn much, much better from real-world examples than from a huge list of abstract rules.
I know - I'm one of them.

To better guide these people, we should show a curated and annotated list of good and bad example questions. This list should be extremely easy to reach - it could for example be a 2nd tab on the "How to ask" page. It should be a separate page, maybe with questions listed in 2 columns.

Each example is a full question, with an explanatory "what's wrong with this question?" paragraph underneath. The paragraph will be very curt and to the point.

What's wrong with this question?

It's a "shopping question" asking for a product recommendation. Questions like this are off topic on Stack Overflow: they frequently lead to low-quality content and subjective opinions.

etc. etc.

Thinking of the extreme problem cases is easy. Too vague, Gorilla vs. Shark / Shopping question, Duplicate / too basic, Multiple questions in one, Wall of code... But we should also cover the more subtle cases where a question looks perfectly intelligent, but is not a good fit for the site.

Thinking of good examples is less easy - I can think of

  • Great title
  • Question body giving a lot of detail, but not too much
  • Good structure (introduction / question)

^--- this obviously needs more thought.

The list could be curated by the community, or staff, whatever works.

Ideally, there would be example questions on a per-tag basis, at least for the big tags.

A "before" / "after" view might also work: How do I fix a problematic question?

The examples should be linkable so they can be used in communicating with authors of problematic questions. I can't see a way to misuse this in a rude way as there was with "What Stack Overflow is not".

Also, I think this would be a great help for dealing with the language barrier. An example is going to be way easier to grasp than a thousand words if you have trouble parsing the language in the first place.

share
    
I assume you're aiming only for the worst of the worst with your bad examples? Because if the language barrier is a problem, it might be difficult to grasp the subtle differences that sometimes exist between a good on-topic question and one that doesn't belong. –  Bart May 7 '13 at 7:45
    
@Bart That's a good point. It shouldn't be only the clearly bad cases - it's the subtleties that are so hard to grasp around here sometimes. We've all seen intelligent people get frustrated with SO because they don't get around those. A list of examples would absolutely have to contain examples that look great, but are a bad fit nevertheless for a less-than-obvious reason. Maybe even with a "before" / "after" view that shows how to fix it –  Pëkka May 7 '13 at 7:49
    
Great idea, but I'm very pessimistic about people taking the effort to read them. I think the problem is not the lack of documentation on how to write a good question or the availability of good examples (if you're willing to look around for a while). The problem is that many people just don't seem to be aware of the fact that their question could possibly fail any quality standards, so why take the effort to make it better? Maybe if your idea would turn into some good question contest (with extra rep, yay!) some people might wake up. –  Gert Arnold May 7 '13 at 9:09
4  
@Gert we'll never reach those who don't read anything - they'll eventually be shut out by the quality filter, and good riddance. But I'm convinced there is a demographic of reasonable, good-faith potential users who can grow into productive members once they "get it". Those are who this would address. (Also, running into a question ban may indeed cause some people to wake up, and develop an interest in how things work) –  Pëkka May 7 '13 at 9:18
2  
But there seem to be so many of them (the non-readers) * sigh *. Nevertheless I'd like to give it a try and I'd love to cooperate in it. –  Gert Arnold May 7 '13 at 9:24
    
For question body, I suggest: "1. Explanation of what is attempted (probably with code snippet). 2. Expected/desired result. 3. Actual result." –  S.L. Barth May 7 '13 at 10:11
    
+1, for catering to good-faith new users instead of ones who are unsaveable. We don't do that enough in our attempts to improve new user flow. –  djechlin May 7 '13 at 14:18
    
+1 I support this. I also shamelessly nominate one of my questions to be on the list as a good question in the javascript/jquery tags. stackoverflow.com/questions/15118057/… –  ryan May 7 '13 at 16:04
6  
Declining this because I think Laura's answer nails the real solution: strive to make every question you see a good example; there's a much better chance of someone new finding a long-tail question in their area of interest and learning from that than of them finding and reading a small, curated list. –  Shog9 May 8 '13 at 21:23
4  
@Shog9 No matter how hard we try, not every question is going to be a model question. The goal here is to make users learn good and bad questions, not good and bad code. –  Emrakul May 8 '13 at 21:48
    
@Knights: of course not; but those are the examples that most folks will actually see, and thus the ones that'll actually influence them. Here are your greatest hits - they don't need to be perfect, but are they good examples? If not, there's where you can start... –  Shog9 May 8 '13 at 21:52
1  
strive to make every question you see a good example ok, but I'll need one of these for that –  Pëkka May 8 '13 at 23:45
    
It's declined for the same reason "what stackovrflow is not" was deleted after many people contribute to the subject, it just reduces the income to SE. –  user209407 May 10 '13 at 0:51
    
This whole not-a-good-fit thing annoys the beejeezus out of me. Who are you lot to decide what questions people can ask? If you aren't interested in a class of question then perhaps you could ignore it and MYOFB. I find this site less and less useful as it fills up with self-righteous wowsers. –  Peter Wone Feb 11 at 5:04
    
@Peter Of course there is a lot of self-righteousness, snark, and overzealous closings, although I don't see how that has to do with the suggestion for an educational tool. Either way, I invite you to partake in, say, the PHP or Android tags for a while (where questions of inferior quality vastly outnumber the good ones) to see the other side. I 100% subscribe to the general belief that without the strict rules, this place would have become another Amazon Askville years ago. (That said, I wouldn't mind seeing things relaxed around recommendation questions... but that's a different discussion) –  Pëkka Feb 11 at 12:41

3 Answers 3

This is a massive amount of work to treat a symptom rather than a disease.

Rather than having the community pour energy into curating a massive list of questions that, to be quite frank, new users will be no more likely to read than they are likely to read the FAQ, there are two things that should happen to get to the heart of the problem.

I. Better, clearer guidance in the infrastructure.

Simply put, we (SE, Inc.) need to make our complex rules more understandable. Our current guidance to new users is spread across MSO, child metas, blog posts, the FAQ, and various other pages like How to Ask and How to Answer, which are difficult to find again even when you're trying.

To remedy this problem, we're working on a new, consolidated Help center - I'll be posting more about this on meta in the coming weeks, but the basic goal is to make a comprehensive AND comprehensible set of reference material that new users can browse and experienced users can easily link to.

The second major problem is that our guidance sometimes uses jargon that doesn't make sense to a drive-by user, or we use common words with a meaning that is different and very specific to our site. We'll be trying to avoid that in the new copy.

II. Demonstrate by example -- not listing examples.

This is a community effort, which I'm hoping will be well-received since you are suggesting a community effort as well. :)

I firmly believe that learning as you go is the best way to familiarize oneself with the Stack Exchange model. Documentation and guidelines are helpful for a better understanding of the nuances, philosophy, and background, but you really shouldn't need them to learn what a good question is (though you might need the reference material to understand why our definition of "good question" is what it is).

The goal of Stack Exchange is to have EVERY question on the site be eligible for a "good question" list.

I think that the solution is guidance on a per-post basis. Examples are nice, to be sure, but people will learn best when the example is their own. To that end, we should be encouraging people to provide helpful comments pointing to useful reference material or asking clarifying questions, or to suggest edits that clean up bad questions so that they meet our standards. A lot of people - IF they bother to look at this annotated list you're suggesting - will probably have trouble translating your examples to general principles that they can apply to their own questions.

Once we have a better Help center, it'll hopefully be easier to point people to a canonical instructional source of how to use the site. That generalized philosophy, combined with a concerted effort from experienced users to provide user-specific suggestions and guidance, should help us make big progress towards educating new users about our complicated but effective model of exchanging knowledge.

We do have a lot of mechanisms built in to deal with the people who deliberately make no effort to learn our system: downvotes, question bans, etc. deal with the worst offenders. For the people who just don't realize how we're different from other sites, better references, browsing the already great content here (people will gain an intuitive sense of what's good by seeing what types of questions have high scores; passive learning like that is important, too), and receiving specific suggestions for improvements will be a more natural way to learn than studying a list of example good and bad questions.

As a final note, I think that listing full questions that we don't want is a bad idea; I don't want people remembering those questions later, even on a subconscious level.

share
2  
The help center sounds promising, looking forward to that, but... to have EVERY question on the site be eligible for a "good question" list every one of the 5.5 million of them, probably 80% of which are duplicates of each other? I don't think that's a realistic goal. –  Pëkka May 8 '13 at 23:43
4  
"This is a massive amount of work to treat a symptom rather than a disease." Even if we allow for some rhetorical grandeur, that is simply wrong. First, it is not a "massive amount of work". It is a limited amount of work which can easily be quantified: The OP lists 5 typical problems ranging from "too vague" to "wall of code" and mentions the need to cover a few more subtle cases too. Add a couple of positive examples and we're done. Secondly, the OP doesn't propose to simply treat symptoms, but rather to teach people to fish, which is very different and much more constructive. –  Monolo May 9 '13 at 0:17
1  
Also, I'm with @Pekka on this one - defining a goal like "... EVERY question on the site be eligible for a "good question" list" (your emphasis) sounds to me like it won't scale in any meaningful sense. Have you made any estimates or calculations of the effort needed to achieve that goal compared to the number of active participants in the community? If you have, it would be really nice if you could share it. –  Monolo May 9 '13 at 0:27
2  
The help center sounds promising (and frankly a bit like the "new user chat room" that I proposed in this extremely controversial answer). That said, I feel that the infrastructure already provides an enormous amount of guidance. I think it's much more important to work on encouraging the community to be a little more proactive in helping new members learn, which I think is what you're suggesting here, but based on how many downvotes my answer received, I'm not sure how receptive the community is to that idea. –  Kyle Strand May 9 '13 at 5:48
1  
"The second major problem is that our guidance sometimes uses jargon that doesn't make sense to a drive-by user, or we use common words with a meaning that is different and very specific to our site. We'll be trying to avoid that in the new copy." -- this is an important point. Is this just going to be in the help center copy, or will it be a more extensive wording change? (I ask because I think some of the close reasons themselves, especially "not constructive", use really specialized jargon that doesn't align well with common usage). –  Ben Lee May 9 '13 at 19:45
    
@BenLee Obviously, we are not rewriting every word of copy across the whole network all at once. Eventually, maybe, but not just yet. ;) The major changes actively being worked on now are the Help pages and the close reasons. There are several other meta posts about close reasons, with a few more changes still in the works. If you have other specific things you think we should prioritize, feel free to make a request here on MSO. –  Laura May 9 '13 at 20:13
    
@Laura, thanks for the links. I'd already seen the 1st and 3rd of those, but that 2nd one is new to me -- and it sums up my feelings on the wording issue exactly! –  Ben Lee May 9 '13 at 20:17
    
The /ask page is ancient and needs to be rebuilt from scratch with better just in time contextual help based on who the user is and what they are doing. I worry that this new help center sounds suspiciously like the old help text that nobody read, either. This time it'll be different, I guess, but users won't read anything you put on the screen. At least not the ones who most need to. –  Jeff Atwood May 10 '13 at 6:20
    
@JeffAtwood You're totally right; we need to provide more just-in-time help, and the /ask page is almost certainly the most important place to do that. The new help center solves a different use case; I think the people who get burned most on our network are the new users who do genuinely try to figure out our system and just can't. If they're directed to (or find on their own) the help center, it should be a one-stop source of any reference/instructional material they need, which the current FAQ doesn't provide. That's definitely a small subset of users, but still important to address. –  Laura May 10 '13 at 14:55
    
I've been wanting to add more contextual help throughout the site for a while now; it's a larger, more complex project that we don't have the resources to focus on right now, but it is definitely still on our minds (and to-do list). –  Laura May 10 '13 at 14:57
1  
@laura it becomes a game of percentages of percentages of percentages -- how many people is this really helping? Remember too that every single new Stack Overflow user is forced to click through the How to Ask page before their first question. (Try it yourself; create a new account then click Ask Question on SO.) You could do that network-wide, with a better consolidated help page, perhaps, if we can avoid wall o 'text. That would reach those willing to read and try, though these users already do well on the network IMO, so then the question again: who is this really reaching? –  Jeff Atwood May 12 '13 at 19:52

A "before" / "after" view might also work

In many areas it might be worth using bad questions/answers that were later edited into great questions/answers for a direct side by side comparison.

As in - A closed question with -10 votes next to the same question edited and reopened with +10 votes.

This would not only show the value of editing, but would also help to show the value of taking the time to formulate a good question/answer and at the same time illuminate the possible consequences of a poor question/answer.

share

Abusing the answer functionality to get better formatting of what could have been a comment. You can add that to your don'ts if you wish :-).

This is a great idea, and even if it will occasionally be misunderstood, misinterpreted, misused and a couple of other misses, it will still help nudge things a step forward.

Just a couple of thoughts for good examples of the down-to-earth variety:

  • Break long paragraphs up into smaller units
  • Capitalise sentences
  • Spell check
  • Reduce code to about 10 lines or less if possible

These are little things, but they help change the impression of a question from being gibberish to at least parseable. That may actually fend off a close vote or two. It is also fairly easy to show the difference in an example and sufficiently mechanical in nature to be practical also for non-native speakers.

share

This site is currently not accepting new answers.

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