2

I've been ruminating on what is a 'good question' and I'm not yet totally convinced that Optimizing For Pearls, Not Sand neatly answers the question. I don't think a question being interesting defines whether it's good or not.

I can break down somebody else's question into one of three categories.

  1. It is a super easy question for me to answer because I learned it last year. Good because I'm the first to answer, I can get points and I want high score!
  2. This is a complicated question requiring abstract knowledge that is way out of my league, perhaps a year from now I'll be able to answer it.
  3. This is an interesting question. I don't know the answer but I have enough knowledge to piece together the parts, I'll load up the Views UI and test my solution before submitting my answer. Answering this question brings me the same pleasure as doing the New York Times crossword puzzle in pen or today's Sudoku.

Nevertheless, what makes an interesting question is completely subjective. I'm sure that whatever the modern day computer science equivalent to the question, "Is cardboard the proper material for a light bulb filament?" which was asked in 1879 would get nothing but down votes for being a ridiculous notion.

I will never ask a question that Mark Trapp thinks challenging or interesting. Does that mean all my questions are not good?

Putting the subjective aside since one expert which this site wants to attract might have a different idea of interesting than another expert. Is there a blueprint diagram of a properly formed question? Is there data to back this up?

I can't speak for questions but I have tested what makes a good weekly email newsletter for a newspaper, publishing website. I can also back up my arguments with data and facts since MailChimp tracks these things. First, putting body, lead or teaser text is a good way to get people to ignore the email. Second, using artsy, creative, wordplay headlines in links does not get people to click through to the website. (This, by the way, is something that will never be able to be explained to an editor coming from print.) I have data that shows writing plain headlines that describes exactly what the user will find on the other side is the best way to get people to click through.

This might be true for Stack Exchange questions. Having a clearly writen title that is neither too long nor too short that describes exactly what the question will be might be the best way to get people to click through to read the question.

What are the other elements of a question and the body of the question that get answers? Can a good question be diagrammed, with all the elements that make it good, labeled demonstrating the pattern? Is there evidence or data that can back up these conclusions the way I can back up my assumptions about the anatomy of a good email newsletter, one that gets a lot of people onto my site?

4

We've been discussing how to improve/ask better questions here:

During writing these questions, I stumbled over https://drupal.stackexchange.com/questions/how-to-ask

I think that link has very good points, I was more or less thinking about similar points but couldn't explain it that well (and short, my questions/answers always get long..).

As you said, a good title is important. One tip to write good titles is IMHO to think about how you'd search for this problem on Google (I guess you will never search for "Why doesn't this code work?" ;)).

Jeff's answer is actually really good. You could ask the perfect question, do everything right. If there is nobody who can answer your question, because there is no expert on that topic coming to this site, it won't get answered. And if it is not answered, it's not useful to anyone.

I personally like specific questions, that are however not about a certain piece of code, are therefore also interesting for many other people and which you/I can actually provide a good answer to. Some examples:

To both of these questions I've linked to in other answers/comments already, because they were also relevant there. If you look at the most upvoted questions ( https://drupal.stackexchange.com/questions?sort=votes ), I think a large part of the questions there are similar.

  • What I'm asking is what do all the good questions have in common. Of course, they must have great answers because that's how we defined that they are good in the first place. However, there are patterns to the good questions. Can you create a template for a good question? I am sure the template for a good question begins with thoughtful and proper grammatical structures and correct spelling for two reasons. First, although not necessary proper grammar usually communicates better. Second, poorly written questions show laziness and nobody wants to help a lazy person. – Adam S Jun 18 '11 at 10:45
  • 3
    There is no template to write a good question. There are only guidelines and tips. As I said, you can do everything right when asking a question when there is nobody who can/want to answer it then it won't be answered. And, by Jeff's definition, then it's not a good answer, as it not useful to anyone ;) – Berdir Jun 18 '11 at 14:10
  • If you laid out a printed copy of the best 500 questions and using colored markers started circling elements the questions share, a pattern would emerge. Saying 'there is no template to write a good questions' isn't useful either especially since like everything else in life there is always a pattern. If the case is doing everything right asking a question and nobody can answer it then we have to accept the fact answer speed shouldn't be a criteria -- as I argued somewhere else -- in defining a good question as some questions will be before their time. – Adam S Jun 18 '11 at 18:45
3

Saying something is subjective (or more accurately, a matter of preference) is a bit of a cop-out: it's not the same as me preferring mint chocolate chip over rocky road ice cream. There are objective criteria being employed to make a sound judgement call about the value of a question. You can easily write a question that's interesting to me (and others): my criteria aren't as inscrutable as my love for mint chocolate chip ice cream.

Basically, a question has to look interesting without me ever looking at the question body. This is what I was getting at in my answer quoted in "Optimizing for Pearls, Not Sand": I have better things to do than to open up 50 questions in hopes that one of them is going to be solid gold.

Experts, or discerning Stack Exchangers, answer questions when they feel like their time isn't wasted by doing so. So how do you know a question got it right? As Jeff alluded to, when it has great answers. Great questions invite great answers. They beckon to them like sirens.

What's an inviting question? In order of priority:

  1. It's got up-votes, or at least no down-votes.
  2. The question title is a fully-answerable question.
  3. The question has interesting tags.

(1) and (3) are self-explanatory, so I'll talk about (2).

The question title is the hook: it's the first line of a novel, the cold open in a TV show, the trailer to a movie. It's what determines whether I've been convinced to find out more. In the context of Stack Exchange, what's convincing is being able to determine if I can answer this question or I want to know the answer to this question.

It doesn't mean I don't need more context, but if I can say "Oh, I probably know the answer to this" by virtue of the title alone, I'm extremely likely to click through and find out the details.

To this end, these question titles suck:

On Webform and Drupal 6

Why is this not working?

How do I build a high-synergy e-commerce solution with Views?

These questions are things I would find interesting:

Show all nested menu links in Drupal 7

Form Redirect not working if 'destination' is in URL - Drupal 7

How do I return the output from a callback function unthemed?

Coincidentally, these are questions I've actually answered. They're not 100% awesome titles, but they explain exactly what the user is looking for without me ever having to read the body: I was pretty confident that I'd be able to answer them just by seeing them in the question list.

This is all to say that you can write the Great American Novel in the question body, but it doesn't matter if I never click through.

If you did happen to write a really useful title but left a really bad body, that's where the down-vote comes in: it's a warning to others using the same process, "don't bother, it's bad."

So the "perfect" question comes down to two things:

  1. A useful, answerable title
  2. A question body that provides enough context to answer the question

It's very simple: I think a lot of people over-think the existential nature of a question, which makes their questions worse for it. You don't have to "smart" up your question, make it funny, or use some silver-bullet question template to attract experts: just convince me I'm not wasting my time by trying to answer it.

All this should go to say that if this was 1879, and I was Thomas Edison himself, I'd answer "Is cardboard the proper material for a light bulb filament?" so hard.

2

Since the nature of your question is philosophical, my answer will be as well.

A great question is largely defined by its answers.

  • Asking questions about questions can be a dangerous endeavor, mostly because being caught in this recursive thought little work gets done. Nevertheless, eventually, you will be invited, if you haven't done it yet, to give a TED talk -- at least I hope. You will probably be expected to discuss the nature of questions and answers. Despite the simplicity, elegance and beauty of this answer, I don't think it will fill the whole 18 minutes. For that reason, I will just leave it unchecked. Hopefully, other people will contribute. :) – Adam S Jun 18 '11 at 8:55
1

This is a comment to Mark Trapp's answer. I feel that I might not have enough space to respond in a simple comment.

It was in second grade that I learned the definition of a sentence. A sentence is a complete thought. The first two titles that you think suck have something in common, they are not sentences and therefore they are not complete thoughts. The first sentence, On Webform and Drupal 6, is a preposition with neither a predicate nor a subject and the second sentence uses the pronoun this without an antecedent.

Of the titles that you like but not 100%, two -- although you called them questions -- are not questions at all, Show all nested menu links in Drupal 7 and Form Redirect not working if 'destination' is in URL - Drupal 7, which might explain why you are hesitant to call them great questions.

All interrogative sentences begin with a pronoun (what, who, whom), a proadverb (where, when, why, how, how often) or a prodeteriner (whose, which, what).

Although we do not currently view a proper grammar in Shaw's 'Pygmalion' sense by adding social value to it, grammar is, nonetheless, a reflection of patterns of thought that are analogous in usefulness as a tool as design patterns are in computer science such as Object Orient Programming and MVP. Let's analyze the two of the six questions you use as an example that are actually questions. This is done by rearranging the interrogative sentence into its declarative equivalent.

How do I build a high-synergy e-commerce solution with Views?

becomes

I build high-synergy e-commerce solutions in this manner.

A question is constructed by taking the missing or uncertain part of a thought, converting it to a proadverb or pronoun and moving it to the front of the sentence.

And,

How do I return the output from a callback function unthemed?

becomes

I return the output from a callback function unthemed in this manner.

Both of these sentences are complete thoughts. The proadverb's computer science equivalent is a variable or SQL placeholder. The syntax structure of a question is no less important than the syntax structure of an SQL query. However, since we are much more forgiving as humans about the syntax structure of questions than compilers are of SQL query syntax we have developed some bad habits.

If we start to delve into the anatomy of a good question, one that gets great answers, on Stack Exchange, the first pattern to emerge is all the best questions are really questions, not declarative sentences or sentence fragments. Moreover, we will also discover that they are complete thoughts, a sentence which is a pure, self contained thought.

A question shouldn't need a body. The body is there add clarity, previously acquired data or knowledge and context to the question which should also be considered a part of the anatomy of a Stack Exchange question.

If Jeff really believes that you, Mark, have 'summed it up best' and what is most important about the question is that 'the question title is a fully-answerable question' then Jeff has made a huge semantic error. The field 'Title' on every question form is slightly misleading. Although, it is technically a title, it is more specifically a question and should be labeled accordingly, 'Your Question.'

I agree with you that a well formed title is key as I explained by my experience with email newsletters. The CEO or Sony Pictures or Paramount, I forget, once said that he only allows 30 seconds for someone to pitch a movie because that is the extent of time his marketing department will have to pitch it to the public. I disagree with you on mint chocolate chip ice cream. It was the best until I discovered Ben & Jerry's 'Everything But The ...'

  • I'm sorry, I thought you wanted to better understand how to get questions answered on a Stack Exchange site. It now sounds like your question is more general and might be better suited for Writers.SE or English.SE. – user7 Jun 19 '11 at 19:13
  • A long time ago when I was a kid I live in Cambridge, MA. There was a joke about Harvard students not knowing how to add and MIT students not knowing how to write. My argument is we could learn a little from each other because these things we are calling questions on Stack Exchange are not really questions nor are they clear, full thoughts. Now, since I have a well formed hypothesis of what makes a good question -- that it is simply a real question -- I can put it to test to collect data. – Adam S Jun 19 '11 at 19:39
  • 1
    @Adam There's a tendency within the Drupal community to re-think problems because they weren't invented within the Drupal community, that Drupal should always be special-cased, and I think that's a real mistake. While you may have only realized there's a problem yesterday, the Stack Exchange network has been going on for a couple of years now and has fielded millions of questions. Drupal questions are not materially different from the other dozens of types of questions Stack Exchange handles, and the lessons learned elsewhere on the network can and do apply here. – user7 Jun 19 '11 at 19:46
  • @Mark I only realized that I should be asking the question, "What is a good Stack Exchange question?", because I had one of my questions down voted and someone asked something about it in meta. So I asked to others and myself what makes a good question. I'm not satisfied with any of the answers so far because they are vague, not concrete and open to argument. Do you really believe that because a large group of people are doing something that it is right or can not be improved on? – Adam S Jun 19 '11 at 20:15
  • 1
    @Adam I'm saying that any test you're about to do is not going to compare in any material way to the sample size already collected. Rather than rejecting or dismissing all the analysis and lessons that have come before, I'd spend my time trying to understand (by reading and introspection) how the system works and why it works so well. – user7 Jun 19 '11 at 20:44
  • I don't believe that the concept is so abstract that an experiment can't be made to test it. My argument is that we can track performance as I did with my email newsletter experiments. Removing the body text from the email increased the site traffic by three over a one month period which resulted in a dramatic increase in advertisement revenue. We later experimented with cleanly written titles in the newsletter which further increased performance. Has anybody done this with Stack Exchange and where is the data? Right now I'm thinking about what the control of the experiment would be. – Adam S Jun 19 '11 at 21:06

You must log in to answer this question.

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