Last year we organized a semi-formal backlog-pruning event called the Winter Refactor. With the StackExchange holiday season rapidly approaching, I'd like to solicit input on running another Winter Refactor event this year.

In particular, I'm interested in the following opinions, if you have them:

  • Should we bother?
  • If we should, what should we focus on?
  • What can we do to drive participation?

Of course if you have other thoughts beyond those, I'd like to hear them as well.

  • 1
    \$\begingroup\$ The majority vote seems to be towards "do the same thing," so I have. I did include some words about duplicates though. \$\endgroup\$ – user1430 Dec 15 '14 at 16:57

How about motivating killing some unanswered questions by having a chart that shows the progress we've made?

EDIT: OK, I made one. Live chart!

EDIT 2: Live chart replaced with how it looked at the end of the event.

number of unanswered GDSE questions

It's number of unanswered questions on GDSE over time, from around midnight December 9th, when I left the data collection running, until right about now. The times are UTC.

I mashed it together from the StackExchange API, D3, Node.js, Inkscape and ImageMagick. The code is on Github. It renders a new image every 10 minutes and tells your browser to cache appropriately.

  • 5
    \$\begingroup\$ This is very cool. \$\endgroup\$ – user1430 Dec 9 '14 at 20:15

One option is for us to do, essentially, exactly what we did last year: focus on reducing the unanswered question queue by either answering questions, voting to close bad questions that have fallen through the cracks, or upvoting good zero-score answers.

This worked pretty well last year, but the big disadvantage is that the broad focus makes it hard to track progress, especially on a per-user level, so offering rewards for participation is pretty much limited to manually awarding bounties for good answers.


Make softball/canonical questions

We get a lot of the same kind of questions: mostly from folks starting out, but the answers would be too broad or opinionated for the Q&A format. Some examples: "I want to make X / do Y, what do I need to know?" "Should I use this or do that?"

Right now we close these questions to prevent attracting low-quality answers, but I think we can do better. One problem with closing (as off-topic) is that, by saying "we don't welcome questions like these", we're implicitly saying "we don't welcome you here", we drive away users who probably don't come back, even when they've gained enough experience to be able to contribute better to this site. Instead, if we build a collection of softball or canonical questions and redirect duplicates to these, we can say "this is a better way of asking questions here", which encourages people to stick around and learn.

Apart from answering common questions, we can also help narrow down those too-broad, opinion-based questions. Often what the asker really needs is information to help her make a decision, and I think for many opinion-based questions, there are ways to address them in a fact-based manner. To make a car analogy, in order to answer the question "what car should I buy", instead of having a huge poll with options like "Toyota" or "Sedan", a good answer might follow up with a list of criteria such as "what's your budget" or "what will be the primary use". Only the asker can make the final decision, but at least she can now make a more informed decision, or follow up with more specific questions.

I think we can use the refactor as an excuse to do this work.

How this might work

  • Find a type of question that keeps getting asked and closed as off-topic, but might be addressable
  • Ask the question in the most generic, broadly-applicable way possible, maybe with a tag like / and marked as CW
  • Answer and revise as usual, avoiding opinion
  • Where suitable, mark other questions as duplicates of these
  • At the end of the event, publish a list of top contributors to these questions (# of questions/answers, score, # of duplicates), with awards like bounties or hats if they can be arranged
  • 1
    \$\begingroup\$ A few years ago, in response to a rash of questions in the form of "What engine should I use to make <Game X>?" (which are explicitly off-topic for reasons that I agree with), I posted a proof-of-concept question: What to consider when evaluating libraries and engines for making a game?, as an attempt to reframe the question in a generic way. \$\endgroup\$ – Trevor Powell Dec 9 '14 at 1:39
  • 1
    \$\begingroup\$ While it received upvotes and some interesting answers, people didn't feel that it served as a viable replacement for the "please just tell me what to do"-style "what engine should I use for my project" questions. Coming up with canonical questions that people will accept as canonical is tricky. Particularly when they redirect responsibility for picking a final answer for their specific circumstances back to the questioner. \$\endgroup\$ – Trevor Powell Dec 9 '14 at 1:41
  • 1
    \$\begingroup\$ While I like (some parts) of the "canonical question" idea, I don't necessarily think it's something appropriate for the short term push of the Refactor. I have a few proposals in mind for a longer-term project to address the issue that I was hoping to surface after the holidays when fewer people are likely to be off on vacation and the like. \$\endgroup\$ – user1430 Dec 9 '14 at 2:46

One option I've been thinking about is focusing on reducing the amount of duplicate questions.

We have a fair number of questions that are effectively duplicates (lots of questions dealing with rotating objects to face other objects, dealing with resolution independence in UI layout, where to put the renderer in a component-based entity system, et cetera).

Many of these questions are standing alone, without duplication links. In some cases, the questions we are using as the duplicate target aren't, themselves, the best example of the fundamental question. We could focus on correcting those problems.

The downside is that this excludes a lot of low-reputation users from participation, beyond flagging or suggesting duplicates in the chat.


You must log in to answer this question.