32
$\begingroup$

As a community, we expect new users of the site to do a lot of work before making their first post (which is on average a new question). They have to

  1. make an account and sign up,
  2. be at least somewhat familiar with how to ask a good question,
  3. do some work to look for other versions or variations of this question (or quickly have the question be closed),
  4. be sufficiently familiar with mathjax (and usually, English) that they can format their question,
  5. recognize that merely asking a self-contained math problem is different than avoiding asking a "Problem Statement Questions",

and more.

Right now, we make users create an account and have a very quick tour through the help documents. While writing a new question, there are three links to the How to ask a good question meta post, two links to meta, and one link to the help center --- and there is a tiny set of suggestions coming from the system on related questions, titles, and formatting.

Then they are let loose to sink or swim.

Should new users be given further guidance? What would that take? An extremely straightforward (but probably effective) change would be to include a sentence next to the other advice that pops up when writing a question saying that isolated problem statements are usually quickly closed.

This is just like having new people join your company, or your department, or your club, or your other group. We have an onboarding process, but maybe it's stale.

I began to wonder:

In an ideal world, what additional/different onboarding would you like for new users to Math.SE?

Let's ignore the fact that we (the site itself and the site moderators) don't have the power to make these changes happen. I'm setting this in a world of fiction. The stars are the limit. While in the long term I hope that something useful might come out of this, in the short term I'm just wondering what ideas there are out there.


Related Posts:

$\endgroup$
5
  • 5
    $\begingroup$ (1) Have many poster examples of "good" and "bad" questions shown to new users before they ask their first question. These show what role source, motivation, definitions etc. play in a question so that users find it natural to mention these in theirs (2) Triage : Have helpful but firm users vet each new question in chat before it is posted, where vetting consists of (a) gentle, non-invasive editing, if required (b) rejection if required (c) not addressing the actual question in any way as much as possible. All of this subject to some fine-tuning, obviously. $\endgroup$ Commented Feb 29 at 2:11
  • 2
    $\begingroup$ @SarveshRavichandranIyer Since you mention examples of good/bad questions, I will point out these previous posts (one of them mine): Examples and counterexamples of good questions and answers, Standard example of well asked question. and Suggestions for examples of well-asked questions. $\endgroup$ Commented Feb 29 at 4:43
  • $\begingroup$ I remember that you previously mentioned triage in your answer here: Three levels of Math. A related posts posted around the same time: What are Triage and Help and Improvement review queue? $\endgroup$ Commented Feb 29 at 4:47
  • $\begingroup$ Is this also related meta posts like this one where it says “ taking steps to improve onboarding (one of, if not the single most impactful area for change right now) to both main and especially meta to make meta more representative” $\endgroup$ Commented Feb 29 at 16:38
  • $\begingroup$ @Tyma I think the actual SE Community Team acts very afraid of adding any hurdles for new users. More generally, I think SE as a company would try to do almost anything they can think of that might bring more people to the SE sites without adding friction. I think Zoeisonstrike (an SO mod, who wrote the onboarding line you mention) and I are both reacting to nearsightedness on the SE Community Team's part. $\endgroup$
    – davidlowryduda Mod
    Commented Feb 29 at 21:43

2 Answers 2

31
$\begingroup$

All questions posted by new users should start "Closed".

I think that one of the major friction points on the site is that new users see close votes as a kind of "punishment", and they believe that they are being judged as somehow unworthy (rather than understanding that a close vote is about the question itself, and isn't generally a value judgement, but a question of fit for the site).

If every question by a new user starts "Closed", and must be opened by a vote of the community before it can be answered, I see two possible benefits:

  1. New users are taught that closure is not the end of the world, and isn't even really a bad thing. A closed question can be worked on and improved, and will often be opened once it meets the requirements of the community. Once a new user moves beyond the "every question asked starts 'Closed'" beginner's sandbox, they make take this philosophy forward with them, and (a) be less hostile to close votes directed at their own posts, and (b) more willing and able to cast close votes in the future and to offer advice to the askers of closed questions.

  2. Veteran users will be required to at least wait for five open votes before answering low-context problem statement questions. Rather than posting a "fastest gun in the west" answer and earning the 25 XP for an upvote and green check from the question author, answerers will be required to, at minimum, wait for the question to be opened. They may take this as an opportunity to invest some time into helping the author improve their question, and/or doing some of the work to search for existing answers to the question, which often exist on the site.

The site should give more emphasis to the "passive search" approach.

The StackExchange network is designed to be, first and foremost, a "wiki"-style site. It is meant to be a reference work, not a social network or tutoring service. This is articulated in Jeff Atwood's post What does Stack Overflow want to be when it grows up?:

Complicating matters further, there are three tiers of usage at Stack Overflow, from biggest to smallest, in inverted pyramid style:

  1. I passively search for programming answers.

    Passively searching and reading highly ranked Stack Overflow answers as they appear in web search results is arguably the primary goal of Stack Overflow. If Stack Overflow is working like it's supposed to, 98% of programmers should get all the answers they need from reading search result pages and wouldn't need to ask or answer a single question in their entire careers. This is a good thing! Great, even!

  2. I participate on Stack Overflow when I get stuck on a really hairy problem and searching isn't helping.

    Participating only at those times when you are extra stuck is completely valid. However, I feel this level is where most people tend to run into difficulty on Stack Overflow, because it involves someone who may not be new to Stack Overflow per se, but is new to asking questions, and also at the precise time of stress and tension for them where they must get an answer due to a problem they're facing … and they don't have the time or inclination to deal with Stack Overflow's strict wiki type requirements for research effort, formatting, showing previous work, and referencing what they found in prior searches.

  3. I participate on Stack Overflow for professional development.

    At this level you're talking about experienced Stack Overflow users who have contributed many answers and thus have a pretty good idea of what makes a great question, the kind they'd want to answer themselves. As a result, they don't tend to ask many questions because they self-medicate through exhaustive searching and research, but when they do ask one, their questions are exemplary.

Search on Math SE is hard, because it is hard to catch notation in a search result. The on-site search is terrible; Google is okay; and approach0, while often pretty good, is not a well known tool. Recently, our SE overlords have spent a lot of time talking about the wonders of generative AI. One place where this might actually be a good thing on this site is in searching up existing questions which might be helpful. This is very pie-in-the-sky, but I think that part of the "Ask a Question" dialog should a stronger search functionality which (a) does a better job of finding related questions, and (b) more prominently encourages users to read those questions before posting a question of their own. It seems to me that this is precisely the place where the kinds of AI tools that SE is touting would be most effective.

However, even if such tools never appear, users should be encouraged to search more, and ask questions less. Perhaps part of the "Ask a Question" process should be an entire screen with a list of related questions which one must pass through before posting a question (something a little more intrusive than the list of questions on the sidebar, which is easily missed). Something like 90%+ of interactions on Math SE should start and end with a search. (For reference, I had a been using Math SE for years, via search, before I ever created an account, and I have never felt the need to ask a question on my own behalf, because every question I've ever needed an answer to has already been answered on the site).

Implement existing tools from SO on Math SE.

StackOverflow already has tools which are meant to help onboard new users, including the Staging Ground (which is currently being used in a limited manner), and the Triage Queue. These would be helpful to have on Math SE, as well.

$\endgroup$
5
  • 3
    $\begingroup$ These are some pretty major restructurings of the site's working, but I think there's a lot of good ideas in there. If implemented, I think the current "Ask a question" framing should be replaced with "Get an answer to a question". $\endgroup$
    – JonathanZ
    Commented Feb 29 at 15:04
  • 2
    $\begingroup$ I like this idea, but I'd wish we go back to using "on hold" (temporarily, if improvements are made by the OP, but if not, a firm close). The Staging Ground and Triage queue are great ideas; too many poor questions remain due to insufficient reviewing. The rate of down votes for PSQ's, e.g., has declined, as the number of users answering them, has escalated. I think users reluctant too close, might warm up to the staging room, etc. or other ways to help new users. $\endgroup$
    – amWhy
    Commented Mar 2 at 1:24
  • 3
    $\begingroup$ @amWhy Sure. I'm kind of indifferent to the language used to describe a question which is still visible, but which cannot be answered. Though I do think that whatever language is used should match the language used later---train user to understand that being "closed" or "on hold" or whatever isn't a bad thing; it is only a step towards constructing a better question. $\endgroup$
    – Xander Henderson Mod
    Commented Mar 2 at 14:16
  • 10
    $\begingroup$ Just because "on hold" vs. "closed" is indifferent to you, @Xander, doesn't mean it makes no difference to the OPs. Words are powerful, and if you feel it important to make clear that such an action is not permanent, "on hold" is far more encouraging to "work on your question" than is the verdict "closed". But I agree whatever the language, it is imperative to emphasize, particularly with onboarding new users, that improvement is encouraged. Note, I upvoted your answer, and responded in agreement. I merely offered a suggestion which would further your goal. Good luck! $\endgroup$
    – amWhy
    Commented Mar 2 at 22:41
  • 3
    $\begingroup$ But, just a reminder to all, just because a user presents as a "new user", it is not at all uncommon that users abandon a username, create a new account, even when they are not new users. $\endgroup$
    – amWhy
    Commented Mar 2 at 22:46
6
$\begingroup$

Posting my comment above as a more detailed answer.


Demonstrate by example

Have many canonical examples of "good" and "bad" questions shown to new users before they ask their first question. These show what role source, motivation, definitions etc. play in a question so that users find it natural to mention these in theirs.

Make sure these are "textbook examples" i.e.

  • The good questions are highly upvoted, have many views and have highly upvoted answers.

  • The bad questions are closed, are rated below $0$ and have poor answers, if any at all.

  • The role of context in each question is clearly emphasised by highlighting comments and answers that make constructive/explicit use of it or mention its importance.

One should ensure that special classes of questions are dealt with under these examples. Examples include (but are not limited to) , , .


Triage/vetting

Have helpful but firm users vet new questions in a chat with OP before they are posted. Vetting consists of

  • Guided editing of the question by OP, where helpers probe them and equip the user with what is necessary to improve the question, be it MathJax skills or addition of context. Helpers should be allowed to edit the question but non-invasively, as a positive demonstration of improvement to the newbie e.g. spacing MathJax equations and text properly, clearing up grammar and translation errors etc.

  • Rejecting the question if required. Take something off-topic like spam or a six-line proof of the RH. Rejection is done with the mindset of educating the user to not post such questions. If the user themselves refuses to make edits or disagrees with the guidelines itself, one should realize that not all users can be onboarded and that sometimes sending people elsewhere is the right thing.

  • Not addressing the actual question in any way as much as possible. Addressing the actual question won't show the users how to create a question, it'll tell them that they can get their question answered at the primary stage itself.

  • Allowing the question to be displayed on the front page when ready.


On request of @D S

"New users are often confused by our need for "context", and a simpler comment template is needed" , keeping in mind that some users may learn only with a chat : maybe a more chatty comment could be useful in some places. A bot could help with this, maybe?

$\endgroup$
4
  • 1
    $\begingroup$ When you mentioned vetting new questions in chat, it reminded me of this previous question: Would a chatroom for new users be feasible? Since it is important to search before asking, I will add that we already have a chatroom related to searching. The main issue here is that a user needs 20 reputation points before they can use chat. $\endgroup$ Commented Feb 29 at 5:19
  • $\begingroup$ @MartinSleziak In response to your comment(s), you're right : some previous discussions have been quite valuable. As another example in addition to your links, triage was implemented on Stack Overflow (without some features mentioned above) and plenty of feedback was given there. Nevertheless, the important point is that, in scale, the number of questions on MSE is small (and has been coming down steadily), so I believe triage would hopefully work out better here. $\endgroup$ Commented Feb 29 at 5:31
  • $\begingroup$ Can you also add this n your answer: New users are often confused by our need for "context", and a simpler comment template is needed. For example, see this. $\endgroup$
    – D S
    Commented Feb 29 at 14:57
  • $\begingroup$ @DS I have done so, but keep in mind that understanding the comment and disagreeing with what it says are two different things. Feel free to edit that part of my answer. $\endgroup$ Commented Feb 29 at 15:06

You must log in to answer this question.

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