22
$\begingroup$

Probably anybody visiting this site regularly noticed that quite often there are posts which are not at MO-level and quite frequently also of very low quality. (Many of them are simply copy-paste of an exercise or an assignment. Users with sufficient reputation can find some such posts in moderation tools or in close vote review history.) And there have been some complaints about these posts here on meta, for example: Proposal to more aggressively delete off-topic questions or Can the clutter of obviously off-topic questions be removed faster?

Some time ago, Mathematics StackExchange started to require registration before asking a question: Let's require registration to ask a question. (Unregistered users still can answers questions and make edit suggestions.)

I do not have any data which could quantify to which extent closed/deleted questions on MathOverflow are posted more often by unregistered users. (And I do not know whether and how such data can be obtained.1,2,3,4) However, judging by the off-topic questions I have seen, this is quite often the case.

I suppose that users who did not bother to read in the help center what is and isn't on-topic on this site, are more likely to choose a site which does not require registration. (If they know about the existence of two sites at all.) So my guess is that requiring registration might help reduce this type of questions. (At least the ones coming from unregistered users.)

Of course, there might be also negative consequences to this change. Maybe som unregistered users will simply post their questions as answers. And probably some good questions have been posted by unregistered users - this would be no longer possible.

TL;DR: Should MO require registration before posting a question? Would this help to decrease frequency of off-topic questions? What are possible negative consequences?


1There is currently one user which is a moderator on both sites. The mods have access to some data and tools which regular users do not have - perhaps moderators from Math.SE are able to assess to which extent this change helped. And moderators from MO might be able to say to which extent this really is a problem.

2Some analysis concerning number of closed questions after joining the Stack Exchange network was done here: Has the switch to SE 2.0 increased the rate of low-level / inappropriate questions? However, in this case it is more interesting to know to which extent unregistered users contribute to the closed questions.

3 Based on the comments here it seems that such data cannot be obtained from SEDE. Some statistics on this were made for StackOverflow in 2011, about the time when SO started to require registration for asking questions: Stack Exchange Community Statistics results: In which ways do unregistered users participate? Looking at the results there, unregistered users were responsible for more than 10% of questions (before adding requirement to register) and about 3-4% of answers. However, it's hard to say to which extent behavior of users on MO and SO might be similar.

4In the absence of better data, we can simply look at the questions posted yesterday. This is a very small data sample - on the other hand, the period is short enough so that this will probably not be influenced too much by the questions which have been deleted in the meantime. (Users with access to moderation tools can check whether there are many recent deletions. However, there are some posts which were closed and deleted in that period - I was able to find one deleted post in my flagging history. But this was a post deleted by OP - so it is not shown in the moderation tools. Only registered users can self-delete posts.) And one-day period is also short enough to go through manually.

Of the 6 questions posted by unregistered users, 3 were closed as off-topic, 3 remained open. Among the questions posted yesterday, 5 were closed; 3 of them posted by unregistered users, 2 from registered accounts. See here for the complete list of the questions. (Of course, this is based only on the question which were are still undeleted.) All 21 answers posted on the same day were posted from registered accounts. (Again with the same caveat - possibly some answers might have been deleted by the community since then.)

$\endgroup$
  • 8
    $\begingroup$ Even sites which are rather not known to care very much about the quality of contributions, like e.g. Facebook or Twitter, require registration before posting. -- I'd support MathOverflow doing the same. $\endgroup$ – Stefan Kohl May 6 '17 at 13:26
  • 3
    $\begingroup$ Alternatively, one could take the Wikipedia approach and give the IP number instead of the user name / user_xyz for contributions of unregistered users. $\endgroup$ – Stefan Kohl May 6 '17 at 13:41
  • 4
    $\begingroup$ @StefanKohl I am a bit hesitated on this suggestion since some of the experts are just too busy to maintain an account yet their answers are very important in clarifying their own works. Like this post mathoverflow.net/questions/266921/… which I cannot imagine anyone except the author can answer. $\endgroup$ – Henry.L May 6 '17 at 18:43
  • 2
    $\begingroup$ ...and I have the same opinion for the askers, someone may just be too busy to maintain an account but that does not necessarily mean that a first post from an unregistered user must be of low-quality. And the questions on the homepage will serve as a good reference for unregistered users to see if their questions are appropriate. So I am thinking that the negative side will outweight the positive side of this suggestion. $\endgroup$ – Henry.L May 6 '17 at 18:49
  • 1
    $\begingroup$ @HenryL I will point out that the discussion whether the unregistered users should be prevented from asking questions. As I mentioned in the post, at math.SE (where something similar was implemented), posting an answer does not require registration. Neither does suggesting an edit. $\endgroup$ – Martin Sleziak May 6 '17 at 18:57
  • 2
    $\begingroup$ @MartinSleziak Erhh, I said that I have the same opinion for askers...is there significantly more unregistered askers than answerers? If that is the case then I will revoke my comment and apologize for ignorance. $\endgroup$ – Henry.L May 6 '17 at 19:09
  • $\begingroup$ @Henry.L Your comments raise certainly a valid point. My guess would be that the situation you describe (an expert on a given area making a post) happens more often for answers than for questions. (I mean mainly the case of posting an answer on something related to their own work.) In general, I would guess that unregistered users more frequently ask questions than post answers - but this is probably unreasonable extrapolation based on the posts I've seen, which is a rather small sample. Let's hope somebody more knowledgeable will know whether it is possible to obtain some data on this. $\endgroup$ – Martin Sleziak May 7 '17 at 6:30
  • $\begingroup$ I should also add that more than 3% percent of answers posted by unregistered users are more than I would expect. (This is from data about StackOverflow from 2011 - I have added link in my recent edit. It might be quite different here at MO - but still, this is the only data I was able to find.) $\endgroup$ – Martin Sleziak May 7 '17 at 6:36
15
$\begingroup$

I do not see how registration will decrease the number of poor quality questions. From time to time I look at MathSE, and the number of poor quality questions there is larger than here, despite the registration requirement.

On the other hand, there is a class of people who really hate any registration, and this requirement can deter them. I am in this class myself. And this set of people is independent of the quality of questions they can ask.

But I agree that some mechanism for more prompt deleting of non-appropriate questions is desirable. Perhaps we should force a person to read the rules before posting the first question? By clicking some button "read and understood the rules".

$\endgroup$
  • 8
    $\begingroup$ Re: Perhaps we should force a person to read the rules before posting the first question? It might be worth mentioning that something like that is implemented at math.SE, at SO and probably some other sites. So it seems that Stack Exchange software has possibility to do this and I guess that this is done on individual sites on the requests of moderators/specific community. (In both cases, the "How to ask?" page is shown to new users.) $\endgroup$ – Martin Sleziak May 7 '17 at 10:41
  • 2
    $\begingroup$ Thanks God, so far I have seen nothing illegal on our site. So it will be fine if we can continue without registration. $\endgroup$ – Alexandre Eremenko May 7 '17 at 15:59
  • $\begingroup$ @AlexandreEremenko, don't be so sure. "European court says linking to illegal content is copyright infringement", especially by for-profit publishers, and StackOverflow is for profit. searchengineland.com/… $\endgroup$ – Matt F. May 8 '17 at 16:58
  • 1
    $\begingroup$ @Matt F.: Tanks for the warning. I will not link the illegal content to my q/a. $\endgroup$ – Alexandre Eremenko May 8 '17 at 18:06
  • 1
    $\begingroup$ @MattF. I think that linking to copyrighted material is only tangential to this discussion which is about low-quality and off-topic posts. (At least that was my intention.) It seems that there were some discussions related to this both here and on tea. $\endgroup$ – Martin Sleziak May 9 '17 at 14:30
  • $\begingroup$ @AlexandreEremenko If you think that making some page automatically shown to new users might actually help, a reasonable first step towards getting this implemented would be making a feature request here on meta. (To me your suggestions seems reasonable and, as I mentioned in my first comment, it is already implemented on some sites.) $\endgroup$ – Martin Sleziak May 9 '17 at 14:35
  • $\begingroup$ The questions has been around for two weeks, I do not expect much more views and votes to come. I accepted this answer since it seems to reflect position of the community - judging both from the voting and from the comments posted here. $\endgroup$ – Martin Sleziak May 21 '17 at 4:27
  • $\begingroup$ Some time has passed since this discussion - I have made a separate post on meta about the suggestion at the end of your answer: Should users be shown some basic information before posting the first question? $\endgroup$ – Martin Sleziak Sep 16 '17 at 11:15
7
$\begingroup$

Let me rephrase the suggestion as I see it. The problem: too many undesired questions asked by new users; suggested tactic: require new users to register before asking a question; the post asks for opinions on whether this tactic will be effective/should become policy.

I think a different perspective should be considered. The problem should not be viewed as an influx of bad questions to be stemmed by additional requirements on the user before they ask the question. Instead, consider how to train this user to become a contributing community member, partly by showing them what a better version of their question would be. In many cases, this might be redirecting them to a different forum, but in some cases it might be better to edit, answer and include links to related and more appropriate questions. Perhaps a generalized notion of duplicate would apply, so that future questioners might be given an answer before they submit their question.

I think requiring registration will not be effective for dealing with (my version of) the stated problem. I think it is better to reframe the problem and use it as an opportunity for building community.

Gerhard "Has More Community Building Suggestions" Paseman, 2017.05.06.

$\endgroup$

You must log in to answer this question.

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