# Suggestion to improve the quality of questions.

Recently, a contentious discussion has been taken place in Removed Question vs Reputation around the opening of the question Determine if $\frac{k-1}{k}+ \frac{1}{k(k+1)}= \frac{k}{k+1}$ holds.

I have noticed that an important issue of the discussion is about how some kind of questions should be edited. Sometimes, the OP doesn't present any effort or ideas to solve his question. Actually, very often he just says something like: " I have no idea how to prove this".

In spite of his lack of effort, some times the question is interesting and even receives some answers. However, it takes a risk of losing educational content (and/or reputation points) if the question is deleted.

It is true that the question can be edited and all people can earn with that, but how to edit questions that don't have any effort presented by the OP? Is it fair taking the OP's place and editing something wrong just for presenting a "fake effort"?

Now, I present a possible solution for mitigating this contentious problem:

When a new question is posed, I suggest that it can be available only for users with +X reputation points just for evaluation (not for answering yet) and also for the OP. If the question is OK, then it is opened (available) for all users to give an answer (as it is at present). If not, the OP should improve his question, otherwise the question can be deleted even before to be available for all people.

The evaluation can be just a voting procedure, where the evaluators vote either to the deletion or the opening of the question. This evaluation could be available for two days (for instance) and some improve suggestions can be reported to the OP during this time. After this period, if the question is not deleted, then it is automatically available for all users.

I don't know what are the implications if this is implemented in the site. What do you think?

• One shouldn't emphasize the "loss of rep" but, rather, the loss of educational content if answers are deleted (or inhibited). The purpose of this site is not to gain rep but, rather, to share mathematical knowledge. Try to never lose sight of that (despite the SE gamification brainwashing). – Bill Dubuque Jan 22 '15 at 15:39
• The issue is that often educational content (at least as far as archiving it is concerned) is non-existant. – quid Jan 22 '15 at 16:12
• @quid: You have absolutely no way to know that. – Brian M. Scott Jan 24 '15 at 22:03
• @BrianM.Scott I have no way to know what? – quid Jan 24 '15 at 22:22
• @quid: What's the confusion? You made an assertion ('non-existant' [sic]), and I pointed out that you could not possibly have the knowledge required to justify it. The meaning of my statement seems perfectly clear. (So does its accuracy, but that's a separate issue.) – Brian M. Scott Jan 25 '15 at 3:00
• @BrianM.Scott I've no interest in a discussion about the meaning of "know" or "non-existent" and how absolute such a statement is. So: I feel competent to evaluate the merits of part of the contributions to this site, and for not few my opinion is that the site would be better without them. I don't know if I "know" this; it is my firm opinion based on my knowledge of the subject and the site. It is not impossible that something I consider as worthless will in some circumstance be of value. Also, it is not impossible your comment now saved me from being hit by a bus in 5 years. – quid Jan 25 '15 at 3:33
• While I agree with Bill's comment, I do think the gamification has a significant impact (I would be lying if I said that I am completely immune). Perhaps users with +X reputation not only get to vet the questions, but to assign a reward value. – copper.hat Feb 2 '15 at 6:41

Something in this spirit is being tested on SO — it's called «question triage». If the experiment is successful, maybe we can ask to turn it on Math.SE too.

• This could work very well. In particular if all the questions posted by users with less than, say 100 rep, are automatically put into this queue (and are blocked from answering and deletion while there). – Jyrki Lahtonen Jan 22 '15 at 16:47
• I'm sure that if it's working well on SO, they'll implement it at least on the busier SE sites (and math.SE would be next on that list). Though I guess it wouldn't hurt to inquire about that... hmmmm.... – user642796 Jan 22 '15 at 20:13
• A significant part of triage appears to be an automatic analysis of the post's content to give it a "quality score". I'm worried this would not work (or would need significant modification) to effectively handle MathJax. – 6005 Jan 27 '15 at 20:28
• @Goos This is, indeed, an important point — but I'm optimistic – Grigory M Jan 27 '15 at 20:53

I have pondered an idea for some time now on a way to get a new user to write a better question. The triage idea is okay but do new users even know their questions are being evaluated? If they don't, how will this method get the user to ask a better question because isn't that the goal?

Would it be possible to have new users see a question that is written and asked correctly when they click ask question? By written and asked correctly, I simply mean a question that shows the problem statement and what the OP has done in trying to solve it so the example questions don't need to be highest voted question just a high quality one. I would propose that the example question is not closeable for 20secs or so this way the new user just can't click past it. A simple prompt afterwords could say does your question illustrate what you have attempted? The example questions could even be related to the tags they select so the example will be related to their level.

If a new user continually post bad questions on their some number of first post, a list of said users could be sent to moderators where they can increase the time limit of viewing the examples questions and not let them leave this new user prompt until they start to take it seriously and show improvement or even short term suspensions.

• A very useful suggestion! – André Nicolas Jan 22 '15 at 20:01
• @AndréNicolas I think so to but it is not according to the downvoter. – dustin Jan 22 '15 at 20:13
• Note that right now when users post too many bad questions, they are temporarily blocked from asking questions. And this happens automatically. Of course, the system does require users to provide data that indicates that these submissions are poor quality (downvotes and close votes, primarily). – user642796 Jan 22 '15 at 20:15
• @ArthurFischer By then the user is in a question ban, wouldn't it be nice to try and reach them and show them what is proper before the ban? – dustin Jan 22 '15 at 20:17
• @dustin: One downvote means one user of meta does not agree. Several downvotes means that several do not agree. But the hundreds of new people would learn something useful, in a more positive way. – André Nicolas Jan 22 '15 at 20:20
• @AndréNicolas that would be the goal. – dustin Jan 22 '15 at 20:22
• +1 because I support every idea of educating newbies about the site norms. My gut feeling is that we get a not insignificant number of bad questions from users, who never come back (or come back once or twice on a throwaway account). But there are many users who would ask better formulated questions, if they were trained a bit first. This would help such users. Also those users might feel more welcome. – Jyrki Lahtonen Jan 22 '15 at 20:43
• Sorry, @dustin: I was completely unclear. I was more responding to the content of your last paragraph. – user642796 Jan 22 '15 at 20:48
• @JyrkiLahtonen I brought this up on meta and I don't think it will gain any traction. It apparently is viewed as too much effort, not worth, etc. – dustin Jan 22 '15 at 21:28
• I sort of see that. The process would need to be automated, so we need help from the developers. Then we should compile a list of model question based on, say the tags picked by the asker. We actually have a surprising number of people who would volunteer to help in putting together such a list (Martin Sleziak comes to mind quickly, but he would not be the only one). The amount of coding is non-trivial though. I tend to think that the scale of similar problems in SO is larger than here, so let's see what they learn from testing Triage. – Jyrki Lahtonen Jan 22 '15 at 21:33
• @JyrkiLahtonen i suggest using triage to find model questions instead since the code is set up. It would just need new search parameters. Send those to a model question vote queue. They fought back against that logical solution. – dustin Jan 22 '15 at 21:38
• @JyrkiLahtonen since this suggestion doesn't seem to be a hit on meta, do you think this is something you guys (as mods) could ask to try just here? Then if it appears to work here, you guys (as mods) could let them know of the success or failure. – dustin Jan 23 '15 at 18:56

If someone feels the need to edit the question to make it look like the OP showed effort, this is an indication of a serious problem where people are closing worthwhile questions simply because they don't want to do other people's homework. The correct solution would be to only close questions based on poor mathematical quality, unclearness, or if the question is a duplicate. If all of the answers to past homework questions are found online, this can actually be beneficial for several reasons:

1. Serious students who don't want to just copy the answer will have verification that they are correct.
2. Professors/instructors may make an effort to create new questions that cannot be found on the internet.
3. Professors/instructors may find better ways to grade students than homework assignments.
• "Context" and "effort" are different things. This is what context means. Notice that it includes many different possibilities. I explained here why context is important. Your reasons 1,2,3 are completely irrelevant to the working of MSE. People object to the website becoming a homework factory that drowns out all other questions, not (only) to students copying answers they asked for online. – Najib Idrissi Jan 24 '15 at 18:20
• @NajibIdrissi I don't disagree that some degree of context is important. But if the OP understands more than the next person with the same problem looking at the question, this will be reflected in the answer and is likely to be detrimental. There may be a disconnect between the context the OP shows and the level of the answer, where it is enough for the OP to understand but not a later reader. The answer should start from the beginning, and a good answer will cover all details rather than just the ones the OP needs. – Matt Samuel Jan 24 '15 at 18:26
• The concept of "all of the answers to past homework questions" is ill-defined. The servers running WebAssign, MyMathLab, WeBWorK, etc, generate new homework problems every day. There are infinitely many functions to integrate or differentiate, algebraic expressions to simplify, and so on. If there is no barrier between getting an assignment from online homework, pasting it into another browser window, and getting a solution there, this site will a real-time log of homework done everywhere. – user147263 Jan 24 '15 at 23:07
• @Fundamental You should review the definition of "ill-defined" and "past." You seem to be claiming it's infinite, which is not ill-defined, and besides that it is not infinite, there have been finitely many homework problems in the past. – Matt Samuel Jan 24 '15 at 23:12
• @Fundamental Here is a better rebuttal, though: this can be solved with aggressive closing of duplicates, where "duplicate" does not necessarily mean exactly the same question. A question that is almost the same with different numbers would qualify. In this vein, we already have some abstract duplicates. I certainly don't think we should do every single WeBWorK problem, since in a class they are all the same with different numbers. One of them would be enough, then if it is asked again direct them to the one that was already done. – Matt Samuel Jan 25 '15 at 0:24
• @Fundamental When it comes to integrals, only a relatively small class of integrands have elementary primitives. While there is a lot of variation within this, the same principle applies. – Matt Samuel Jan 25 '15 at 0:26
• I'm all for closing duplicates, but this is not a solution to the homework flood. Take integration of a rational function, for example. One could try to write a canonical Q&A pair to serve as a duplicate target, covering all the possible cases... but it would be a copy of a chapter from the textbook that's already on the student's desk, so what's the point? – user147263 Jan 25 '15 at 1:56
• @Fundamental Well in the specific case you mention I wouldn't write just one canonical duplicate target. I would cover it in separate questions. I think the major problem in that case would be finding a good way to search the similar questions, which I believe has been brought up on meta before. – Matt Samuel Jan 25 '15 at 2:03
• @Fundamental This also wouldn't be done all at once. If we had a good way to search questions and we allowed generic questions without extra context, there would be no need to create special abstract duplicates, as they would be created as they were asked. – Matt Samuel Jan 25 '15 at 2:05
• @Fundamental With regards to your last comment, I don't think it is that clear that that's how things work. A general procedure will have arbitrary constants instead of specific numbers. You may not remember it, you may not even ever had such issues, but the difference between specific examples and general ones can be huge to the uninitiated. – Git Gud Jan 28 '15 at 23:03