# The return of the homework question

It has become a ritual: three times a semester (around 2nd week, midterms and finals) there is a flow of simple, homework-like questions (mostly in the field of automata theory; US education system is to blame).

The ceremony goes like this:

1. some new user asks "Are the following languages are regular for: (1) $L={a}$; (2) $L={b}$; (3) $L={a and b}$. Sometime they will also mention "This is not homework!" (oh, so you're posting your take-home midterm question in SE. just great).
2. Then someone will add a comment like "what did you try" ; "where are you stuck" ; "what exactly can't you understand" ; "yes" ; "please use Latex"
3. In the meantime, either someone posts a short answer with partial hints, or some 2-lines formal proof.
4. WAIT 10; GOTO 1;

To me it seems this was not what this site is for, and all of these questions are duplicates of each other. I am a big supporter of answering (also) simple question, but some of those are ridiculously simple. Moreover, it looks like many users don't even try to search the site for similar question, that solves theirs as well (maybe because searching latex doesn't work well; or maybe there are small differences that render the search useless). Not searching for existing question, and posting too many similar trivial questions is my main concern.

We have a homework policy, but it feels we need some policy to deal with this non-constructive flow of questions.

What do you think we should do with this flow of homework questions?

• Yea, I was frustrated too by the same user posting five (?) questions yesterday without demonstrating any understanding. If in doubt, look at the chosen tags: the questions that are a problem are often tagged with any word the came up in the same course, however unrelated. – Raphael Feb 17 '13 at 8:46
• Did we have a reference question for deciding regularity, just like this one for recurrences? If we did, couldn't we just direct the user to that question, or ask what in the new question is not covered already? – Juho Feb 17 '13 at 18:52
• @Juho: We have this one, and myriads of "give me [automaton|grammar|regexp]" questions. So yea, we can do that. It's important that regular users cast these votes, though. – Raphael Feb 17 '13 at 18:54
• @juho, something like that is what I had in mind (but extended in a away that will help more people). I'll process my thoughts and try to post a suggestion for a new policy as an answer. – Ran G. Feb 17 '13 at 18:58
• Would it make sense to have a meta post that would both collect all the reference questions under one post, and furthermore promote these reference questions as something people should direct new users to when they ask a "common" question? Right now, it is hard to find and remember what kind of reference questions we even have! – Juho Feb 17 '13 at 19:03
• @Juho that's a good idea. – Ran G. Feb 17 '13 at 19:09
• @Juho Excellent idea, I'm on it. I have most of them as favorites and/or posted them myself. – Raphael Feb 18 '13 at 8:51
• @Juho Done. – Raphael Feb 18 '13 at 9:09
• @Raphael Nice! I posted something on the chat regarding another possible reference question, hopefully you got it (I have never used the chat before) :) – Juho Feb 18 '13 at 21:47
• We can also have a faq tag so collect the reference questions, it is a meta-tag but is there another way of distinguishing frequently asked questions? Maybe we should look into how the issue of repeated similar question is dealt with on SO and MSE. Or make a future-request to do something about questions that get many duplicates. – Kaveh Feb 23 '13 at 7:36
• @Kaveh What purpose would that tag serve that the meta question does not fulfill? – Raphael Feb 23 '13 at 13:12
• @Raphael, the idea was to make them more visible to the new users so they don't ask already answered questions. – Kaveh Feb 24 '13 at 5:00
• @Kaveh: New users won't find either; we have to point them out to them. So we don't need a meta tag, the list does all we want. – Raphael Feb 24 '13 at 15:44

I agree with all you say: these questions are crap and the askers (often) lazy; if anyone thinks some independent effort is too much to ask, I suggest they read this. However, our homework policy mandates us to endure and deal with them, mostly because we can not reliably tell lazy homework-dumper from inexperienced self-learner.

There is a simple solution:

1. Downvote,
2. post one of the standard comments and

The last item is vital -- too many of our users post quick answers (often as uninspired as the questions), thus reinforcing the unwanted behaviour. If the question presents an interesting puzzle that deserves an answer (good exercises to that!), hold back your answer a week or two. It's immediacy the asker desires most. If you think you can give a helpful hint, do that instead of posting a complete answer (but include the full solution later, maybe behind a spoiler tag).

Of course, if there is a (very) similar post and/or reference question, vote to close mercilessly as long as there is no own thought. See also here.

• The solution you suggest, is indeed what happens today. I agree with this solution as a concept, the problem is that it doesn't work. Those questions do get answers, and the fact that they are being closed after 2-3 days is meaningless. – Ran G. Feb 17 '13 at 18:49
• Oh, and BTW, I oppose putting "what did you try?" comment as a standard. There are too many of those comments also on very valid questions. This is not welcoming! Even worse, I recall a "why do you ask this?" comment. IMHO, nobody is required to justify his interest in a specific question. One asks 'cause one wonna know. – Ran G. Feb 17 '13 at 18:55
• @RanG. That means that our core users have to be faster. Vote to close, comment on the question and answers. Most (active) users will eventually realise they hurt the site, and adapt their behaviour. (I am against downvoting answers on such questions just because they are there, btw.) Again, it is important that regular users do this. – Raphael Feb 17 '13 at 18:55
• @RanG.: Quite the opposite, a clear motivation and/or history of the problem is one of the things every SE question should have. On the sciency sites, we are more lenient about this because many questions have inherent value of their own. A question a la "What is a CFG for this language?" is usually not, though, and should come with some meat. (I don't feel obligated to welcome homework dumpers. So far, even when encountered in a friendly and helpful way, they have never come back. It's the professionals and self-learners that return.) – Raphael Feb 17 '13 at 18:56
• @RanG. Also, I'm not sure it does not work. The answers they get are mostly hint material, and the dumpers rarely return. I think we may have to endure their bursts, but as long as they learn "cs.SE is not a good place to get homework solutions", mission accomplished. – Raphael Feb 17 '13 at 19:00
• Raphael, I don't agree. People encounter things that they don't understand and are very welcome to ask it here. If they give extra information (and motivation) their question will be better and will probably get better answers. However, not giving motivation doesn't damage the validity of their question. A valid question would be "I'm trying to analyze an algorithm and got stuck with the Master theorem" both when this is HW algorithm, or research-level algorithm. Both are welcome! – Ran G. Feb 17 '13 at 19:01
• @Raphael I agree. Also, often the answer can be a lot better and more useful when the asker explains what he has tried, what didn't work, and just provides more detail in general. I think it also shows respect and the question is more likely to get a good answer. – Juho Feb 17 '13 at 19:10
• @RanG. You are correct: the question itself is valid, and that was the motivation for our current homework policy. The problem is, as Juho phrase quite well, that homework dumps are not questions, but delegation of work. The spirit is important. Similarly, questions are downvoted/closed for exceptionally poor form (if the question is incomprehensible). Of course, every user may judge the spirit themselves, in the spirit of our policy. – Raphael Feb 18 '13 at 8:54
• Maybe these duplicate questions should be merged into one place? – Kaveh Feb 23 '13 at 7:41
• @Kaveh: Which are you referring to? – Raphael Feb 23 '13 at 13:12
• @Raphael, my suggestion is that if these questions are answered before getting closed as duplicate then it might be reasonable to also merge them. – Kaveh Feb 24 '13 at 4:59
• @Kaveh Not at all; preemptive, localised solutions to the "homework" question don't make much sense on the reference question. – Raphael Feb 24 '13 at 15:43
• Unfortunately that's not the official SO policy. Instead, the official one seems to be that trivial and low/no-effort questions are ok as long as they are clearly formulated because those generate a lot of views from similarly inclined individuals. And ad-views/clicks are what keeps this site alive. Also see what some high-reps on math.SE answer. – Fizz Feb 15 '15 at 21:00
• @RespawnedFluff Lucky for us, every SE site is entitled to create its own policy (within some bounds). Afaik, SE staff is well aware of our policy. (Note that the level of the question is not the issue here, it's the effort. Even Stack Overflow requires you to put in some effort yourself.) – Raphael Feb 16 '15 at 7:46
• @Raphael: Indeed, physics.SE turns out to have a much tougher stance/policy on homework questions. – Fizz Feb 18 '15 at 23:24

Stick to the homework policy, close duplicate questions, and move along; nothing to see here.

Realistically, students at US universities represent one of this site's largest potential user bases. That doesn't mean we should tolerate bad behavior, but it might mean we should err on the side of letting the community (most of whom won't read this question) moderate itself when it comes to these kinds of questions.

We do want people to ask questions (even simple ones), but we don't want duplicates of the same question (or with multiple copies of close variants).

A possible way to handle it is the following:

For any subject that we recognize having multiple small-variant questions (like: "Is this $L$ regular"; or "What is the solution of this recursion using the Master Theorem"), we

1. create a thread with exemplary question.
(this will be the "reference question" @juho suggested in his comment).
• the question should be formatted in a way that will be very easy to search ("How to prove the following L is regular"), very simple and straight forward.
• however the question will not contain any specific details
(i.e., "L" will not be defined within the question)
2. we post several CW answers. Each answer will be one specific detailed instance for the "reference" questions. E.g., each answer-post will contain one $L$ and the solution for that specific $L$.
3. new posts with questions that have a "reference questions" will be handled in the following way:
1. Immediately closed.
2. a comment will refer to the relevant "reference question"
3. If the specific "L" doesn't already appear in the reference question, a new CW answer will be added to the reference question, with that specific L
4. any answers on the original post will be deleted and their content will be moved to the answer-post in the reference question.
5. Eventually, the closed question should be deleted (Why not just let it be? because such questions are not searchable anyways, and there's no point in keeping them).
4. In meta, there will be one post that aggregates the "reference questions" in one place, so we will be able to find them fast and refer to them (thanks @Juho for this clever idea!)

This way, there is a single well-organized place that gives extensive information on the question (with all its variants). New users will be encouraged to search before they ask, and the community will be encourage to refer to exemplary quesiton/answer rather than repeating the same answer in multiple places.

• A good idea. I think the reference question should not contain say all possible L's someone might ask about. Instead, it should have general methods and techniques and somehow well chosen examples that try to capture the essence of the different possible L's. If the answer is too long or has too many answers, people will wanna read it less. Emphasis should be given to explaining common and expected misunderstandings. Textbooks are already full of examples. – Juho Feb 17 '13 at 21:52
• I agree with Juho's sentiment. Collecting huge numbers of languages is not meaningful. The idea would be that people go to the reference question, digest the material and can then ask a good question, if they still have problems. – Raphael Feb 18 '13 at 8:55
• @Raphael Although there are many examples out there, exactly the one that the OP needs (or similar in its nature) cannot be found, and the OP is helpless. This is why I prefer to have several examples for different $L$'s, and if the OP, seeing that thread, still doesn't know how to apply the method on his $L$, than that new $L$ should be added as well. Win-Win. – Ran G. Feb 23 '13 at 7:43

I personally don't really mind homework questions. What I do mind are people who register a new account, ask a poorly worded and badly formatted question with that account, yet seem to expect an answer instantly (perhaps because it is an exam question).

A homework question asked by a professional or a self-learner can often spark interesting answers, expanding on the standard textbook answer. The difference is that the OP in such cases is around to address comments and to accept a good answer, instead of leaving the answers orphaned.

It might therefore be worth asking SE for a stronger filter when new users post a question.

The current filter is geared to SO and doesn't seem effective at all to discourage the kinds of poor questions seen here. For instance, very short questions by new users could be blocked until approved through the review process for First Posts. Right now review seems to be a means of closing the barn door once the horse has bolted and a bunch of other new users have started asking similar poor questions in imitation of the first. The process of posting a question could also be made to require more attention. Right now some people are clearly ignoring the helpful suggestions of similar or identical questions that the system presents, which I think is quite rude. It might be worth having an explicit disclaimer that says something like "This question appears to be a duplicate or very similar to these other questions already asked: (...) Your question may be interpreted as rude by the regular users, and may be put on hold. Do you still want to submit it to a moderator?" Then any First Posts reviewer can allow it.

• The tools you propose would be nice, if they could be implemented in a reliable fashion. Sadly, site metas are hardly monitored for feature requests; I suggest you search Meta Stack Exchange and post your suggestion(s) there if needed. – Raphael Jun 22 '13 at 15:59
• Similar ideas have been thought about meta.stackexchange.com/questions/170586/… though the "robo-reviewer" problem seems to make this less effective on SO. – András Salamon Jun 22 '13 at 16:25
• In that case, we have little chance until Oversight not only acknowledges that not all SE sites are hellpits like SO (among others) but also that this should influence the platform. – Raphael Jun 22 '13 at 16:32
• By the way, I agree that one does not have to have a problem with (lazy) homework questions as an individual (I know I don't). However, they can become a problem for the site if they drown everything else, since those people who don't like them have little way to evade them and thus have a poor (perceived) signal/noise ratio. (Wow, am I arguing for a homework tag just now? O.o) – Raphael Jun 23 '13 at 15:58
• @Raphael: Ranking hellpits is probably hardly useful, but Superuser is even more so (than SO) in my opinion... – Fizz Feb 18 '15 at 23:54
• As for SO being a hellpit for CS questions, there's no doubt in my mind about that. Recent example: stackoverflow.com/questions/28593290/… – Fizz Feb 19 '15 at 4:03