Is it just me, or is there a major burst of homework questions popping up ? Fall semester automata theory, anyone ? Unfortunately, we still have people answering questions instead of voting to close. Any thoughts on how to close these questions BEFORE people start answering them ? I could bring the AMP hammer down I guess.

Update: I just exchanged some email with a TA for a theory of computation class at Princeton, who confirmed that three questions posted by a single user were part of a take-home midterm given to students there. This is a lot more serious IMO than homework questions. In all cases, the questions were closed, but not before a few answers slipped in, which essentially makes the closing irrelevant.

  • 3
    $\begingroup$ Some said in a comment (I don't know where) that they were referred here from StackOverflow??? Maybe I got that wrong, but if another site is steering people here, that might explain the sudden increase. $\endgroup$ – Aaron Sterling Nov 2 '10 at 20:05
  • $\begingroup$ Most of them (3) were posted by the same user. $\endgroup$ – Kaveh Nov 3 '10 at 19:21
  • 2
    $\begingroup$ @Aaron: I found a message on MathOverflow Meta suggesting cstheory.stackexchange.com as one of the alternatives for non-research-level questions: meta.mathoverflow.net/discussion/698/molevel/…. I added a comment asking not to include us in those alternatives: meta.mathoverflow.net/discussion/698/molevel/…. $\endgroup$ – Tsuyoshi Ito Nov 3 '10 at 19:40
  • $\begingroup$ I liked the format of his comment: $\endgroup$ – Kaveh Nov 4 '10 at 6:13
  • 2
    $\begingroup$ This site is for research-level questions in theoretical computer science, that are likely to have short well-defined answers. "Research-level" means, roughly, questions that might be discussed between two professors, or between graduate students working on Ph.D.'s, but not usually between a professor and the typical undergraduate student. It does not include questions at the level of difficulty of undergraduate homework. You can try MathOverflow (research-level math) or Math.SE (elementary math/theory) or StackOverflow (programming). $\endgroup$ – Kaveh Nov 4 '10 at 6:14
  • $\begingroup$ @Kaveh: I like the parenthesized purpose for each alternative, but I am afraid that some people will not read it. As a result, your comment may lead some people to incorrectly think that MathOverflow is for undergraduate homework questions. That is the last thing I want to do, especially because I asked people on MathOverflow not to suggest that cstheory.stackexchange.com is for homework (I do not want to do the same thing to them “in return”). I am afraid that it is next to impossible to come up with an all-purpose canned response which is not prone to misunderstanding. $\endgroup$ – Tsuyoshi Ito Nov 4 '10 at 13:51
  • $\begingroup$ @Tsuyoshi: how about this: MathOverflow (research-level math), Math.SE (elementary math/theory), StackOverflow (programming) $\endgroup$ – Kaveh Nov 4 '10 at 21:28
  • $\begingroup$ @Kaveh: There are cases where the question is not about TCS but it is about research-level math, and there are other cases where the question is not research-level. My point is that it will not probably be wise to try to come up with a single canned response which works in both cases. $\endgroup$ – Tsuyoshi Ito Nov 4 '10 at 21:34
  • $\begingroup$ Does anyone else see this as indirectly related to allowing anonymity? $\endgroup$ – Maverick Woo Nov 5 '10 at 2:53
  • 1
    $\begingroup$ @Maverick: yes, the students will not post their assignment questions under their real names. But that does not mean that anonymity is bad, there are high reputation users that prefer to remain anonymous. Anyway, there doesn't seem to be a reasonable way to restrict anonymity, it seems to me that the only thing we can do is to be more considerate in answering when we see an elementary questions is posted by a drive-by/new low-reputation user using a nickname, specially when the question is badly written and does not provide any motivation on why the person is interested in the question. $\endgroup$ – Kaveh Nov 5 '10 at 3:46
  • $\begingroup$ @Tsuyoshi: I have posted a comment similar to the following when voting to close a question: "this is not a research-level question and is too elementary for this site (probably a homework exercise)." I think the comment above is more informative and nicer. We can remove MO and use it for questions which are too elementary for this site. $\endgroup$ – Kaveh Nov 5 '10 at 5:24
  • $\begingroup$ @Kaveh: Might I suggest use of "perhaps" instead of "probably"? I think it would cut down on potential arguments, while still making the point. $\endgroup$ – Aaron Sterling Nov 5 '10 at 16:04
  • 1
    $\begingroup$ If the cheater used their own on-campus IP address to post the mid-term questions the campus IT department can probably figure out who the cheater is. They need that the IP-to-name association for forwarding copyright infringement notices after all. $\endgroup$ – Warren Schudy Nov 5 '10 at 23:12
  • $\begingroup$ @Suresh: About the update.. even if you close questions as soon as they appear, someone might answer the question in a comment. $\endgroup$ – Robin Kothari Nov 5 '10 at 23:25
  • $\begingroup$ @Robin, I'm working under the assumption that people providing answers are doing so innocently, i.e they don't think the question is a homework question. Thus, if the question is closed, with associated comments, they won't post an answer. $\endgroup$ – Suresh Venkat Nov 5 '10 at 23:28

It's important to remember two things:

  • Speed is of the essence with such questions: it does no good to close it after 24 hours, if someone posts an answer within 3 hours
  • Closing is reversible: a mistake can be rectified very easily by a moderator.

So I'm proposing that I use my AMP to pre-emptively close questions that smell like a homework, pending OP clarifications. If you see such a question and want to draw my attention, you can always flag it. If it turns out I was wrong, I'll reopen.

The cons (the pros are clear) of this approach are that a poster with a real question might get annoyed. I'm willing to take that risk though. What say you all ?

  • 1
    $\begingroup$ I have no idea what AMP means but moderators closing suspicious questions immediately sounds like a good idea to me. $\endgroup$ – Warren Schudy Nov 5 '10 at 22:32
  • $\begingroup$ @Warren: AMP = Awesome Moderator Powers (see a comment by Suresh on meta.cstheory.stackexchange.com/questions/431/…) $\endgroup$ – Tsuyoshi Ito Nov 5 '10 at 22:37
  • $\begingroup$ :). sorry Warren, I got so used to referring to it that way that I forgot to expand. $\endgroup$ – Suresh Venkat Nov 5 '10 at 22:42
  • $\begingroup$ (1) I think anything confirmed to be homework, midterm, etc., should be deleted so it cannot be searched in the future. (2) I agree that prompt closing of obviously non-research-level questions is the solution, as long as it is accompanied by a diplomatic explanatory note. However, (3) I strongly disagree that this should be a "Suresh thing." That way leads to burnout, frankly -- being a policeman instead of a builder of a research community. It should be an "everybody who has enough rep to close questions immediately" thing, which will hopefully be quite a few "meta people" over time. $\endgroup$ – Aaron Sterling Nov 5 '10 at 22:50
  • $\begingroup$ Also, if we are going to close more aggressively, then in addition to the language Kaveh suggested above, I would suggest a link or a standard comment explaining how a question can be improved, and the process for reopening it. $\endgroup$ – Aaron Sterling Nov 5 '10 at 22:52
  • $\begingroup$ We could also, as a community, decide to downvote answers to HW questions, so that no one feels inclined to HW questions. This will be tricky to implement, and might annoy a lot of people. $\endgroup$ – Robin Kothari Nov 5 '10 at 23:24
  • $\begingroup$ @Aaron, thanks ! Actually I only suggested moderator control because it appears to take some time to close questions. If we have swift question-closing, that's perfectly fine. $\endgroup$ – Suresh Venkat Nov 5 '10 at 23:28
  • $\begingroup$ @Suresh: Is it only you and Ryan who can close with one vote? I thought that above a certain threshold, anyone could do it. $\endgroup$ – Aaron Sterling Nov 6 '10 at 0:04
  • $\begingroup$ you need mod powers. I don't think anyone gets those powers till 10k. $\endgroup$ – Suresh Venkat Nov 6 '10 at 3:18
  • $\begingroup$ Thanks. It only takes 500 rep (at least in beta) to vote to close a question, and there's two pages of users at that threshold. So it seems to me that we are moving toward policy/request that users with >500 rep should aggressively close homeworky questions. (This relates to point 2 of @Lev above, also.) I'm fine with that, certainly, but maybe we could work out some language so things don't seem draconian. Maybe @Kaveh could post his language in an answer or in a separate question as a starting point to work out an official (or at least officially suggested) message. $\endgroup$ – Aaron Sterling Nov 6 '10 at 12:50
  • $\begingroup$ right: we do have enough users to close questions, but some questions get closed really quickly (usually the vague ones) and HW questions don't $\endgroup$ – Suresh Venkat Nov 6 '10 at 15:57
  • $\begingroup$ @Suresh: I believe that AMP is given at 2k during beta, but 10k when the site "graduates" (see blog.stackoverflow.com/2010/10/when-will-my-site-graduate). There are 13 people who could presumably verify this. We could graduate as soon as next week, but presumably they will let us simmer for a little while longer. $\endgroup$ – Rob Simmons Nov 6 '10 at 22:24

Bumping this thread --

I'm sure part of what is going on is the crunch at the end of the semester. Also, there's less pressure to "prove we're experts" because we made it out of beta. But... aren't the way-below-research-level questions dominating all new questions? Also, several new questions look like homework problems, or requests for advice on class projects or what could be class projects.

I answered one "too-basic" question, because I thought it was politely and enthusiastically asked, and I thought I could provide nonstandard information. But maybe that was an error, especially given the direction things are heading.

What do you think? No big deal, ride it out until the new year? Or, downvote and close questions?

  • $\begingroup$ Why are people asking for ideas on class projects this late in the semester? $\endgroup$ – Peter Shor Dec 11 '10 at 16:20
  • 4
    $\begingroup$ I've been biting my tongue because I didn't want to seem too hardcore, but I completely agree with Aaron: there are way too many 'basic' questions, and I also agree that because we're out of beta, we've relented slightly on the pressure to maintain quality, which is a BAD IDEA. I'd advocate getting more hardline on these posts $\endgroup$ – Suresh Venkat Dec 11 '10 at 16:42
  • 1
    $\begingroup$ @Peter, I gave a class project in September, and people asked me their first question about it a week before the deadline. $\endgroup$ – Suresh Venkat Dec 11 '10 at 16:43
  • $\begingroup$ When I give class projects, I always have a deadline much earlier for the topic. I guess I didn't realize how important this was. $\endgroup$ – Peter Shor Dec 11 '10 at 17:11
  • 4
    $\begingroup$ One factor may be closing a question needs more time now than during the beta. At the end of the beta, 66 users had 500 points required to vote to close. Currently only 11 users have 3,000 reputation points required to vote to close. $\endgroup$ – Tsuyoshi Ito Dec 11 '10 at 17:32
  • $\begingroup$ @Tsuyoshi: Indeed. And that may also relate to the problem that we have relatively few questions than other SE sites with similar traffic rate, which may reduce the total upvotes generated. This worths a new thread of discussion. $\endgroup$ – Hsien-Chih Chang 張顯之 Dec 11 '10 at 18:00
  • $\begingroup$ I agree with Tsuyoshi, the number of users who can vote close has dropped considerably. Another change is that we now try to ask the OP for his/her motivation before voting to close (in many cases where it is not very clear that the question is elementary/homework) and wait for the OP to reply, so it takes longer to close them (if at all, becuase the users with close votes don't check them later). An alternative to deleting the questions is: close the question and delete the answers (or flag them), but I haven't thought about it carefully and don't know if it is good idea or not. $\endgroup$ – Kaveh Dec 11 '10 at 20:34
  • $\begingroup$ I've been wielding my AMP more liberally for this reason, when the question looks like it really needs to be closed. So far, no major complaints. It is a pain that we don't have enough users with the right rep: I feel that the SE overlords are being a little bull-headed about this: not every site has the scale of SO, and the numbers need to be tempered accordingly. $\endgroup$ – Suresh Venkat Dec 13 '10 at 8:44

When a homework question is asked, as long as someone is willing to answer it, the questioner gets what he came for, and it provides incentive for him to ask even more homework questions. All it takes is one "defector." This is complicated by the fact that there can be a difference of opinion as to whether something is homework, and again, once one person decides a question is not homework and answers it, the restraint among the remaining members is for naught.

We have a couple options, for example:

  1. We could not to worry about helping with homework, as the ethical violation is mostly on the person asking the question, not the person answering. We would still close non-research questions, but not worry about someone answering homeworks. The problem with that approach (in addition to helping people cheat) is that we might incentivize a flood of questions that will have to be closed.

  2. We could have moderators constantly monitoring the site and aggressively closing/deleting questions they believe to be homework. The downsides of this is that we'd need more moderators and that we'd get lots of false positives. This might discourage new people from asking questions in the first place.

  3. I don't know if this is possible, but perhaps if enough high-rep users decide a question is homework, anyone answering it gets a reputation penalty. This would require a change in site mechanics (probably not possible) and also have the downside of making people weary of answering even legitimate questions. Perhaps this can get implemented by an ethic of downvoting such answers, but I am not sure I like where this would go.

More worrying is that once a professor becomes creative with homework problems, and assigns harder or non-standard questions, we would not be able to recognize them as homework. I've had professors give homework problems that would easily pass as a legitimate questions on this site. There's not much I can think of to deal with this problem, except perhaps have this site become sufficiently well-known, so that profs are aware of this potential "resource" for students.

  • $\begingroup$ You call someone who wants to help a "defector". Also, you assume that low-level questions are inherently bad. Please note that only some people feel that way, others do not. $\endgroup$ – Raphael Nov 8 '10 at 10:48
  • $\begingroup$ @Raphael I put defector in quotes to recognize some irony of using the term. But questions that aren't research level are indeed inherently bad, at least according to our FAQ. Also, while there is indeed some disagreement about the propriety of answering homework questions here, ultimately the site will have to be governed by some rules that a minority may not agree with. $\endgroup$ – Lev Reyzin Nov 8 '10 at 13:13
  • $\begingroup$ That is certainly right. Can you quantify which opinion is in the majority here? I have not seen a vote so far but only individual opinions. If we want this site's scope to be defined by a reputation/founders/whatever elite, then we should clearly state so. In my opinion, as long as people answer low-level questions they seem to be in-scope by design of this software, if not by FAQ. $\endgroup$ – Raphael Nov 8 '10 at 15:40
  • $\begingroup$ We had a site proposal which defined this as a research-level site. We had people commit and join based on this premise. Not all questions that someone decides answer are automatically in scope just because the software allows it. For example, someone may choose to ask cooking questions and someone else may even answer them, but this is no reason to redefine our site. It is why we have moderators -- to make sure the site is used properly, etc. But if you think lots of people think that homework questions are okay, then I challenge you to find a couple > 1000 rep users who agree with you. $\endgroup$ – Lev Reyzin Nov 8 '10 at 16:26

I don't know what the motivations of homework-answerers are, but from my perspective the question closing process must seem bewildering and downright mean to the homework-question-asker (I know, obviously, that this is not how it's intended.) It makes one want to be more helpful than the official closing process.

The message people get when a question gets closed is: "Questions on Theoretical Computer Science are expected to generally relate to Theoretical Computer Science, within the scope defined in the faq." Not only does that not really explain that the question was closed as homework level, but the FAQ link doesn't go to the right place - it goes to the Stack Exchangey FAQ here, which then has a link to the "Offical FAQ" which then has only an indirect reference to "Typical homework problems in textbooks" being too basic. So that's two or three links to get the first oblique reference to the real reason why a question question was closed.

Is there a way to make Tsuyoshi Ito's comment linked to above or Kaveh's paraphrase actually appear on closed homework questions?

  • $\begingroup$ Agreed: I think the general FAQ is very long and focuses on the site mechanics, which seems to be something that only people who have been on the site for a while are really interested in. Since the FAQ should really start with the first questions (is this question appropriate, should I use my real name, should I register an account) it would make sense to try to make "our" FAQ the real one, with the SE one pointed to from ours. $\endgroup$ – András Salamon Nov 4 '10 at 14:29
  • 1
    $\begingroup$ one way is to link to a question/answer on meta that states the official policy on homeworks. Any one voting to close should add a link to such a question/answer. $\endgroup$ – Suresh Venkat Nov 4 '10 at 15:58

Can we somehow force new users to read the portion of our FAQ that describes the scope before posting questions? This would at least prevent people from accidentally posting off-topic questions.

  • $\begingroup$ we'll have to take it up with the SE overlords or meta.SE. It seems that for some of the sites that made it out of beta, they were able to rewrite their FAQ. $\endgroup$ – Suresh Venkat Nov 5 '10 at 20:02
  • $\begingroup$ There's no point in campaigning for new software features on meta.SE until we have consensus on what features we want to ask for. Hence the suggestions here. $\endgroup$ – Warren Schudy Nov 5 '10 at 22:56

This is more of a comment than an answer, but anyway this is meta.

Of course, the asker is ultimately responsible for asking the question. However, the current situation that the questions like homework are likely to be answered before getting closed is unfortunate. It is probably tempting for some students to post homework questions given to them, because it seems to work. We have to close quetions faster.

Because deleting as well as closing is discussed here, I would like to point out one drawback of deleting homework questions: it will hide the evidence of cheating. To be honest, I do not think that we should delete homework questions. I understand that the asker wants to delete the question to hide the evidence of cheating, but we should not try to hide the evidence of cheating for the asker.

  • $\begingroup$ You make the same assumptions as Lev. Those who take offense could just ignore the questions, you know. $\endgroup$ – Raphael Nov 8 '10 at 10:50
  • $\begingroup$ @Raphael: You are missing the point. This discussion is about those who ask the homework questions given to them. I do not want to participate in the website which helps cheating. Ignoring bad questions does not solve the problem. $\endgroup$ – Tsuyoshi Ito Nov 8 '10 at 11:18
  • $\begingroup$ Then we should talk about how to prevent cheating. Scope is not the correct parameter to manipulate; people might cheat on research-level, too. In other words, cheating and level are not correlated in general. As stated elsewhere, we cannot determine who is cheating and who is not and therefore should responsibility stay where it belongs, that is with the asking person. Or would you say Wikipedia helps/promotes cheating? You can answer almost any high school homework assignment using Wikipedia. And we are talking about grown-ups here, mind. $\endgroup$ – Raphael Nov 8 '10 at 15:45
  • 1
    $\begingroup$ @Raphael: Yes, scope and preventing cheating are different. On this page, we are discussing preventing students from using our website for cheating. $\endgroup$ – Tsuyoshi Ito Nov 8 '10 at 16:24

How about adding the following minimal restriction: users with reputation <= 500 cannot answer questions asked by an asker with <= 150 reputation unless a user with reputation >= 500 first approves the question by upvoting it.

This does require software support of course, but if the stackexchange owners aren't willing to make minor changes to the software to suit our needs (or expose an API for customization) perhaps we should look elsewhere for a more responsive host.

  • $\begingroup$ Warren, I'm not trying to stonewall, but we have no control over such rules unless they are globally enforced by the SE system. All we can do is campaign for such things in meta.SE. We need other kinds of solutions that we can implement within the constraints we are under. $\endgroup$ – Suresh Venkat Nov 5 '10 at 22:25
  • 1
    $\begingroup$ This idea is like killing a fly with a cannon, and I think it would be harmful to the site. $\endgroup$ – Karolina Sołtys Nov 5 '10 at 22:33
  • $\begingroup$ Also note that HW questions have been answered by users with reputation over 150, so this is not going to solve the problem. $\endgroup$ – Kaveh Nov 6 '10 at 9:34
  • 1
    $\begingroup$ I'm with Karolina on this. Such a rule would be massive overkill. $\endgroup$ – Joe Fitzsimons Nov 7 '10 at 16:54
  • $\begingroup$ At least it would be a clear cut rule as opposed to a dozen high-rep people judging, making it their job to keep the site clean wrt to their definition of that term. $\endgroup$ – Raphael Nov 8 '10 at 10:52

You must log in to answer this question.

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