# Handling low-quality questions from low-rep users

Please see the end for my feature request - forcing users with [what would be] negative rep to have subsequent questions approved before they're viewable by the public.

As a fairly regular MATLAB asker/answerer on SO, I notice that there is a disproportionately high number of basic homework-help questions to legitimate MATLAB questions. What I mean is, a lot of questions might get asked that could be answered VERY quickly if one took the time to search (it's typically something immediately available in MATLAB documentation); is only tangentially related to MATLAB; or the asker has clearly made no attempt to solve the problem himself.

Examples:

Highpass Filter In matlab for a Image

I want to apply Highpass filter to a image containing a line in matlab . So anyone can u help me out to designing a highpass filter in matlab and how to extract the frequency components in it

plot smooth CDF using matlab

using some dataset example: data = [1,2,2,3,4]

how can I plot a normalised (0,1) smooth CDF (some call it s-shaped cdf) using matlab?

How can i return a image in the M-file? (Matlab)

My function file starts with `function drawline(point1,point2,color,img)`, at the end, i'm supposed to return a image. How to code the return line. i post codes matlab:color issueThanks.

How to number to alphabets in matlab

The function arithenco needs the input message to be a sequence of positive integers. Hence, I need convert a message into a sequence of numbers message_int, by using the following mapping: ‘A’→1, ‘C’→2, ‘G’→3, ‘T’→4.

Another question I was about to link to just got deleted; the OP asked how to find the points (x,y) that define an ellipse. (Note that this isn't actually a MATLAB question.) It was there until I mentioned to the OP that he/she needs to show that there has been some attempt at answering the question.

In my undergraduate and graduate experience as a non-programmer (my major is aerospace engineering, with a minor in computer engineering), any first language can be a hard one to learn, especially in an engineering class where the primary focus is not how to learn to program, but how to create numerical methods. At least in aerospace and mechanical engineering, MATLAB is introduced in the junior year and not really given a good treatment aside the absolute basics. Furthermore, SO is by design a site for programmers and most programmers don't use MATLAB as their base language; so I understand that there will be a lot of "homework-help" type questions as opposed to "real" questions. But there is no reason you can't attempt a homework and post what you've tried and asked for a helping hand, rather than "give me answer kthxbai"

I also see that a lot of people will just answer the question even though it's obvious the asker has not attempted the problem himself. Maybe it's because it's easy rep (if it ever gets awarded), but this flies in the spirit of answering homework questions on SO. I personally pretty much refuse to answer questions like this, but so long as users feed the fish, so to speak, I feel like MATLAB questions will get progressively worse and worse. It's hard to stay engaged in the community when I have to weed through these types of questions before finding a gem that truly deserves attention and effort.

Is there anything that can be done to curb this issue?

My idea/feature request: If a user has what would be negative rep (i.e., he was a new user and asked a question which was downvoted - I will without hesitation downvote any question in which no effort was made beforehand to solve the problem), any further questions he asks would be flagged for review. If the question is not of sufficient quality or smacks of "plz gimme answer", the moderator should inform the user that the question will not be posted and he needs to improve its quality. This reinforces that SO is not a dumping ground for the homework you can't be bothered to do; if you want to use it for homework, make sure you've attempted it first and don't just ask for an answer, but ask for guidance or tips or "can anyone see why what I'm doing is wrong".

-
I think this one probably meets your criteria as well. Hot off the griddle :) – Tim Post Oct 6 '11 at 11:30
Sometimes you don't understand the problem well enough to search for, or understand, the documentation. It can be hard to tell lazy questions from "I'm out of my depth" questions. That said, I just ignore quite a lot of Python questions like this, where they're not quite bad enough for me to flag and comment. – agf Oct 6 '11 at 16:14
@agf - I do ignore these types of questions more often than not, but others tend to just give the answer immediately without any thought for what seems to be easy +rep. That's part of my concern, anyhow – Dang Khoa Oct 6 '11 at 17:44
@TimPost - :sigh: – Dang Khoa Oct 6 '11 at 17:52
"I feel like MATLAB questions will get progressively worse and worse" -- I think it just hit rock bottom. – Charles Oct 6 '11 at 18:22
@Charles That link is now 10k+ only (I whacked it from orbit) – Tim Post Oct 6 '11 at 19:02

In my experience, we always get a wave of low-quality questions when a new MATLAB-using class starts. Some of the questions are really a failure of putting in a little bit of effort, but for quite a few people, the problem is simply that they don't get it. As it turns out, MATLAB is being taught to people who are non-programmers (unlike e.g. Haskell), and thus, they have no clue at all as to even how to understand the help. Also, it turns out that some people are really not very good at using Google (some of the students I've taught almost made me cry).

I have previously tried to find a solution that would make it easier to close low quality questions as duplicates (rather unsuccessfully).

In the end, one goal of Stack Overflow is to answer people's questions, and if someone wants to spend time doing someone else's homework, why not let them?

In the end, what you may be looking for is to have an option to filter questions from users with low rep. Maybe that can be implemented?

-
It's an intriguing idea, but there are low-rep users I legitimately want to help - they've showed that their question is truly homework and they've attempted the problem themselves. I like my idea for queuing questions from users with [what would be] negative rep and have mods look at questions from users who have already demonstrated their unwillingness to attempt questions themselves. At least you have someone looking at the question and judging if it's someone who doesn't care vs someone who is truly doe-eyed. – Dang Khoa Oct 7 '11 at 0:05
My guess is since Friday is when homework is typically due, we'll see a barrage of these kinds of questions on Thursdays. :/ – Dang Khoa Oct 7 '11 at 0:16

Thoughts from someone who is at least on the "Low Rep user" end of things, and isn't so far removed from what I suspect people consider low quality questions:

This happens, in my experience, whenever you mix two particular things: programming and non-programmers.

When I was a teaching assistant in a class in a program that went from relying on simple, easy, cut-and-paste code where there was very little reliance on understanding what was happening to much more sophisticated techniques in the same language, students became really, really uncomfortable. Even pretty basic programming language concepts - how a `for`-loop actually works, when to output data, etc. proved to be problematic. We even got questions about how to google programming questions when we suggested they use Google. And this was for an exceedingly well documented language.

I think the problem stems from three different things:

1. Unfamiliarity with concepts and terms. It's often hard to even figure out what you're asking for at the very beginning, let alone producing useful Google terms or a properly formulated question for Stack Overflow.
2. Urgency. Often, if you're posting a question here (or searching for it on the Internet), it's because you've run aground on a project. The problem is that the value of searching, finding an answer and figuring it out - even for a basic problem - has less of a clear and immediate payoff if you're not a programmer. This isn't so much a valid concern, but it nevertheless is one when at the end of the day, spending an hour or two learning the language feels like a waste of time.
3. Translation. This was the biggest problem for me when I was starting. I could find a solution I knew was close, but either because my problem was slightly more complex, or I didn't have a good grasp of the language, I couldn't get from A to B. That's profoundly frustrating, and one of the big appeals of Stack Overflow is it's easier to get a direct answer for your particular problem.

It is frustrating. But I think it may be a problem with the nature of the site, and perhaps you can feel a little better about the idea that someday, some of the people asking those questions might convert to active, productive users.

-

I like the idea, but I wonder if it will really scratch your particular itch. Of the three questions I can access, none would likely have been prevented with this scheme. One of the askers has a reputation over 400 and no downvoted questions, another is a first-time question and the third (reputation 14) probably slips through the cracks. It seem likely that an overwhelming portion of homework questions are asked by people who will never return to the site and won't be screened out with this procedure.

Going back to USENET days (when I was a student), I've always wondered what would happen if some of the student's teachers could be notified of their students questions on public programming forums. I wonder if it's possible and practical to trace these students back to their institutes of learning? I wonder if some of the teachers could be recruited to help deal with these types of questions?

-

Are these duplicates? Could they be considered duplicates? Could existing posts be edited to have higher quality and be considered canonical? There's only so many of these questions that can get asked, and if you're vigilant about closing them, hopefully, the behavior will at least slow down.

I use this 3D spectrum (just squint a little) to determine whether a question should be closed:

``````High quality                                        Low Quality
<- It must be a verbatim dupe to be closed
On-topic                                              Off-topic
Close it if at all possible ->
Interesting                                       Uninteresting
``````

Obviously, most questions are in the middle. Use your best judgement and make a case.

For instance, your first question seems interesting and on-topic. MATLAB is often used to create filters, and there are several ways to do it. There should be at least one quality question about "What tools exist in MATLAB to create filters, and how do I use them?" comparing the merits of the `filter()` functions, DSP toolbox wizards, and manually computing the convolution. This question should have a high-quality answer with links to the documentation. A question or questions for the import and export of various file formats could also be good. Then, close low-quality "How do I create a highpass, lowpass, bandpass, bandreject, Butterworth, etc. filter for audio, images, this matrix, this function, etc." as a dupe of the canonical question.

See also the Dr Strangedupe and Are some questions too simple blog posts for more about determining quality.

-
Some may be considered duplicates, like the high-pass filter question. But others, while not "duplicates" in the sense that they've been asked before, certainly request information that is already widely known and freely available without too much though. It's the latter questions that I really get annoyed about. – Dang Khoa Oct 6 '11 at 17:42
@strictlyrude27 The policy of SE though is to strive to be the place to get answers. So even if it's easily Google-able elsewhere, it should also be available on SE. – agf Oct 6 '11 at 17:47
@agf - I agree with you to a point - the point being SE should not be a congregation of links to the appropriate documentation, which is what the latest MATLAB questions are turning into. Take a look at the latest questions with the MATLAB tag, for instance. – Dang Khoa Oct 6 '11 at 17:48
@agf and even if the answers ARE basic, that doesn't mean the asker can't put some serious thought or effort into his question. I'm sick of questions like "how do i find max, min, sum of matrix without using `max()`, `min()` or `sum()`" and no code posted showing what he attempted to do so far. – Dang Khoa Oct 6 '11 at 17:52
@agf - I'm not sure that's entirely true. Did you read the "Are some questions too simple" blog post? If the answer is already easy to find via a search engine, the top results (such as Matlab documentation) are high-quality, and the question isn't interesting or complex, then we don't need to be the place to get answers to that question. – Kevin Vermeer Oct 6 '11 at 19:38
@KevinVermeer I didn't mean to imply there is no lower threshold at all. But to quote my comment on the question "Sometimes you don't understand the problem well enough to search for, or understand, the documentation." The answers to out-of-my-depth questions can be helpful. – agf Oct 6 '11 at 20:03

As with any public resource there will be some less-than-stellar contributions, but that's what flagging and close votes are for. In my experience, low-quality questions do tend to get closed eventually, although sometimes it might take a few days.

If you see a post that is of very low quality, flag it for moderator attention and/or vote to close the question.

-
Right, but it's turning into the vast majority of MATLAB questions, which is simply getting infuriating. I feel like I'm constantly flagging, or editing, or asking for more information, prodding the user to make the question better. – Dang Khoa Oct 6 '11 at 15:44