25

I would like to start a discussion on how to deal with posts where the OP has written up a clear question, but s/he has also clearly not done his/her homework.

There have been several cases recently. A couple of random examples:

All of these take a simple search in the doc centre to find the solution, yet most of them receive very detailed replies that are essentially repeating the docs.

However, these posts usually do get answers. There's always a nice and helpful soul who will write up one, often in great detail and repeating some parts of the documentation. Since the questions are answered, this encourages the asker to post more of these. I am worried that this is going to be harmful to the community in the long term and might lead to burnout in some of the regulars (without them realizing the danger).

Do you think this is a problem? What can we do to ameliorate this and encourage people to do their homework before asking?

We cannot (and shouldn't) stop people from answering, but I worry that too may of these will both dilute the site and might induce a tedium in some regular answerers that might make them go to the other opposite or simply leave. Burnout is a real danger---it has happened to me in the past, and I stopped visiting the forum in question for months because of it.

I would especially like to hear from those who have been there when other beta sites have started, and have some actual experience.

Somewhat related, though not the same: Help Vampires: A Spotter’s Guide.

  • Neil Butterworth is the perfect example of burnout, which he did twice. – rcollyer Feb 29 '12 at 14:59
  • 1
    It also needs to be said that Neil... had issues. – rm -rf Feb 29 '12 at 15:32
  • @R.M yes, very true. – rcollyer Feb 29 '12 at 15:38
  • 1
    You might as well add this question to the list. I voiced my objections somewhat there.Did you folks think it was appropriate? – Sjoerd C. de Vries Feb 29 '12 at 16:33
  • 2
    Also related: meta.stackexchange.com/questions/56817/… – Sjoerd C. de Vries Feb 29 '12 at 16:48
  • 1
    +1 for "Help Vampires." lol – Mr.Wizard Feb 29 '12 at 17:04
  • @Sjoerd Actually some of these questions don't seem like badly written or hit and run questions---it's just that the poster might get used to getting answers to everything and anything they ask, no matter how trivial, which encourages them to ask even when they could easily solve the problem themselves. They just don't realize that something is wrong and writing all these questions and answers is not very productive (it takes quite a bit of time for both the asker and the replier to write up those posts, – Szabolcs Feb 29 '12 at 17:09
  • @Sjoerd but ultimately neither the question not the answer serves people better than the docs would have. There are exceptions too, for example I loved MrW's answer to the DeleteDuplicates question---it's certainly not a waste of space, but it isn't even the one that got accepted) – Szabolcs Feb 29 '12 at 17:10
  • I think "-1 for RTFM question" comment and a close vote are a good start – David Mar 1 '12 at 5:56
  • 4
    I was going to ask what RTFM meant but instead I read the manual (googled it) :) – Mike Honeychurch Jan 8 '13 at 21:45
30
  • Horrible low-quality questions with unreadable "leetspeak" and incomprehensible illogic should be closed as unclear what you're asking.

  • Duplicate questions should be closed quickly so as to avoid replication of efforts.

  • Rude gimme the codez NOW!!!1!11 questions should be downvoted, closed, and otherwise ignored.

If these rules are followed then I think that questions are really what the community makes of them. From your own list of examples, the delete-duplicates question gave me an opportunity to post this which I was glad for and over a dozen people appreciated.

If the only answer to a simple question is a quote form the documentation (which I have done myself on more than one occasion) I think this shows a lack of imagination. If the question is really so drab and clear-cut that no other answer is appropriate edit the question into something better, and answer that. If upgrading the question would result in it being a duplicate, then perhaps it should be closed as one anyway.

Of course there are going to be exceptions to this where the volume of low-quality questions cannot be handled gracefully. This then becomes the moderators' problem to keep the site running smoothly and there is another forum for that discussion.


In the two and a half years since penning this answer I believe I have observed a shift in community reaction to be more likely and faster to close simple questions. I believe this is partly due to the introduction of the "simple mistake ... or else it is easily found in the documentation" close reason. I have matched this apparent norm with my own behavior.

This does at least to a degree contravene the policy I described above however it should not nullify it. Rather it is my hope that the quick closes will help prevent wasted effort on questions that are trivial.

The principles above may still be applied to closed questions.

  • Users are encouraged to transform "simple mistake / easily found" questions into more interesting ones whenever possible, again so long as a duplicate is not created.

  • At any time if a user has a nontrivial answer of general interest to give to a closed question he should flag the question for moderator attention and state this, requesting a reopening.

15

At Szabolcs' request I thought I could weigh in here: I pretty much agree with Mr. Wizard. Answering RTFM questions is not necessarily going to lead to widespread burnout, as long as

  1. the questions being answered are those where it's clear that the asker has put some thought into formulating the question
  2. it is possible to post a useful answer that does something more than just quoting the documentation.

True help vampire questions are obviously bad. They're basically targeted spam, in the sense that they are posted in large numbers with little attention to quality. If someone puts in the time to ask a well-formatted question, they're not a help vampire, they're trying to ask a legitimate (in their eyes) question, and they can probably take a hint that they should search the documentation more carefully next time. I think it's important to realize that it's a lot easier to RTFM when you know what you're looking for - so even if a question looks like it could be answered by a simple documentation search, that may not necessarily be the case to the person asking it. A lot of people don't know how to use search engines effectively. Perhaps they didn't know what to search for, or they were overwhelmed with results and didn't know which one to look at to get the answer. In these cases, it's useful to describe the process of searching the documentation in the answer so the asker sees how to be self-reliant in the future.

The other thing that can be done, although it's a matter of community preference whether you want to do this or not, is establish a convention of downvoting RTFM questions. Downvoting is appropriate for pretty much any sort of behavior you want to discourage. The most prominent such behavior is, of course, being wrong (when answering), but you can also use it to discourage asking questions that you think should have been handled by other resources. Naturally, it is best to leave a comment explaining the reason for the downvote, although there is no obligation to do so. (Of course this paragraph also applies only to questions which show some effort on the part of the asker; actual help vampire questions should of course be downvoted, closed, and their askers suspended once the signs of abuse are clear.)

  • This more or less coincides with my own feelings. I would add, though, that although downvoting probably is appropriate, enough people still need to do it for it to be useful as an expression of the community's views. One or two downvotes might be seen as coming from a minority of less tolerant users, while ten would unambiguously signal that such questions are not welcome. So in that sense, you're absolutely right that there needs to be a firm convention on this if it's to work at all. – Oleksandr R. Mar 1 '12 at 22:03
14

Many of you who have posted comments and answers to this question are what I consider Mathematica experts, no wait, gods. I have to agree with you that the Mathematica documentation is close to complete. I will even add that help is built to provide streamlined access to volumes of information it contains, that is, if you know what you are looking for.

As a recent newbie, okay somedays I still feel like a newbie, I have a different point of view of the Mathematica documentation. The documentation is sadly lacking. This is especially so when a newbie has some prior experience with other programing languages and they know what they want to do, but are not familiar with Mathematica.

For example, lets say a newbie wants to remove Nulls values from a list. The newbie just needs to find that magic function(s) that will do it. So the newbie fires up Mathematica help, types: "remove Nulls from list", and MMA help returns 6 pages of results. BTW not one result is DeleteCases. The most promising result is the first one, "Adding, Removing, and Modifying List Elements". The newbie clicks that result, looks at the page, and sees nothing about removing Nulls from the list. There is Delete, but it uses indexes to delete values from the list and not the values themselves. So now the newbie now needs to find the indexes for Null values. So back to MMA help, the newbie types in "find indexes of Null values". MMA help returns six pages of results, the first page is not too promising, and no mention of Select in any of the results.

The newbie gives up on MMA help and tries Google, "Mathematica remove Nulls from a list". The number one link is the same page that MMA help returned. Actually, this is a bad example as there are links to stack overflow, and some other sites on the first few pages. Initially when I goggled questions about how to do things in Mathematica, the majority of the searches usually returned links to various versions of the same Mathematica documentation at wolfram.com, at least for the first few pages of results. It was not until I started putting -wolfram.com into my searches did I see more useful results.

So the newbie finds stack overflow or mathematica.se, great, and posts his/her's first question. The answer to the question is obvious, RTFM.

What do you do?

Slam the door in their face and make a great first impression.

-or-

Answer their question and encourage an undesired behavior.

I am not sure if this is technically feasible within the se framework or not, but what if ...

The question is open to the masses. Someone (with a certain reputation level) tags the question as a newbie RTFM question. Once tagged as a newbie RTFM question, the question is removed from the general public's visibility. A comment is automatically posted with a link to a FAQ on how to use MMA help, a link to a FAQ on how to use mathematica.se site, and a challenge to the author for he/she to answer their question.

If the author answers their own question within the initial period of time, they earn a badge, some reputation, and most importantly some experience on how to effectively search for answers to their questions.

If the initial period of time passes and the author doesn't answer their own question, the question's visibility is open for a second period of time to the other newbies (users with some reputation less than some threshold amount) to link a preexisting question/answer that answers the question.

If a newbie makes a link to question/answer that answers the question within the second period of time, that newbie earns a badge, some reputation, and most importantly some experience on how to search for questions that have already been asked.

If the second period of time passes and the question remains unanswered, the newbie RTFM tag is automatically removed and the question is once again visible to the masses to answer. The newbie RTFM tag cannot be reassigned to a question once it has been removed.

  • wait... if your argument is that Google search doesn't help, then why would you suggest hiding an RTFM question from the general public's view? Shouldn't you be doing the opposite instead? I think this is an overall very complicated solution (involving changes to the SE engine that are not possible) to a rather simple question. – rm -rf Mar 7 '12 at 11:18
  • @R.M I don't disagree that it is a very pie in the sky solution, and unlucky to be implemented because of the changes required. However if the question was important enough to be asked in the first place and to address what seems are two legitimate concerns people have raised (burnout of experienced users and disenfranchising newbies), maybe outside the box thinking is needed. Granted, what was proposed is way out of the box, maybe on the floor next to it, but maybe there is something in there that can be used as a catalyst to address the concerns. – mmorris Mar 7 '12 at 14:02
  • @R.M As far as hiding the question from the general public goes, the general public I was referring to was the general users of mathematica.se. I was only suggesting (but did not type it) to hide the question prior to the question being answered or the second time period elapsing, to elevate noise. Eventually the question (and answer) would be completely visible. The argument is not that the Google search does not help, it does, but only after applying -wolfram.com to the search does one see a broader range of answers than what is available in the Mathematica documentation. – mmorris Mar 7 '12 at 14:17
  • 7
    These are very good points. We just have no idea what is it like for someone new to Mathematica, to even formulate the question right. And Mathematica is getting bigger and more complex. For me, it is enough to recall my attempts to code some stuff for my Master's thesis in 1997, with M3.0 - boy, was that a nightmare. So, I think we should not ignore this - these problems do exist. Just a simple observation in support of this: why are there so few "middle-class" Mathematica users - those who are not yet experts but no longer beginners? I attribute that partly to the inadequacy of the docs ... – Leonid Shifrin Mar 8 '12 at 1:02
  • 4
    ... to serve as a learning device for most people, who don't have the luxury to spend excessive time with Mathematica. And the problem is, IMO, that the docs are optimized for referencing huge Mathematica functionality, not for teaching. The appropriate literature to teach the language in the modern way is sorely missing (I made some attempt to remedy that with my book, but it is by far not enough), and I believe that the only way to conquer the M language is by thorough understanding of its core. So, I agree that we should treat beginner-level RTFM questions with care and patience. But,.. – Leonid Shifrin Mar 8 '12 at 1:09
  • 4
    ... your specific suggested solution might not work in practice. Anything which involves enforcing multi-step procedures is hard to implement without a police force. For this to work, it should be in the engine of the site, I think. The practical point I take home is that these situations with RTFM should be considered on the case by case basis, I don't believe there is a universally applicable solution. – Leonid Shifrin Mar 8 '12 at 1:11
  • 2
    @Leonid Maybe there should be a question about what's the best resource to learn Mathematica (there's this). The problem is that the only people who can give a really good answer are those who learned it recently, because they have practical experience. I learned using this, and I'm quite fond of it, but today it is awfully outdated. – Szabolcs Mar 8 '12 at 8:15
  • @Szabolcs I think this has been asked many times, and IMO there isn't a fully satisfactory answer. I also think that we should view it as a community problem: we don't yet have effective knowledge transfer mechanisms, that would reach beginner or intermediate -level audience. My approach would be (and I've been thinking about it for a while) to fight Mathematica complexity by using its own power to create new types of learning media (organize the material differently), which would both make learning from others easier and emphasize self-learning. No easy task though. We also lack ... – Leonid Shifrin Mar 8 '12 at 13:05
  • @Szabolcs some tools, both developer-level and used for automation in transferring and sharing the stuff. I have a number of tools under development, but it is a bit early to discuss that at this point. But, I can at least say that very soon, I will release one tool which has been requested for quite some time (particularly by you), and I think will be quite useful to all of us - Mathematica code formatter written in Mathematica. As to the source you linked to - looks very useful, and I was completely unaware of it, so thanks for the ref. I would not worry so much about stuff being outdated. – Leonid Shifrin Mar 8 '12 at 13:19
5

I'm not sure whether this has already been implemented on the site (perhaps as part of low-quality posts), but I'm thinking that we should have a specific flag for RTFM questions.

The difference would be that flagging as RTFM would require a higher minimum reputation (say at least 500/1k). Once a question is flagged by two or more users as RTFM, no-one except the user who asked the question would be able to post answers to the question for a cooling-off period (of say two days). Comments would still be open so that other users could give hints to the user who asked the question.

If the original user is able to post an answer to his own question within the two days that receives sufficient up-votes, the question will be unlocked to allow other users to contribute answers. Otherwise, the question would appear in the review stack for other users to decide whether it is indeed RTFM or should be re-opened.

4

I would suggest we employ a threshold procedure on RTFM questions where the first person to ask a particular question gets a bye. But, upon the second asking, we rework one of the questions into a canonical question, merge them, and tag it with . There is precedence for just such a tag on SO c++-faq. Mind you, on SO the high reputation members are deliberately writing up questions to go in the FAQ which I don't think we should do here.

  • 5
    I don't mind answering any question once, especially if the question itself is reasonably clear. I think many of us have asked questions at some point that could have been answered with a bit (more) of time spent searching. I do think we should avoid voting up such questions, though. (Not to say we should down-vote mercilessly, but something in the range of -2 to 0 or maybe +1 seems reasonable to me.) – Brett Champion Feb 29 '12 at 16:48
  • @Brett you bring up an interesting point. But, what do you make of this question where the current docs aren't immediately clear. It's searchable for the solution, but the example differs from the method described up top in the docs. – rcollyer Feb 29 '12 at 16:56
  • Please expound on the last sentence of your post. – Mr.Wizard Feb 29 '12 at 17:07
  • @Mr.Wizard I'm not entirely sure why I feel that way. I'll come back to it later, and see if I can clarify the "whys" behind my thoughts. – rcollyer Feb 29 '12 at 17:11
  • "LogCount" was a documented syntax in V7, before it was made more general. The documentation for Histogram and friends has a few areas that could be improved. – Brett Champion Feb 29 '12 at 17:11
  • @Brett I'm aware that it was documented in v7, I was just curious as to how it fit in with the "a bit more time spent searching" mentality when the places we would normally have a user search aren't clear. So, I guess my question was, how does that change things on such low level questions? – rcollyer Feb 29 '12 at 17:16
  • 4
    The less obvious the documentation is, the more slack I'd cut the user. But it also depends on what the user tells us. His response to High Performance Mark indicates that there may be some other issue, but without knowing what he's tried, it's hard to say. Writing good questions takes work -- too little information is no good, but people don't generally want to wade through huge blocks of text either. – Brett Champion Feb 29 '12 at 17:27
  • @BrettChampion, agreed on all points. I just wished to clarify. – rcollyer Feb 29 '12 at 17:28
  • 1
    I worry that in practice we would end up with a FAQ that is less than useful through being made up of mostly "if you want to do this, use that function" Q&As, simply because of the many thousands of functions Mathematica has. I'm totally agreed on the point that if the documentation is unclear, the question can be treated as genuine. But the downvote button tooltip does include "question shows no research effort", so I wonder if, in fact, we should downvote mercilessly? I haven't downvoted at all so far because I've wanted to be ... – Oleksandr R. Mar 1 '12 at 21:46
  • 1
    ... constructive, but it's very clear that no-one likes these questions, and that there is a general view that allowing them might be harmful to the community. Since all it takes for this problem to go away is a concensus to downvote (and then close when the question has been downvoted enough times, unless someone has managed to provide a genuinely useful, non-textbook answer), perhaps that is the way to go after all? – Oleksandr R. Mar 1 '12 at 21:46
4

I agree with Brett that many of us have asked questions that could have been answered by a simple search. Some of those answers may have gotten surprising answers, though, so even that isn't that bad. But generally lack of visible effort annoys me to no end. I know I will be gone if the majority of the questions will be of that type.

Perhaps we should allow any user three RTFM questions, leaving warnings in comments and suspend the user if s/he keeps on showing the same behavior.

  • I have asked such questions myself. I understand that learning a new system can be overwhelming and one might end up writing up a question which will seem trivial in hindsight. Also, it happened to me that I searched a lot, but couldn't find a solution. But when writing up the question, I somehow ended up choosing a title which would have give the solution right away if I type it into Google. But my point is: I have a feeling these questions are encouraged by providing very detailed answers, and often not pointing out what is wrong with them, so some askers might ... – Szabolcs Feb 29 '12 at 17:03
  • 1
    ... end up asking more and more of these, simply because they're not pushed to at least try to solve problems on their own. I think most of the time just a little push would help. But instead they get many very detailed answers about something that's already better explained in the docs anyway. I am worried that the people who write these answers might get burnt out, lose motivation, leave the site eventually. It's funny how rep can motivate people btw, and how SE exploits the quirks in our psychology. – Szabolcs Feb 29 '12 at 17:06
  • 8
    I do not support suspending a user over such questions. Vote down and/or close the questions as appropriate. I believe the system already has measures in place to prevent new questions from one with a very low question quality score. – Mr.Wizard Feb 29 '12 at 17:09
  • I agree with @Mr.Wizard on the point of suspending. Mathematica.SE is a very friendly site, mathematica was one of the friendliest tags on SO, and I am hoping it will stay that way (I really really don't want to end up like the c++-tag or ##C++ chatroom on freenode where beginners usually receive very bad treatment ...) – Szabolcs Feb 29 '12 at 17:30
  • 2
    I know I might sound somewhat inconsistent. The reason I wrote this post was not because I have a strong opinion on how to deal with this. I just wanted to ask the opinions of others, and especially wanted to know if it has any bad effects in the long term to proceed as we are doing now (i.e. if a question is not a hit and run, then answer in great detail even if it is obvious how to find the solution). Most programming sites are not nearly as friendly to this question, but on the other hand I am really glad that Mathematica.SE is not like those. – Szabolcs Feb 29 '12 at 17:34
  • 1
    @Mr.Wizard I believe such auto-bans are in place only on SO/SU/SF and Programmers. – rm -rf Feb 29 '12 at 18:31
  • @Mr.Wizard for this to work, a little bit of an increase of the current level of downvotes would be necessary. Of the four questions listed by Szabolcs only one received a (single) downvote. – Sjoerd C. de Vries Feb 29 '12 at 18:41
  • Sjoerd, that's true. Also if as R.M says this is not in effect here it once again becomes a moderator's problem. – Mr.Wizard Feb 29 '12 at 18:47
  • 1
    In light of the fact that apparently the auto-question-block will not take effect on this site, I am voting for your answer because in the end, a suspension would be the only way I know of to keep a stream of bad questions off the site, if a user was doggedly posting them despite warnings. – Mr.Wizard Feb 29 '12 at 18:50
  • @R.M: I believe that the repeated-low-quality-auto-ban has been applied on math.SE, so I think it's system-wide, but I haven't looked for an authoritative answer on the subject. – Isaac Mar 22 '12 at 22:18

You must log in to answer this question.

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