15
$\begingroup$

My question Libraries for Siamese networks put on hold as off-topic because, as gung claims, "questions asking for software, libraries, or code are off topic here".

There is no sentence "questions asking for software, libraries, or code are off topic here" in https://stats.stackexchange.com/help/on-topic .

If my question is off-topic, the following question: R libraries for deep learning also should be off-topic. Both ask for libraries which do a specific task.

$\endgroup$
  • 2
    $\begingroup$ +1 You make a good point. The R libraries thread is popular, too. $\endgroup$ – whuber Feb 17 '17 at 21:11
  • 1
    $\begingroup$ The R libraries thread should be closed as off topic. It is not a question about neural networks or machine learning. We have had a (largely) consistent policy of closing pure software questions for some years now, but in the site's initial years, the policy was unclear & inconsistently applied. $\endgroup$ – gung Feb 17 '17 at 21:20
  • $\begingroup$ Question "what library does X" is not a "pure software question". It would be off-topic at stackoverflow.com, for example. $\endgroup$ – user31264 Feb 17 '17 at 21:38
  • 10
    $\begingroup$ Just because something's off topic at SO doesn't necessarily make it not pure software, but in any case gung's assessment of it's off-topicness here would seem to be the broadly accepted position, whether one categorizes it as a software question or not. The on-topic help attempts to describe the practice of what is on topic, but not being explicitly listed as off topic there doesn't itself make something on-topic. $\endgroup$ – Glen_b Feb 18 '17 at 0:06
  • $\begingroup$ I don't know how much is visible to whom, but let's underline that being put on hold was the result here of 5 votes to close from users with high enough reputation. (I didn't see this post at all, but I would have supported that vote on the grounds summarised by others.) $\endgroup$ – Nick Cox Feb 18 '17 at 8:48
  • $\begingroup$ @NickCox - "I didn't see the post at all" - The question was "Which libraries/implementations for Siamese networks do exist? I am most interested about R, C, and C++ implementations." "let's underline that being put on hold was the result here of 5 votes to close from users with high enough reputation." - The problem is, there is no vote against putting on hold. If 5 want to put it on hold, and 4 others are against it, it will be put on hold. $\endgroup$ – user31264 Feb 18 '17 at 11:14
  • 1
    $\begingroup$ The key point now is that a vote to put on hold in no way rules out people editing their question to try to make it (more clearly) on-topic. Your specific complaint here is that you think the decision was wrong, but if so you need people to agree with you and sufficient votes to reopen. A general complaint about the voting system belongs on Meta SE. $\endgroup$ – Nick Cox Feb 18 '17 at 11:26
  • 2
    $\begingroup$ I think I am correct in saying that if five people with sufficient rep now vote for it to be re-opened it will be. $\endgroup$ – mdewey Feb 18 '17 at 12:14
  • 1
    $\begingroup$ In the close vote review queue, people have the option to vote to leave open. If four people had voted to leave open (or if even only two had), the question would not have been closed & would have exited the close vote review queue. In this case the vote was 5 to 0. $\endgroup$ – gung Feb 18 '17 at 13:21
  • 2
    $\begingroup$ Closing a popular thread that has stood for a long time (over a year in this case) is not the best action. We have a mechanism specifically to handle such a case: "locking" a thread for "historical significance." The effect is practically the same but the explanation is more informative. I have therefore reopened and locked the R libraries thread. $\endgroup$ – whuber Feb 18 '17 at 14:33
  • 3
    $\begingroup$ @w.huber, not only that, but you've changed your name. I can't even pronounce it the same way now. $\endgroup$ – gung Feb 18 '17 at 20:50
  • 2
    $\begingroup$ Most likely nowhere on the SE system, @Maddenker. There have been a couple of attempted proposals to make an R SE site, but they went nowhere. This sort of question can always be asked on the r-help listserv, of course. $\endgroup$ – gung Feb 21 '17 at 2:54
  • 2
    $\begingroup$ Personally I have a softer spot for "what libraries are available that deal with problems of type X" than for "how do I code/debug this particular thing". But currently consensus is against library questions being on-topic (and I would vote to close, in line with this). I do think this consensus might be revisited at some point. I think there's an argument that library-request questions are very similar to reference-request questions: "what resources are available that deal with problems of type X" is on-topic for written resources, online lectures etc but not for software libraries! $\endgroup$ – Silverfish Feb 21 '17 at 12:13
  • 3
    $\begingroup$ To be really cheeky, someone could try asking "What written resources are available to help me understand how to implement X in practice? I speak R." And the answers could be links to the docs and vignettes of the relevant R packages (and sometimes their accompanying book or journal article). Not quite sure if you'd get away with that, but this analogy is precisely why I'm more sympathetic to library-request questions than other software-specific ones. $\endgroup$ – Silverfish Feb 21 '17 at 12:20
  • 2
    $\begingroup$ @Silverfish: I tend to agree with you. The similarity to reference requests is greatest for someone who wants to study a new area of Statistics: the choice of software isn't trivial, & a recommendation would often be in tandem with that of a book - Diggle & Ribeiro, Model-based Geostatistics & geoR; Harrell, Regression Modelling Strategies & rms; &c. $\endgroup$ – Scortchi Feb 22 '17 at 15:51

You must log in to answer this question.

Browse other questions tagged .