6

Last week, I mentioned:

This private beta needs to demonstrate that a dedicated site opens the door for questions that can't survive on our flagship site.

In order to evaluate the site on those terms, I categorized each and every question base on which site I think it should be asked on:

Site              Count Percent
----              ----- -------
Emacs                63  41.18%
SO                   51  33.33%
TeX                  16  10.46%
Unix                 12   7.84%
Apple                 3   1.96%
Ubuntu                5   3.27%
Super User            2   1.31%
Network Engineering   1   0.65%

Total               153 100.00%

If you want to compare notes, take a look at the spreadsheet I used. Note that I did not try to guess whether the questions would be closed, downvoted, left unanswered, etc. I erred on the side of assuming that Stack Overflow would take questions about using Emacs as an IDE as well as Emacs Lisp questions. I also spot checked 28 of the questions I categorized as SO, and found 18 had been answered there. (While some were the same question phrased differently, others getting at the same underlying issue. It's a judgment call.)

The upshot is that I see 63 questions on this site that would have no home on another site on the network. Reading the answers, the quality of content ranks up there with the best on the network. I personally have learned a lot about Emacs already just reading through other peoples' questions. Everything looks good so far. Keep up the good work.

However, we aren't quite yet ready to move this site into public beta. We'd like another week to see a bigger sample of questions that belong on Emacs.SE and this site alone. Now is the time to ask your Emacs questions, especially if they are unrelated to using Emacs as an IDE or a shell.

12
  • 2
    I have no idea what makes you choose Emacs and site-other-than-Emacs.
    – Gilles 'SO- stop being evil'
    Commented Sep 29, 2014 at 21:39
  • @Gilles: Do you mean you don't understand why some of the Emacs questions were not assigned to other sites? Or do you disagree with my assessments? In any case, I tried to categorize based on audience. I have some confidence that I got the SO questions right since I found many of them already asked and answered on Stack Overflow.
    – Jon Ericson Staff
    Commented Sep 29, 2014 at 21:46
  • 2
    I think that for any question on this site, I could say “this would work on SO” or “this would work on SU” (and in the latter case some would work better on U&L, Apple or TeX), with some overlap. I could also say for any question that it is best suited to an Emacs specialist site. I don't see any difference between the questions you classify as being best asked here and the ones you classify as being best asked on SO/SU/….
    – Gilles 'SO- stop being evil'
    Commented Sep 29, 2014 at 22:25
  • 1
    +1. Nice analysis. I'm impressed that you looked at each question so closely.
    – Drew
    Commented Sep 30, 2014 at 5:19
  • "63 questions on this site that would have no home on another site" Why is that? Do you mean that they would be removed as OT if posed on, say, SO? Which questions are those 63? Could you maybe call them out in your spreadsheet? To me, this category of does-not-belong-elsewhere should be key to the question of whether this new site is needed or a good idea. It should not be the only consideration, but a key one (IMO).
    – Drew
    Commented Sep 30, 2014 at 5:24
  • @Drew They're the questions listed as "Emacs" in the spreadsheet.
    – Malabarba
    Commented Sep 30, 2014 at 11:35
  • 2
    "should", or "could"? I'm not aiming at the big guys (SO and SU) here, but rather the smaller sites: a question about AUCTeX surely could be asked on tex.sx, but does that mean it should be asked there? It seems to me that these questions are on-topic on tex.sx purely because there is (atm) no better place to ask them; and maybe this "on-topicness" would be reevaluated should the emacs site go out of beta. (I see SO and SU as exceptions, because their scope is much broader: usually, if a question can be asked there, then it should.)
    – T. Verron
    Commented Sep 30, 2014 at 13:52
  • OK, but "Why is that? Do you mean that they would be removed as OT if posed on, say, SO or SuperUser?"
    – Drew
    Commented Sep 30, 2014 at 15:16
  • 1
    @T. Verron: That's a philosophical question. ;-) But the answer, I'm afraid, is very pragmatic: questions should be asked on the sites where such questions are asked and answered.
    – Jon Ericson Staff
    Commented Sep 30, 2014 at 16:18
  • 5
    This question appears to be off-topic because we are already in public beta.
    – nicael
    Commented Oct 7, 2014 at 18:15
  • 1
    When are we going to choose mods?
    – nicael
    Commented Oct 7, 2014 at 18:41
  • 1
    @nicael: The Community Team will be appointing moderators in the next few weeks. Please see: Moderator Pro Tempore
    – Jon Ericson Staff
    Commented Oct 7, 2014 at 18:52

5 Answers 5

Reset to default
9

First of all, I agree that we're not ready for public beta: right now the quality and complexity of the questions and answers is rising, and I think it's totally worth waiting some more.

Second, I'm a pure "answerer". I don't intend to ask questions in the foreseeable future, because this workflow doesn't work as well as IRC for me. So my point of view is strongly biased.

Now, I would like to challenge a bit the evaluation criteria. To me, it is essential to incorporate factors like:

  • the fact that questions actually get answered
  • how quickly they are answered
  • the quality of the answers

The fact that there is potentially another home for the questions doesn't mean that they'd be treated equally there (and that goes both ways).

And I'm not saying it's easy to incorporate those factors. Actually I think it's probably awfully difficult. But it's too important to be ignored in a serious evaluation.

Let me take 2 examples of questions that have been "answered" in SO:

  1. Disable ido-use-filename-at-point under specific cases
  2. How to achieve code folding effects in Emacs?

The first one is a low-quality answer that's way too specialized to be generally useful (and wasn't accepted).

The second one (scored at 94 and accepted) is explaining how the feature is unneeded. I have no idea how that could satisfy anybody looking for a real answer (even a "that's not possible" would be more acceptable). And seriously, code folding is not needed? That's rather short-sighted.

In both cases, I strongly think we can do better (and actually did) here.

I don't intend to do a full analysis, but I strongly suspect there is a pattern of sub-optimal answers that I think we have a chance of avoiding here. The reason is simply that the density of people who deeply care about Emacs, and want to provide the right answer is bound to be higher here (as you said, Emacs is a way of life, not a tool :))

Then there are the unanswered questions. I was surprised at first to see that Accessing remote tags file with Emacs was asked and not answered, despite it being fairly straightforward. And then I remembered my experience of the last 2 weeks.

Since I committed to the private beta, I resumed a bit my activities on SO, on the Emacs tag. And that was painful. I cannot really quantify it, but if feels so noisy, so unfocused, that I really wouldn't want to go there on a regular basis to answer questions. As soon as the beta started, I happily gave up, and don't intend to go back.

Maybe there are more like me, who feel much more comfortable browsing randomly new questions or unanswered ones, without any additional filter (therefore without relying on any imperfect classification), because they feel like there's always a good chance they'll be able to answer. If that's the case, that means emacs.se brings value that nothing else can. Which would be good news.

And maybe it doesn't matter at all, maybe the numbers are in favor of SO, and maybe the fact that answers are not necessarily true to "the Emacs way" is of relatively little importance for those who ask. I just don't think so :)

16
  • "The density of people who deeply care about Emacs, and want to provide the right answer is bound to be higher here". Why is the density important to the purported higher quality you think one should find here? Surely what counts is the absolute number of people who care deeply about Emacs, not their size in relation to other, non-Emacsers. More and better Emacs eyes is what counts. Just because the proportion of Emacs users will be higher here does not mean that there will be more & better ones here. S.O. is big: lots of Emacs-question visitors, users or not. Lots of potential users.
    – Drew
    Commented Sep 30, 2014 at 5:35
  • 2
    @Drew the density matter as well. Although the absolute number of people who care would usually determine the likelihood of getting good answers, the total number of people willing to answer (without caring as much) determines the likelihood that the first answer will be a great one. Ideally, the best answer would be voted up, but in reality (in a world of people who don't care much) the first decent answer gets voted and accepted. On top of that there's a also second order effect, where the people who care are demotivated from giving great answers because the decent one will always be on top.
    – Malabarba
    Commented Sep 30, 2014 at 7:25
  • 1
    I am a living proof. I had asked this question stackoverflow.com/questions/19870133/… in Nov 2013 on SO and I did not get a workable solution. I just lived with that nagging issue for these many months. Finally it is resolved with the help of emacs SE community (thanks Sigma!) when I ask the same question ( emacs.stackexchange.com/questions/107/… ) here.
    – Kaushal Modi
    Commented Sep 30, 2014 at 12:51
  • 2
    I second @kaushalmodi. This question is one I had asked on SO months ago and got a single negative response. Emacs.SE got me a solution within the day.
    – Malabarba
    Commented Sep 30, 2014 at 13:10
  • @Drew I would agree if everybody was looking at everything. In practice, a question that receives a couple of answers (and in particular an accepted one) is much less likely to get others. That's why I believe the density counts. I also believe that it's unlikely that hardcore Emacs fans already active in SO or elsewhere will come here too (having our own kingdom is too tempting ;)), so that the absolute number should not decrease significantly.
    – Sigma
    Commented Sep 30, 2014 at 14:27
  • Anyone interested in answering (or asking, for that matter) Emacs questions can simply filter on the [emacs] tag (or the [elisp] tag, or both) - on any site (SO, SuperUser,...) or any combination of sites, including all SE sites. In effect, that means a virtual Emacs site. The problem, as I see it, is that some people interested in Emacs have not learned about SE tags and filtering. Is that reason enough to create a separate site? Maybe, but doubtful.
    – Drew
    Commented Sep 30, 2014 at 15:21
  • Thank you for the feedback. I do appreciate your impressions comparing the SO [emacs] tag to this site. It's important to understand that the private beta is really a special time. It's likely that questions will start going unanswered and the rate of new questions will slow during public beta. So if we can make SO a better place to ask about Emacs, it'll help a lot more people (probably) than splitting the topic.
    – Jon Ericson Staff
    Commented Sep 30, 2014 at 15:34
  • @Drew if that was true, any SE site could be implemented as a tag query on a central hub. And I agree that would be mathematically equivalent, I just think you underestimate the social aspect of that issue. I do know about tags filtering, and I can assure you I'm bored to death on SO/Emacs. I'm definitely not saying it's reason enough to create anything new and disruptive. But there's a trade-off to be made, and the SE staff needs to do the maths.
    – Sigma
    Commented Sep 30, 2014 at 15:39
  • @Drew I'd like to concur with Sigma: The social aspect matters. I go through the [emacs] tag on SO, and I have a filter for [emacs] questions on all sites, but I frequently don't bother to answer or to vote, because I don't feel “at home” on any on these sites.
    – user227
    Commented Sep 30, 2014 at 20:36
  • If everything about Emacs were on one site, that would be one thing. The only downside would be that people on other sites (SO and SuperUser, in particular) would not see it. But everything will not be on one site, AFAICT. To find Emacs questions & answers people will now have one more site to check. And since it will likely have most of the Emacs questions and answers (except for past Q&A, before the new site existed), people expecting to find Emacs stuff in SO or SuperUser will miss most of it. I too appreciate the social/community aspect. But I don't see the site creation as a panacea.
    – Drew
    Commented Sep 30, 2014 at 21:03
  • 1
    @Drew it's definitely not a panacea. But I don't think it's as bad as you describe. To find questions to answer, well, except if you want to find all questions, it's mostly up to you whether you want to go to SO, SU, emacs.se or what have you (I know I wouldn't bother with the first 2). To find answers, I don't believe people use SE search directly: instead they ask google, and I bet they don't care if google redirects them to SO or emacs.se. That leaves people asking questions, which might get confused. But if questions migration work properly, that should solve itself "automatically"
    – Sigma
    Commented Sep 30, 2014 at 21:55
  • @Sigma: SO, on the Emacs tag...was painful. I cannot really quantify it, but if feels so noisy, so unfocused, that I really wouldn't want to go there on a regular basis to answer questions. How so? Here are the most recent questions from the last couple of days of SO questions tagged emacs:
    – Drew
    Commented Sep 30, 2014 at 22:28
  • Emacs dired slow on cygwin, Path with a folder named tilde(~) can't be recongized by load-file in emacs, emacs24 SMTP NTLM authentication, Emacs C-Mode Indentation in Windows Frequently Stops Working, For emacs snippet, how to expand another snippet inside one snippet, error: package.el not yet initialized, Error using Google two factor auth with mu4e & Gmail, How to get rlwrap to work inside Emacs' shell / eshell?, popwin customisation for small frames, Error updating emacs packages: Failed to download 'gnu' archive, Centering display equations with org-latex-preview?,
    – Drew
    Commented Sep 30, 2014 at 22:32
  • How do I get Emacs projectile-mode to ignore most of the files that show up in a virtualenv Cython project?, How do I make named arguments in ELisp?, How to disable the highlighting of faces in Emacs using web-mode.el, Emacs org-mode: increment equation numbers with latex preview, Emacs: contextual margin settings or filling text to narrow range of columns?, Convenient way to start a custom OCaml toplevel in Emacs with tuareg, Running racket with emacs?, How can I copy-paste between emacs and other applications on mac? [migrated], org babel tangle file without main function,
    – Drew
    Commented Sep 30, 2014 at 22:33
  • 2
    I don't intend to ask questions in the foreseeable future, because this workflow doesn't work as well as IRC for me. So my point of view is strongly biased. This might have been mentioned already, but I just wanted to say: if you really want to help this site out (really as in even more) post your emacs questions here and then go ask in IRC. When you get an answer come back an answer your question, or better yet, encourage the IRC user who answered your question to come post it! Not only will this add more useful questions to the site, it could help bring in more users.
    – Seth
    Commented Oct 7, 2014 at 21:43
8

This may be a terrible place for it, but I will jump in with my two cents about a benefit that Emacs.SE can provide, but has not been mentioned yet: Moderation.

Emacs questions on any other SE site are almost guaranteed to be moderated by non-emacs users. This includes not only site moderators, but the review queue as well. For example, I made (what I considered to be) a helpful edit to a SO post that has helped me in the past. My init.el recently broke, and I had a link to this SO post as being the source of the (now broken) code.

I was able to do some investigation and find a solution to the problem. I decided that I would help others out who might come back to that post, like I did, when their init files broke. I submitted a concise edit that fixed the problem, and I provided a link to the GitHub issue where the problem was discussed. However, much to my dismay, my edits were quickly rejected as "not correct."

I visited the profiles of the users who rejected my edits and found that they knew nothing about emacs. At least, their activity on SO did not show any activity on emacs/elisp related tags. I realize that this can happen on any SE site just by dumb luck, but I was amazed how quickly my edit was shot down as "not correct" despite the fact that I provided a link to an "official" answer from the package developer.

This sort of poor moderation in the past has led me to lose heart in participating in small SO "minority" communities. I was very active on the Matlab tag for a while on SO, but eventually got tired of being moderated by people who knew nothing about Matlab.

If Emacs.SE does not survive, and you are looking for ways to make the Emacs community feel more at home on SO, then please make it so that Emacs questions can only be moderated by somebody who has some rep points in the Emacs-related tags.

5

If you don't mind, I'd like to kindly criticize the TeX related choices which have been made.

The following questions have been filed under tex, but they're not about LaTeX at all. You can unarguably bump the "emacs" count by 3.

  • How to view and edit pdf files in emacs?
  • Why is my function not identifying the end of the region correctly?
  • New line in Title of an org-mode exported html document.

These two are also plainly and clearly not LaTeX questions. Despite the title,they only mention an example use-case where LaTeX is involved. In fact, I could have included them in the list above, but I'm trying to do this very cleanly.

  • Emacs function to convert an arbitrary ORG PROPERTY into an arbitrary string
  • Overriding fill-paragraph in LaTeX mode

These two involve LaTeX. But the relationship is so marginal they run a decent risk of being considered off-topic at tex.se. I'd find it very hard to say they "have a home there".

  • How can I preview TeX formulas inline in arbitrary modes?
  • Exporting beamer with org-mode

The following three are probably on-topic at tex.se, but the elisp involved is advanced enough they would very likely go unanswered (unless one happened to run into one of the auctex developers). Should we call that "having a home"?

  • controlling display of latex equations in org-mode
  • Aligning multi-lined equations and tables in latex-mode
  • See LaTeX code and math preview at the same time with preview-latex

The following can be regarded to have a home on either sites.

  • How to call pdflatex under TeX-PDF-mode?
  • Speed up TeX-insert-macro on \input
  • Predictive mode, Auctex not playing nicely
  • How can I hide/display LaTeX section just like org-mode does with headlines?
  • Do I use preview-latex correctly?
  • TOC of tables for org-mode long tables

It seems that any question with so much as a brief mention to tex was filed under that category, but most of them (yes, most) do not have a home on that site. The emacs count can be bumped up by 5 without a second thought, and even 10 if we're being sensible.

To clarify, I'm not saying there are 5-10 which simply should be in emacs, I'm saying these questions have no place in tex.se.

I noticed the focus of this was more heavily on stack overflow and programming questions. But it's important to remember LaTeX represents a good chunk of emacs usage.

Hopefully I didn't come off as angry. Just trying to help.

2
  • 1
    I do admit I spent far less time on LaTeX questions than on programming ones. And you correctly guessed that if I saw the user was editing a *.tex file or using AUCTeX, I was likely to sort it that direction with little extra thought. But I would point out that some of the questions you identified as not belonging on TeX - LaTeX would likely have been sorted into the SO bucket instead. It's more helpful to show that Emacs has a broader audience than just programmers. (By the way, I really appreciated reading your posts here.)
    – Jon Ericson Staff
    Commented Sep 30, 2014 at 0:39
  • This answer would be even more useful if links to the relevant questions were added.
    – Faheem Mitha
    Commented Feb 3, 2015 at 10:14
2

Thanks for taking the time to quantify this.

StackExchange sites work because they attract experts. I think that many 'IDE' or 'TeX' questions regarding Emacs are often answered most effectively by people with a good understanding of Emacs.

For example, in this TeX question on Emacs.SE, the user was able to solve their problem as a result of help with Emacs debugging. You may not have as many users with elisp debugging knowledge on TeX.SE.

This is not a criticism of the excellent TeX.SE site, just observing there's an intersection problem here.

1

I don't know whether or not we should move to a Public Beta yet so I won't comment on that.

Like Sigma I am mostly/solely an "answerer". Up until now I'd mostly restricted myself to questions on SO and SU, with an occasional glance at . I've found most of my answers through Google in the past (which did lead to SO fairly often).

I would like to see more of the questions end up on emacs.se since I feel that would be a better place for them. When I was keeping up with questions on a daily basis there would be occurences of org-specific questions being sent to tex.se because they mentioned latex or pdf, yet were specifically about org-export-to-latex export. Since I had no rep on tex.se I could not comment or push for them to be moved back. Questions also end up split betwen SO and SU with the majority on SO, mostly because there is a stronger org-mode community there to answer the questions.

Drawing questions regarding ,, or any other larger package to emacs.se will both increase traffic on the site, and ensure that questions about emacs-packages are answered in an Emacs context, not looked at by their other tags (tex,git,statistics) where it may provide an answer, but not provide an actual solution.

In the org-export-to-latex question, the answers mostly focused on how to modify the .tex file to add the required information, when the solution would have been to modify org #+HEADER: arguments or #+EXPORT: options.

Like Malabarba I have a few disagreements on the classifications:

Those are all (from what I can remember when looking at them) regarding emacs-specific keybindings/shell processes/tramp access. They may be making reference to shell/term/TRAMP but within an Emacs framework.

Sending them to Unix would not improve the answers, sending them to SO would probably have a few sent to SU (Configuration of Eshell / Overriding Keybindings). Having them in emacs.se ensures an emacs-centric view to them as opposed to where you might get someone with more experience with Shell/Term but less with emacs.

Specifically for Configuration of eshell. Running programs from directories in PATH env variable, there is

  • A Linux/Unix answer : Add the directory to $PATH in ~/.profile.
  • A Windows answer : Add the directory to %PATH% either for the user or the system
  • An Emacs answer : Add the directory to the exec-path list in Emacs.

Assuming you keep certain tools in your home directory that are cross-platform and that you only ever use from within Emacs, the third option is most portable since you only need to ensure that ~ (%HOME%) is properly mapped on Windows.

Another question: How does emacs manage its packages? could go to Ubuntu because it deals with that specific package manager + package.el, but it is better served as an emacs.se question since it can be ellaborated on to explain just how package.el works with regards to it's repositories, EmacsWiki .el files and third-party packages to manage packages (cask and el-get for retrieval, use-package for configuration/loading).

1
  • Thank you for the feedback. Your analysis of the eshell question is particularly helpful. I did waffle on which site to assign those questions to (they all ended up being categorized as Unix). The tricky bit here is that asking on other sites will likely get you useful answers, but not Emacs answers.
    – Jon Ericson Staff
    Commented Sep 30, 2014 at 15:47

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