13

Emacs—Stack Exchange has really grown in the last 302 days. We've simplified the graduation criteria lately and Emacs.SE is starting to reach that bar. However, I'm still wonderering whether this site serves Emacs users better than Stack Overflow. To that end, I'd like to attempt an experiment in introspection. (N.B, the results of this experiment won't have any practical effect on the schedule of site graduation. Please attempt to give honest feedback.)

Below are listed 10 randomly chosen questions from the last 90 days. (This query shows my method, but the result set is likely to be different when you run it.) Please reply to this question with answers that evaluate these questions and their answers on the following criteria:

  1. Has the question been asked on Stack Overflow? You can search within a tag by adding [emacs] to the search string. If it has been asked, are the answers here more useful and understandable than the ones on Stack Overflow? Does the voting reflect your evaluation? Which question has more views so far?

  2. What could be fixed about the posts on this site? Go ahead and make any changes you think would help and leave your reasoning as an answer.

Please copy the questions into your answer and add your comments. Also, I'd appreciate any overall comments you have after going through this exercise: both about the site and the process of evaluation.


Inline verbatim and code with quotes in Org-mode

How to determine operating system bits (32 vs 64-bit) in elisp

Can emacs close open LaTeX tags automatically?

Evil: Is there a default register that does not get overwritten by later yanks to explicit registers?

Technical considerations for package maintainers to not use the Emacs package manager?

Is escaping backslash in regexp string expressions optional at times?

Automatic package install is not available, package exists

Zip/Map functions ending when shortest list runs out

Instructions on how to work with Evil Mode

Automatically pressing enter and y in emacs when prompted for input

  • 8
    I'm ok with this as long as we conclude that emacs.stackexchange is worth keeping. – Aaron Hall Jul 24 '15 at 23:05
  • It makes me sad to say that I probably won't be able to take the time for this. I'll try to sneak in one or two when I find a free hour, but hopefully there'll be other great users like Verron who can help out with this. – Malabarba Jul 29 '15 at 10:09
  • @phils: Those are really good points. Would you mind reformatting these comments as an answer? Also, you might be interested in: Does it pay to spin off sites? I wrote that around the time this site was launched. – Jon Ericson Aug 6 '15 at 22:59
11

I don't know if it is expected that each answers examines all 10 questions... That sounds like a lot of work, and I would be afraid of making it more of an overview than a detailed analysis. If that was the expectation though, I can delete this answer and let someone more courageous look at all the questions.

I have looked at the two that interested me most when reading the list: Can emacs close open LaTeX tags automatically? and Instructions on how to work with Evil Mode .

Has the question been asked on Stack Overflow?

Apparently, the LaTeX one has never been asked on stackoverflow. I also searched on tex.sx (there are 400 questions tagged there, for 1000 questions tagged and mentioning LaTeX on stackoverflow), and there, a couple questions do mention the trick, but they don't directly answer that question.

The answer to this question is very stackoverflow-style: a short description of the appropriate key-bindings, depending on the mode you use, without much explanation. On the other hand, this question does not call for explanations, so the answer is complete in my opinion. This question could probably have been asked on stackoverflow and recieved a similar answer.

The evil question has a sister on stackoverflow: evil-mode best practice. It is not exactly the same question, but it would appear to call for the same answers.

The question on this site has only one answer, with introductory style. No code snippet, and instead explanations about what is evil and how it integrates with emacs. It links to a configuration example should a reader be interested, but the primary content is the explanation.

The question on stackoverflow has a lot of answers, each giving some configuration to either integrate evil into emacs or get rid of the conflicting emacs bindings. Most answers contain little to no explanation to accompany the snippet. The question is asked by a vim user coming to emacs, is he expected to be able to read emacs-lisp fluently?

What could be fixed about the posts on this site?

In my opinion, this site could benefit from more questions and answers like the second one: questions asking for explanations as much as code. Answering these questions properly takes more time, which means that, asked on a high-traffic site like stackoverflow, they may remain answered longer than necessary.

I couldn't help but notice that the second question and answers had been left unvoted, maybe it means that this kind of questions is bound to remain unpopular here. I surely hope not.

I have read many times that there are two conflicting communities here: guys coming from stackoverflow, used to quickly answer specific questions to solve a specific problem, and guys coming from lower-traffic sites (tex.sx, math.sx, prog.sx...) who like answers to be as explanatory as possible, even if it means that the answer will take longer to write.

Both approaches are compatible, because most questions belong to one or the other category: people with a problem to solve want it solved fast, curious people asking more general questions will prefer to understand what they read.

Specific problem-solving questions are the main flow of questions: they are many, they are easy to ask, they bring traffic and experts. And they could be asked on stackoverflow if this site didn't exist. But in my opinion, the real added value of this site is in its more down-to-earth questions, which are answered in english instead of elisp.

So there are three suggestions I would make to further improve our site:

  • for askers, don't hesitate to ask more questions which are not "I have this problem, please give me an ad-hoc solution" (as long as they are on-topic of course)
  • for answerers, try to recognize it when the question asks for explanations rather than (or together with) a mere solution; answer/edit answers accordingly
  • for readers, don't be afraid of reading 30 lines of text before voting ;)
  • Don't worry about analyzing just some of the questions; the general feedback you provided is helpful. Thank you for the insight into the two communities. – Jon Ericson Jul 23 '15 at 14:43
8

(n.b. converted from comments to an answer, by request)

Vote counts aren't likely to be directly comparable (cross-site) as a measure of quality. In my experience, E.S. generally attracts higher vote counts than S.O. does for Emacs questions. (I think that's a good sign for E.S., mind you -- it suggests there are more people actively looking at the general Emacs activity here than at S.O.)

To put some numbers against this statement: I've written 102 answers on E.S. garnering 495 up-votes (average 4.853). In the same time period I've written 60 S.O. Emacs answers garnering 96 up-votes (average 1.6).

Even when I use my entire history of S.O. Emacs answers over five years (i.e. with much more time for most answers to accumulate views and votes), the average number of votes per answer is still higher at E.S. than at S.O.

I also don't think Views are very comparable between the two sites. The two questions may have existed for vastly different time periods, and so you would need to try to factor that time into the equation (which is probably really hard to do). Search engines will presumably continue to favour the more established of the two.

Furthermore, the users answering questions may well favour the more established Q&A. If a solid answer exists elsewhere on the network, then pointing to that is arguably the best way to respond to duplicate questions which crop up here. And if that pre-existing Q&A (which is already known to search engines) can be usefully improved, doing that is of greater benefit to the end users (who are searching for a single definitive solution) than dividing the knowledge by answering duplicates on other sites.

I'm sure there is useful data to be obtained by comparing the two sites -- I'm just dubious that it's quite so easy to interpret the numbers.

(n.b. Jon also referenced his Does it pay to spin off sites? question in the comments.)

6
  1. Inline verbatim and code with quotes in Org-mode - I asked this same question eons ago, when I still had an SO account. This is the original question for the reference: https://stackoverflow.com/questions/16443386/how-to-escape-double-quote. There's no surprise this question is asked over and over again, since this is an inconvenience with no obvious cause or solution. The answer on Emacs SE also indicates that the question has been asked before. Same goes for SO Emacs tag.
  2. I think that the answer on Emacs SE is very thorough. Rather exemplary of what a good answer should be. Perhaps, I'd like to see more tags on the question: something to make it easier to find in the midst of all other Org-related questions. Possibly, this https://emacs.stackexchange.com/questions/tagged/syntax. This is more of a general commentary. I'd like to see tags in Emacs SE which are not too specific to some Emacs feature or mode, but address a category of problems s.a. perhaps, "interface", "accessibility" etc. I know there is an argument against "meta" or generic tags because, allegedly, readers don't subscribe to them, but having a rough idea of how text search works, I'd assume that such tags would implicitly help in ranking of search results.

  1. Automatically pressing enter and y in emacs when prompted for input apparently, this question has been asked before: https://stackoverflow.com/questions/19194619/how-to-auto-say-yes-when-run-a-command-in-emacs.
  2. In both cases answers aren't good (they try to address the more specific problem rather than treating it in general). I do like the SE Emacs answer better though, since it at least gives a generic procedure for dealing with such problems (although some work would be required to deal with each specific case). An all-encompassing solution is possible, but is tedious to write.

Commentary

I'm biased against StackOverflow model of social interactions, and I had my doubts before joining SE Emacs site. I don't have this kind of statistics, but I imagine it to be common for SE users to subscribe to several SE sites at once (I do too). For me, however, Emacs site is the one, where I spend most of my time. Joining this site, for me, was a combination of making a bad "political" choice with being able to participate in life of the community I care about. Obviously, I decided differently when my doubts were between staying with Emacs community posting to SO and having to put up with site's "policy".

Some of the SE family of sites have a character very distinct from its flagship site. And I hope that Emacs will be different. There are sites, whose community I can barely stand, but I participate because of uniqueness of the resource. These would be Math SE, Theoretical Computer Science SE, Code Review SE - I don't like these for elitism and hunting for karma points. To contrast that, LaTeX SE is a very friendly community. Sometimes I see efforts of introducing behavior patterns to Emacs SE similar to SO, Math SE and the likes, but hopefully these are few and far in between. There probably aren't many people with the same dilemma, so my vouching for separate site rather than blending in with SO aren't that important in general, but since you've asked for an opinion, I thought I'd give one :)

You must log in to answer this question.

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