What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 127 Stack Exchange communities.

Based on your experience in the tags that you're active in, do you think that activities[*] in certain tags more than others need improvement? What do you think are the problems, and how would you suggest we try to improve the situations?

[*] Activities includes how questions are asked, answered, voting pattern, how comments usually go, etc.

share|improve this question
2  
I was curious where this was going, tbh. I think maybe what you're asking for in the question is a little vague/rantish? Can you focus the question a little bit? –  Jon Seigel Jul 5 '10 at 18:52
2  
Excellent edit, voted to reopen. –  devinb Jul 5 '10 at 19:12
    
Do you mean "certain tags ... need improvement" (i.e., tags on questions) or "questions in certain tags ... need improvement"? –  Jon Seigel Jul 5 '10 at 19:16
    
Too wide, in my opinion. There are regular efforts of retagging, for particular tags, in separated questions. It's not really something to be done in only one question, I think. It's not "noise or pointless", though. –  Gnoupi Jul 5 '10 at 19:28
    
@Gnoupi: I think we can use this question as a brief survey of all the "problem" tags, then based on votes we can see which efforts are worth undertaking, and focus those individual efforts in their own meta [discussion]. How's that? –  polygenelubricants Jul 5 '10 at 19:33
    
This is looking a lot better now, although it should probably be CW. –  Jon Seigel Jul 5 '10 at 19:44
1  
@Jon: CW-ed. Now let's get the discussion going. –  polygenelubricants Jul 5 '10 at 20:55
add comment

6 Answers

It looks like there's a lot of room for improvement in [regex]. There are too many questions where people keep on asking "I need regex for this", "I need regex for that", and the highest voted answers are often the ones that just says "here you go" and "there it is". The ones that actually explains HOW and teaches things takes time to write, and thus usually gets buried with little reward, or even worse, downvoted because it's just "too long" of an explanation instead of just getting "straight to the Fish™".

Other "problems" with [regex]:

  • People often fail to mention which flavor they're using
    • Also evidenced by many "which flavor is this?" comments
  • The insistence of using it despite existence of better alternatives (e.g. parsing HTML), and the awkward situation that this puts for answerers
    • Also evidenced by many "regex isn't the right tool for this" comments
    • Sometimes even a best effort answer gets downvoted, perhaps because people think it's "harmful" to provide regex when it's really inappropriate for the situation
    • Sometimes what should've been a "don't do this with regex" comment is given as an "answer", which then acquires numerous votes (because it is an excellent point). This reward propagates the cycle which leads to more and more "don't do this with regex"-type "answers".
  • The duality between the actual regex pattern vs the String literal to represent said pattern (e.g. \w\/ vs "\\w/")

Possible solutions

  • Perhaps whenever someone asks a [regex] question we can trigger a language check prompt (e.g. "you have not identified which flavor this is for, we have identified that this information is usually crucial", etc)
  • Perhaps a [regex] reference "FAQ" is a good idea. It's essentially a CW index to various questions on stackoverflow that covers most of the basics. It can supplement the search functionality, which can be hard to use effectively if you don't know what to look for, but a comprehensive index can cover the most typical scenarios, answers to common mistakes/problems that gets asked repeatedly, etc.
share|improve this answer
3  
I'd also think it's frustrating when the assumption is that a regex is the only solution and they're not willing to consider something else. –  ChrisF Jul 5 '10 at 17:24
    
In a similar vein, I would love to reject any posts to [perl] using code snippets that are obviously missing use strict; use warnings; -- it seems like half the questions we have would be rendered moot if the "programmer" simply made use of the diagnostics and debugging features available (or more realistically, pop up a "don't you dare think about posting this question if it doesn't pass a warnings check" dialog. –  Ether Jul 5 '10 at 19:31
1  
@Ether: I don't know anything about [perl] to comment on that, but please express your points as a voteable answer and see if people feel the same way. –  polygenelubricants Jul 5 '10 at 19:32
add comment

Anything mis-tagged with a version number when it has nothing to do with a version, which I find to be a pretty decent percentage of occurrences.

For example, here's a c#4.0 question right now.

share|improve this answer
    
Or that lists the version number (and only the version number) in a tag by itself - ie: .net and 3.5 –  Joel Coehoorn Jul 6 '10 at 2:25
add comment

[algorithm]

The way questions are asked:

  • Bad Tagging: I suppose this is a problem with other tags too. Questions not really relevant to this tag are tagged as such. I suppose nothing much can be done, other than re-tagging and educating the question asker.

  • Asking the wrong question: Usually people generalize their questions and this makes their problem much harder. There is a 'clarification' dialogue which follows because of that. No good solution other than letting the asker know to specify any constraints etc next time they ask a question.

  • Best Algorithm: The questions sometimes ask, 'which is the best algorithm for this?', which is usually meaningless without specifying what defines best, or the underlying model/constraints etc.

  • Too many homework problems: I believe this tag has a good percentage which are homework. Educating the askers might not help and might even get ignored.

The way questions are answered:

  • Quick, correct looking but wrong answers: Sometimes we get quick answers which may look right, but are wrong and either fail for certain input, or have the wrong stated time bounds etc.

  • NP-Complete looking problems: People usually look at the problem as a specific case of a known NP-Complete problem and quickly claim NP-Completeness, which is quite wrong.

  • Only code answers: People just post code without any hint as to what they are trying to do.

  • Benchmarking: Sometimes people write benchmarks comparing two algorithms and claim that their algorithm is superior. Without knowing the actual constraints/scenarios/data etc, this could be a pointless and misleading exercise. Even when it could be relevant, sometimes they have bad implementations which make it pointless again.

Voting patterns:

  • Voting up wrong answers: Quick, correct looking but wrong answers are voted up quickly, while well-thought out and correct answers which come a few minutes/hours later are buried under. This is probably a problem with other tags too.

  • Not enough downvotes: People seem to downvote very rarely in this tag. This is all the more needed because of the above issue.

  • Voting up answers with benchmarks: If your answer has a benchmark, you are sure to get a couple of upvotes. I have even seen comments saying 'voting up because of a benchmark' without even checking to see if it is needed, and if that benchmark's results can even be trusted.


Do we need a CW-built index to common [algorithm] questions?

Yes. As an example, there are dozens of dupes on generating permutations and combinations alone, and some people couldn't search for these effectively because they didn't know what these concepts are called.

share|improve this answer
add comment

[.net] is frustrating, because it does and doesn't pick up [C#], [VB.NET], [F#]. If it either picked up all or none it'd be ok, but with it the way it is you never know what you'll get.

share|improve this answer
2  
In my opinion, that's because we not only use these tags to mean, "this question is about [tag]", we also use them to mean, "I happened to use [tag], though it's not relevant to my question". –  John Saunders Jul 6 '10 at 0:50
    
@John, I totally agree. Not sure what the solution is, outside of lots of manual re-tagging. –  C. Ross Jul 6 '10 at 10:29
    
Ross: The issue is that "the Management" doesn't see this as a problem. From my point of view, as soon as you've got one mechanism doing two things, you've got a problem, whether or not you realize it. –  John Saunders Jul 6 '10 at 19:48
add comment

[subjective] is frustrating.

The tag is often applied in a very subjective way, and often contradicting the dictionary meaning of the English word subjective.

share|improve this answer
add comment

[staus-deferred] is currently used to say two different things:

  • We'll think about that later
  • We think that's a third party problem and we hope they'll do something to fix it.

See Joel's comments in this question for more info.

IMHO, it should be split out into two more accurate tags.

share|improve this answer
add comment

You must log in to answer this question.

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