-13

If I ask or answer a good question in a very popular tag it gets a lot of votes, and thus a lot of reputation.

However a question or answer in a less popular tag or category will get less votes, and less reputation, even though it may be much more specialized, well thought out, and show more knowledge than the popular tag.

Surely there must be some way to address this reputation imbalance. One possibility is to apply a reputation multiplier for less popular tags. So for example, an answer in a less popular tag gets 3x reputation per vote. And the moderators make sure people don't abuse this by adding multiplier tags to questions they shouldn't be.

Thoughts?

The idea above is only a suggestion, but I'd also like to hear other better ways to address the problem. I am not in any way proposing the idea as a completely thought out, fully baked solution.

Edit for all those saying that there's no problem

Consider this scenario:

Suppose a new user joins the community because the are really interested in a really obscure, specialized tags. They participate, and spend lots of time analyzing and thinking through an excellent, well thought out answer that helps out the person asking it immensely. Now all they get in return is an acceptance and 2 votes, and their total reputation at this point is 35.

Now as they are browsing the site, they see other tags where mediocre answers are getting 20+ votes (and 200+) reputation. How motivated is this going to make them feel?

I agree that the proposed solution is not ideal, but surely if we all discuss this there must be some way to provide the user with that missing motivation?

|improve this question
  • 2
    Sigh. Votes on Meta are different – Oded Aug 22 '12 at 19:29
  • How is this multiplier to be determined? What are the caps for it? Should it be applied retroactively? Which of the tags should be considered for this? What happens if a tag gets popular (or less popular)? – Oded Aug 22 '12 at 19:31
  • @Oded OK but is it too much to ask for a better idea to address the problem? – zooone9243 Aug 22 '12 at 19:32
  • 3
    @ZaidMasud There's no problem. – yannis Aug 22 '12 at 19:32
  • The downvotes so far seem to suggest that there is no problem. – Oded Aug 22 '12 at 19:33
  • 3
    @Oded, Yannis - the problem absolutely exists. There's no doubt of that and it's been pointed out many, many times. But as Oded says, this solution would bring more new problems than it fixes... – Pekka supports GoFundMonica Aug 22 '12 at 19:33
  • 2
    An approach more in keeping with SE patterns might be to have a badge (multiple awards possible) for answers satisfying a certain ratio of vote:view counts. – Monica Cellio Aug 22 '12 at 19:34
  • 1
    @Oded and Yannis If 1,000 people regularly view answers on tag A, and 50 people regularly view answers on tag B, won't tag A answers get a lot more votes? – zooone9243 Aug 22 '12 at 19:36
  • 1
    @Pekka - If you see the problem as: less popular tags will get less views therefore less answers. To motivate people to answer those, we want to introduce some sort of shifting reputation scale, then I agree that would be one way to try and address the issue. – Oded Aug 22 '12 at 19:37
  • 8
    as an avid contributor in few low-activity tags I for one find the situation fair enough. I've got my Unsung Hero badge easily and I don't need to fight FGITW which feels really comfortable – gnat Aug 22 '12 at 19:38
  • 1
    @gnat - I bet that the vote to accept ratio you have is also significantly higher than for those answering in popular tags. – Oded Aug 22 '12 at 19:40
  • 5
    @ZaidMasud Obviously, I just disagree that it's a problem. You get less reputation, so what? On the other hand, there's less competition and possibly better crafted questions and answers (if the tag is truly a specialized one). That's far more important to me than reputation. – yannis Aug 22 '12 at 19:40
  • 2
    @Pekka I'm (primarily) a PHP developer and I rarely visit the [php] tag on SO. Popularity isn't necessarily a good thing, certainly a good PHP question will probably get a lot more votes than, for example, an equally good Haskell question, but I honestly don't have any energy left to swim in the vast ocean of crap that is [php] questions on SO until I find that one good question. – yannis Aug 22 '12 at 19:49
  • 1
    @rabmcnab Too late ;P – yannis Aug 22 '12 at 19:50
  • 2
    I value my silver plone badge just as much, if not more, than my golden python badge. The plone tag doesn't gain me as much reputation, but because there isn't much competition I am the only person holding a silver badge for that tag. I answer plone questions because I am an expert in that field, not because of reputation. I don't want to have to deal with crap answerers flooding my 'valuable' specialty tag either, trying to find answers just for the reputation, thank you. – Martijn Pieters Aug 22 '12 at 19:58
9

Three reasons I expect this won't happen:

  1. A lot more data now needs to be tracked (multiplier of each tag) and kept up to date (new tags, tag suddenly becomes more or less popular)
  2. Large possibility of abuse. Someone could go around retagging questions to give themselves (or others) more (or less) reputation. "And the moderators make sure people don't abuse this" - ask a moderator if they would be willing to keep an eye on a system like this. I highly doubt it.
  3. Reputation recalculations. Whenever the multiplier of a tag changes, all questions tagged with that tag (and all their answers) need to be revisited, and the reputation awarded for them recalculated. Whenever a question is retagged, reputation needs to be recalculated. This is A LOT of recalculation.

What is the real benefit of this feature? Sure, it helps people who participate in obscure tags get more reputation, but why is that a good thing (I'm not saying it's bad, I just don't see how it helps the site nearly enough to overcome the three reasons I've listed here).

|improve this answer
  • These are good points. I'm hoping to see some alternative solutions to the problem as well. – zooone9243 Aug 22 '12 at 19:43
  • 4
    Another reason: people with no expertise what-so-ever flooding 'high rep' tags to gain reputation, decreasing signal-to-noise. – Martijn Pieters Aug 22 '12 at 20:02
6

Your underlying premise (Popular tag = upvotes) isn't supported by the data.

If you rank the tags by how rewarding they are on average (R_Rank) and how Popular the are (P_Rank) you'll find that the tag with the highest P_Rank in the Top 50 by R_Rank is

Pearl (R_Rank 35, P_Rank 58)

The highest R_Rank of tags in the Top 10 of P_Rank is C++ with an R_Rank of 107. C# has a R_Rank of 307. My best tag, SQL has an R_Rank of 515 (ouch, shoulda learned to program in haskell I guess).

Top 50

Rank Tags                    Upvotes Answers Avg Upvotes / Answer TagSizeRank 
---- ----------------------- ------- ------- -------------------- ----------- 
1    haskell                 85019   16988   5                    183         
2    operators               29300   5982    4.9                  787         
3    functional-programming  44322   9291    4.77                 471         
4    branch                  14374   3319    4.33                 859         
5    clojure                 32517   7646    4.25                 392         
6    mathematica             19857   4701    4.22                 614         
6    c++11                   22239   5266    4.22                 545         
8    scala                   86679   20580   4.21                 120         
9    ocaml                   9226    2215    4.17                 1142        
10   lisp                    24304   5960    4.08                 667         
11   git                     141577  34931   4.05                 65          
12   c++0x                   10112   2725    3.71                 1048        
13   f#                      31666   8657    3.66                 363         
14   automatic-ref-counting  6515    1788    3.64                 1085        
15   syntax                  52890   14577   3.63                 271         
16   vim                     59431   16468   3.61                 191         
17   interview-questions     37364   10584   3.53                 561         
18   const                   14173   4101    3.46                 1047        
19   floating-point          31290   9121    3.43                 452         
20   enums                   33913   9984    3.4                  407         
21   ggplot2                 6829    2021    3.38                 905         
22   resharper               10349   3075    3.37                 890         
22   version-control         63045   18711   3.37                 221         
24   boolean                 15414   4591    3.36                 852         
25   constants               10552   3187    3.31                 1088        
26   mercurial               27601   8362    3.3                  313         
27   ienumerable             9624    2941    3.27                 1130        
28   lambda                  26433   8158    3.24                 419         
28   stl                     46639   14400   3.24                 295         
30   elisp                   7789    2412    3.23                 1041        
31   coding-style            51771   16076   3.22                 321         
31   warnings                10087   3130    3.22                 908         
33   closures                10982   3460    3.17                 898         
34   programming-languages   47768   15101   3.16                 384         
35   perl                    155729  49393   3.15                 58          
36   clr                     11431   3640    3.14                 821         
37   standards               11369   3642    3.12                 996         
38   scheme                  12697   4078    3.11                 762         
39   extension-methods       10388   3355    3.1                  1017        
40   erlang                  16032   5183    3.09                 531         
41   language-agnostic       74790   24304   3.08                 245         
42   operator-overloading    14285   4685    3.05                 810         
43   integer                 18073   5954    3.04                 698         
44   comparison              20213   6708    3.01                 620         
44   compiler                44720   14838   3.01                 261         
46   types                   40522   13494   3                    267         
46   iterator                20781   6934    3                    553         
48   null                    23718   7929    2.99                 490         
49   try-catch               11177   3791    2.95                 1057        
50   naming-conventions      18099   6181    2.93                 791         
50   emacs                   34831   11906   2.93                 244         
50   r                       74981   25574   2.93                 80

Data.SE Query

|improve this answer
  • Science always wins. – yannis Aug 22 '12 at 23:11
  • Thanks for compiling this, but I am trying to understand how avg upvotes / answer are related to my premise. Popular tags = upvotes is sort of the premise, but the premise is better stated as "popular tags have higher frequencies of significantly upvoted answers." So statistically I would be much more interested in the distribution ... for example "Percentile of answers with > 25 votes" per tag. Unfortunately I don't think the "avg upvotes / answer" measure is helpful at all. @YannisRizos I encourage you to do some critical thinking as well, even though you have made your bias very clear. – zooone9243 Sep 7 '12 at 16:01
  • @zooone9243 You claim If I ask or answer a good question in a very popular tag it gets a lot of votes, and thus a lot of reputation. However a question or answer in a less popular tag or category will get less votes, and less reputation, even though it may be much more specialized, well thought out, and show more knowledge than the popular tag. So if a good post gets more votes in one tag than other, it must follow all posts will get more votes in one tag than another. So the averages would show that if it were true. – Some Helpful Commenter Sep 7 '12 at 16:48
  • @zooone9243 as an aside (if I understand what you mean by frequency of good answers) you can take a look this frequency query – Some Helpful Commenter Sep 7 '12 at 17:27

You must log in to answer this question.

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