Looking at the oldest (2008...2010) questions it's easy to find those with 50+ votes and answers with 100+. This trend seems to be changing, new questions (especially with certain tags) pretty often have only a few votes (or they stay unvoted even when the question itself is good and not just a "What's wrong in my code?"). The same goes for answers, I see a lot of very good answers but very often they don't earn more than 5 votes (in the first 10 minutes of life).

I know that an old question has been around for years so it has accumulated votes over the years but for some tags (maybe with more disciplined users) everything still works as it should.

So my question is: Stack Overflow rules were good for a limited number of users and Q/A (thresholds for badges, for examples). Are these rules still valid? If the assertion "users are very interested in reputation" is true then to change something may increase both the number of votes and number/quality of answers. For example:

  • Add rep +2 for each upvote on answers and +1 for each upvote on questions (all users would be encouraged to vote/check/read and other users would have more satisfaction posting a good answer). This change may simply mimic what happens with downvotes. I understand that it could start a "random voting" season but a few restrictions may be applied (just as an example, the rep change may be applied only if the question reached a given number of upvotes and it has no downvotes).
  • Rep change when someone downvotes a question is -1 for downvoter and -2 for the one who posted the (bad?) answer. Maybe many people do not downvote for the -1 but no one is afraid to post a random answer (untested, wrong, out of topic, answer instead of comment) as scoring -2 isn't that bad when you consider that one upvote recovers all the lost reputation with 5 downvotes (and it's not common to see answers with five downvotes, even when they may deserve it).

  • The point then is: Why don't upvotes and downvotes have the same weight? +10 for an upvote, -10 for a downvote (or at least -5). This may even stop posting quick and incomplete answers to be the first one (and to catch, maybe, some upvotes when the question is in the period of high visibility).

After some clarifications in the comments I would point my topic better:

  • Why votes per question and votes per answer plotted per year seem to show a decreasing trend? Is it just about users' discipline?
  • Has it a negative impact? Is there a way to improve average quality using reputation system? Specially for the second point I think it would increase average quality for answers.

Quick test for old pro users (I would be really interested to know the results!): pick randomly five of your simple (= no more than syntax check) oldest (< = 2009) answers. Calculate the average score per month (twelve upvotes in one year = one upvote per month). Please note that often this isn't really good because that kind of answers aren't reviewed many times (a better but more complex test would involve the number of views too). Now pick five new answers, calculate their score per month and compare them.
Moreover do the same for five pro answers. Something that involves a complex topic or something you're proud of.
What's the result of this comparison?

  • 6
    It's worth noting that many of those older questions accumulated a lot of their votes over time, not immediately after they were asked. I know I've seen a very long tail on votes for my older questions and answers, with some of the higher-voted ones having starting with no votes for a week after I posted them originally. – Brad Larson Jun 25 '12 at 20:52
  • @BradLarson I agree but pick a good answer of 2008. Let's say it has 50 up votes (but you'll find many answers with higher score). In theory 2008...2012 = more or less 10/12 u.v. per year (rounding, 2012 is just in the middle). Now pick a good answer of 2011, you expect to see a score around 20 but it's pretty uncommon (again, for some tags) to see answers greater than 5 or 10 (when lucky). – Adriano Repetti Jun 25 '12 at 20:59
  • @Adriano well this one got 2K upvotes and its only about a year old. And this one has 1K upvotes and its less than a year old – Some Helpful Commenter Jun 25 '12 at 21:07
  • @SomeHelpfulCommenter I talked about trend, thanks God it's not true for every single answer/question. Moreover take a look to this: meta.stackexchange.com/questions/198/…, initial reputation of the one that answers does matter. – Adriano Repetti Jun 25 '12 at 21:11
  • @SomeHelpfulCommenter see my comment for Ernest's answer. Some users are more disciplined. – Adriano Repetti Jun 25 '12 at 21:13
  • 1
    @SomeHelpfulCommenter - that one answer got 2248 upvotes, and 1 downvote. That one guy is hardcore. – Ernest Friedman-Hill Jun 25 '12 at 21:38
  • Some related older posts: Should there be more incentive to upvote? or What is our reward for voting? What is the metric for voting? It's quite probable that you can find other similar discussion. – Martin Aug 11 '19 at 4:47

Maybe I'm stating the obvious, but one possible explanation for questions and answers with such massively high upvote totals is this: someone comes to the site, and decides to peruse through questions sorted by vote totals, or a user's answers sorted by vote totals (such as in the screenshots below), and will subsequently discover old contributions with tremendously high vote totals.

In that situation, who wouldn't be curious, and click on those questions/answers? And how many of those users would be similarly impressed, and pile on?

All I'm saying is that it's natural for the "rich to get richer." (That's not a bad thing – it's merely an explainable phenomenon that perpetuates over time, simply because SO allows users to sort by vote totals, making these "classics" easy to find, so that their numbers eventually climb even higher.)

enter image description here

| improve this answer | |
  • I agree but now I see that any change to rep system isn't really welcome, it was good (I think) in old times but now in my opinion it doesn't fit the needing of new users. Of course older people here can't feel that 'cause when they were 100 rep they did get 50 up votes for "add a semicolon at the end of the line" (same for questions too!!!). Oh it sounds like generational change! LOL – Adriano Repetti Jun 26 '12 at 7:18
  • The rep system already has a 200 pt/day cap, so 50 upvotes for "add a semi-colon" won't get that lucky answerer more than 200 points and a badge anyway. I don't think the rep system is perfect, either, but I don't think it's in need of major reform. I pointed out that SO users can sort by vote totals; it also lets anyone (even you) sort by date as well. Your idea of going through old questions and finding thoughtful but neglected answers is interesting; no one says you can't do it on a weekly basis; I think what's being downvoted here is an overhaul of the rep system merely to encourage that. – J.R. means 'Just Reinstate' Jun 26 '12 at 9:02
  • I think current limit is higher than 200 pt/day (for multiple answers) anyway even in one week is a little bit too high rate! ;) I do not worry about down votes here (I guess on meta +1 = it may be useful, -1 = it may not be useful; I'm not English speaker so I hope didn't appear rude or so) but I would see some answers like "point 1) can't be applied because..." or "it's not useful because as you can see on this stats...". By the way I did +1 'cause you at least tried to give some explanation! – Adriano Repetti Jun 26 '12 at 10:11
  • Example of what I mean: this question had 5 votes on the question itself and 8 votes on answers. How many simple/naive/talkative Q/A get the same feedback on SO? – Adriano Repetti Jun 26 '12 at 10:27

I think the real answer to your first question is that (a) there used to be a lot fewer questions; each question got more of an audience because it had less competition. Now, questions fly off the front page in just a few minutes. Rep is much harder to come by these days; and (b) it used to be much easier to ask a question that had never been asked before. Now most questions are dupes.

Not that I want to dis Isaac Newton or any other great scientist of the past, but you notice how hard it is to become a household name by discovering a new fundamental law these days? Not as easy as it used to be, is it?

Anyway, I think you've pretty much already given the answers to your other questions.

  1. If you got points for upvoting, then some people would upvote for no reason.
  2. If you lost the same number of points for a downvote as you gained for an upvote, then some people would hesitate to give answers unless they were really sure they were going to nail it.
| improve this answer | |
  • 1
    Nature's laws won't change, languages, frameworks and protocols do. Quickly. Moreover this is not true for every topic (for what I saw, for example, Java users are more disciplined than, say, C# users). 1) I see, that's why there should be some rule to prevent this (or a day limit, even very very low). 2) If they hesitate then...good, reached the target. Average quality of answers will be increased and there won't be tons of "noise" answers. – Adriano Repetti Jun 25 '12 at 21:04
  • 1
    @Adriano why do you think that Java users are disciplined? – Some Helpful Commenter Jun 25 '12 at 21:15
  • @SomeHelpfulCommenter I know/knew that an example is always a bad idea! LOL Raw statistic: filter per tag (C# and Java), order by votes, compare by total votes, number of views and age. But again, this is just an example not a critic to C# users (or whatever). The point (I would be happy if someday SO will provide custom statistics) is that in the past both question and answers did get more attention (= votes and because rep is one of the bricks of SO then it would be nice if it comes back to its original meaning). Otherwise...why it's not anonymous (with rep but without name)? – Adriano Repetti Jun 25 '12 at 21:22
  • 3
    @Adriano does this do it for you – Some Helpful Commenter Jun 25 '12 at 21:23
  • @Adriano also the stats guys looked into how much rep plays a roll in upvotes. it doesn't – Some Helpful Commenter Jun 25 '12 at 21:27
  • 1
    @SomeHelpfulCommenter great link, thank you! I didn't know about data.stackexchange!!! Ok, Java users aren't more disciplined :) Let's say Haskell users compared to flex4 users :) – Adriano Repetti Jun 25 '12 at 21:43
  • @SomeHelpfulCommenter OK for the analysis about the link I pointed in the comment but my question is still about changes in rep system. Why Haskell (lol) users are more disciplined than C#/Java users? Why this changes over the time? Has it a negative impact? Is there a way to improve average quality using reputation system? – Adriano Repetti Jun 25 '12 at 21:54
  • 1
    @SomeHelpfulCommenter So the message is to stick to functional programming languages for the best SO experience :) – Lorem Ipsum Jun 25 '12 at 22:14
  • @yoda that and avoid Crystal Reports (as if one needed another reason) – Some Helpful Commenter Jun 25 '12 at 22:20

You must log in to answer this question.

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