-15

How about complete revamp or removal of reputation system?

The issue I'm having with current system is that as far as I can tell, current system:

  1. Does not reward knowledgeable people and difficult questions (and answering those questions).
  2. Rewards trivial questions that can be answered by opening any tutorial on first page. Also encourages energetic people (who are not necessarily knowledgeable) to answer.

Direct consequences I've observed so far:

  1. Most difficult question I ever encountered in last 5 years or answer that took 3 hours of research - zero or one upvote in month.
  2. Simple question that can be answered with 3 lines of code and requires you to open any programming book and think for 15 seconds - 50 upvotes in next 3 minutes.

In my opinion, this is pure lunacy. I think this will be harmful to community in long term, although it will attract many people to the site.

I think that current system encourages beginner programmers who are incredibly active, but do not necessarily posses enough experience. I've already seen person that hasn't passed fanboyism phase in C++ (and needs 5 more years of practice) pass 10k rep barrier in one year. I would prefer a system that would reward experienced people and would encourage programmer to grow and hone skills/knowledge further (internet is already flooded with questionable beginner material), because in my opinion programming needs more professionals. Current system "aint it".

Of course, it is possible that I misunderstood purpose of SO, and it was originally supposed to help hobbyists without encouraging them to grow and learn more. In this case I'd appreciate if somebody pointed me to section of faq that states that SO wasn't meant to promote getting better at programming. Also my opinion can be unpopular.

Unfortunately, I can't suggest how reputation system could be improved (since there's difficulty of automatically measuring question complexity), but it doesn't mean that there is no way to do it.

12
  • 8
    I agree about how often easy questions/answers get up voted. My highest reputation answer on SO is an embarassment to me. However, I don't think that invalidates the reputation system overall. It's dumb luck whether you hit an "easy reputation" q/a. and the daily rep limit puts a very effective cap on the initial flood of votes on those. Meanwhile, the good stuff does attract votes, if more slowly. Feb 11, 2012 at 16:23
  • 1
    If there was no reputation system, what would Pekka do for a user name??
    – Lix
    Feb 11, 2012 at 16:27
  • 1
    Is it a coincidence that the SO answer I speak of above just got another up vote? lol Feb 11, 2012 at 16:27
  • 1
    @Lix Pekka'sFantasyRepBong Feb 11, 2012 at 16:27
  • 1
    @Andrew - Haha - birlliant! You think the Pekka signal we sent out will catch his attention and he will come shed his wisdom on poor answerless Sig here?
    – Lix
    Feb 11, 2012 at 16:30
  • 2
    I award bounties (from 50-500 reputation) to outstanding answers that haven't got the attention they deserve.
    – Ambo100
    Feb 11, 2012 at 16:30
  • 3
    The biggest issue with a revamp like this is its effect on current contributors. Past reputation changes were surprising, disrupting and in many cases not welcome to existing contributors.
    – Oded
    Feb 11, 2012 at 16:33
  • 3
    @Ambo100 Excellent point, and a great idea I plan to use more often. (I'm always forgetting the bounty system). Even the lowest bounty is worth 5 up votes, and uniquely decorates the answer. Feb 11, 2012 at 16:34
  • 2
    @AndrewBarber I consider most of my top twenty answers to be rather over-rated. Oh, some of them are good answers, but only about three of the top twenty show any insight (as opposed to just some knowledge). That's crowdsourcing for your: most people aren't equipped to judge answers to hard questions. Feb 11, 2012 at 17:43
  • @dmckee Indeed. I know I have looked at very complex, involved answers and hesitated to click up vote because I wasn't sure if I understood things well enough to judge the answer useful. I try to take the time, but there is clearly more effort involved in casting a proper vote on a more difficult question/answer. I know I also sometimes see questions/answers that I know are "bad" in some way, but are just long/detailed - and which got up votes apparently because of the length/detail. Feb 11, 2012 at 17:57
  • @AndrewBarber (1) " It's dumb luck whether you hit an \"easy reputation\" q/a" : Bad thing doesn't become good only if it happen by chance. (2) "the daily rep limit puts a very effective cap on the initial flood of votes on those" Thats good, but what about the 100-10K upvoted accumulated in months/years? (3) "Meanwhile, the good stuff does attract votes, if more slowly" upvotes=vote rate*time, what you need to compare is vote rate, not total vote. (4) Conclusion: your first comments's statemant can't support the opinion "However, I don't think that invalidates the reputation system overall"
    – jw_
    Jan 15, 2020 at 5:22
  • @Ambo100 I guess that bonus is neglectable compared to the "wrong upvotes" based on current rule.
    – jw_
    Jan 15, 2020 at 5:25

4 Answers 4

3

The complaint here, it seems to me, adds up to a corner case. If all you do is answer obscure questions, you are liable to toil in obscurity.

To some extent, the wicked flourish like the palm tree -- some people harvest a material amount of rep by answering easy questions. However, virtue more or less triumphs. The folks at the top of the graph do much more than that.

They also serve, to abuse another cliché, who only stand and wait for questions to answer. Answering many, many, questions, however simple, is a serious time commitment and a service to the community, and deserves plenty of reputation.

If you are willing to put in hours of work to research and answer questions here, that's very nice of you. However, I doubt that the site is really designed around that activity. The point of attracting experts is that experts can answer many questions without very much time and effort, because they, well, know the answers.

Instead of proposing to replace the rep system, you could imagine either (a) more use of the ex-post-facto bounty pattern to reward the deserving, or (b) making a concrete proposal for some other scheme for rewarding special efforts.

1
  • 1
    It does bother some of us that our supposedly "best" answers, like this one Why declare a struct that only contains an array took a full 10 seconds to write. At the same time we can have more than a dozen accepted answers with zero upvotes. Not fair! :-)
    – Bo Persson
    Feb 11, 2012 at 20:03
1

How about complete revamp or removal of reputation system?

I don't really have a say in this, but this is not going to happen. The reputation system (either in its current or an improved but similar form), is the very core of Stack Overflow.

Unfortunately, I can't suggest how reputation system could be improved (since there's difficulty of automatically measuring question complexity), but it doesn't mean that there is no way to do it.

I don't think that there's any form of reputation system that could prevent popular answers from being upvoted. Basically, reputation is popularity.

Simply put, people upvote what they like. The most upvoted answer I know is Bobince's answer to the question RegEx match open tags except XHTML self-contained tags. It got 4435 upvotes. The answer is true poetic genius and the most Unicode-laden mention of Tony the Pony I've even seen, but only the last line is really answering the question.

Is there any way of fixing this? I don't think so. Bobince's answer is popular. Extremely popular. There's even a T-shirt. You might consider this everything but fair, but I can't think of a single way to prevent this without taking the community-driven out of the community.

Also, in any reputation system only one of the following statements can be true:

  • Reputation means knowledge and insight.

  • Reputation means helpfulness.

A good answer to a difficult question might require a lot of effort, but extremely difficult often means extremely localized.

Also, a simple answer to a simple question will get read and understood by more people. Therefore, it will help more people.

1
  • 1
    "understood by more people. Therefore, it will help more people." You might want to read "The Sack" by William Morrison (short story).
    – SigTerm
    Feb 12, 2012 at 2:09
0

Reputation don't bring the user much real benefits (e.g. can be exchanged for goods), so whether the present rule about reputation is fair does't matter a lot. Then as some post above suggested, there is no need to risk the future of the site to revamp the basic rule.

But here is a method to revamp without changing the basic rule:

(1) Keep every rule about reputation and privilege as before. Users still gain/loose reputation like before and the reputation bring every benefits to you like before. Then this change can't somehow harm the site.

(2) Don't display the reputation value (to other users) any more, since due to votes based on popularity and the randomness of votes-or-not of common users, the reputation value doesn't represent the common meaning of "reputation" i.e. goodness/hard working/contribution well, it only represent a mixture of common meaning of "reputation" (call it gold) and noise (call it sand).

The numbers that should be displayed to other users in the new system are:

<1>:"votes" : total vote of an user, this is near the current reputation value divided by 10

<2>:"reputation" : This is not the current reputation value, but a new one. What should be displayed here it up to decide, one choice is:

int N=5;// this is a tunable parameter to be decided
int displayed_reputation=0;
for(each question/answer of the user){
    if (votes>0){
        votes-=votes_from_hot_question_context;
        votes_fixed=votes_in_first_year(votes);
        if (age_of_QA_in_years>2)
            votes_fixed+=(votes-votes_in_first_year(votes))/(age_of_QA_in_years-1);
        displayed_reputation+=(int)((log(votes_fixed)/log(5)+1)*10);
    }
    else if (votes<0) displayed_reputation-=10;
}

This method both consider "more votes mean more value", but still remove most of the noises. The result is that most question in the site only give the user 10-30 displayed reputation.

Examples of benefits of the above method:

(1) When you get the "we trust you" thing, your displayed reputation won't be 101 every site which is weird, but 1 every site, and you still have the right to comment every site since your classical reputation is still 101 every site.

(2) Displayed reputation can finally be negtive now.

(3) When you get -2 down votes and 1 upvotes, the displayed reputation change will be -10 instead of +6 which is obsurd.

(4) Displayed reputation gain of a Q/A doesn't increase with its age. Only votes in first year and average votes per year after first year is counted.

(5) Votes from hot question link not countd in displayed reputation.

(6) Can differentialize like-upvotes and undo-downvote-upvotes. For example one post is downvoted for 2 times for some serious problem, and later the problem is fixed and get 3 upvotes then finally 1 upvotes. The intention of the 2 of the 3 upvoted may just to undo the downvotes. But in the present system the user get 26 rep gain which doesn't reflect the real situation. With the above system, the displayed reputation only depend on the final 1 upvote which is a like-upvote

-1

The current reputation system--modulo various tweaks--has been part of Stack Overflow from the private beta, and is one of the features that distinguishes this site from other sites that exist to provide answers.

Certainly it has a number of known flaw, and rewarding repetitive answers to basic questions is one of them.

However, the least hypothesis must be that the reputation system---flaws and all---is part of what has made the site so successful.

To throw it out now would be, well, silly.

4
  • While it would be unwise to do global change without testing, it is possible (and makes sense) to test alternative approaches to reputation system in another stackexchange website. Also, "successful" depends on your definition of success. If original goal was to attract large number of people, then yes, it is successful. However if the site was created as a place for programmers to learn/improve, there there is still a lot of room for improvement.
    – SigTerm
    Feb 11, 2012 at 17:54
  • @SigTerm If you think the current reputation system attracts a large number of people, may I respectfully request data backing that up? Because lacking evidence, my guess would be that at best, it is neutral, but actually has negative effects overall. Feb 11, 2012 at 17:59
  • @SigTerm: The system has been tweaked, and probably will be again. If you have a suggest for that make it. But suggesting a major change is going no where. Fast. Feb 11, 2012 at 20:16
  • @AndrewBarber: No data backing, it is personal impression I have based on all sites with reputation I've seen. (IMO) upvote/downvote is addicting and attracts people that like having high numbers in counters. Same applies to rep. Also, people will form groups that will use reputation to supress opinions they dislike instead of supressing opinions that are not useful. Rep also produces worse results compared to 5 star rating, However, 5 star system is vulnerable to extremely large (same opinion) or small (obscure question) groups and also aren't reliable. So there's no perfect solution.
    – SigTerm
    Feb 12, 2012 at 2:03

You must log in to answer this question.

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