Should reputation points expire? [duplicate]

Question

While looking at the SO question How can I make a redirect page in jQuery/JavaScript? It occurred to me that this was an extremely simple question with an extremely simple answer. And yet, it has garnered the OP thousands in upvote points, simply because he was the first one to ask this question in the early days of SO. There are many examples of these simple questions and I believe they don't accurately model a user's participation in the SO community. People shouldn't be getting thousands or even tens of thousands of points from a question just because they were lucky enough to get it in the system a while ago.

Another related issue is that the non-expiring point system rewards users for seniority. Simply put, the longer you're around SO, the higher your score will be. While there's some merit to this, it seems to have a disproportionate effect. So what can we do to limit these effects? I'm just brainstorming but here are a couple ideas:

1. Automatically close voting on questions after a certain time period. Good questions will still be rewarded, but they won't give disproportionately large rewards simply because they are really old questions.
2. Expire users' points after a certain time period. If you expire users' points after, say 3 months, then this will force them to stay active and their points will be more a reflection of "How much have you participated lately"? It forces old users to stay active rather than resting on their laurels, and allows junior members a chance to catch up to senior members of the community by putting in a lot of effort. Currently, there is no chance at all for junior members to ever catch up with senior members and their only 'sin' is that they didn't register 5 years ago.

Answer 1

Automatically close voting on questions after a certain time period.

This will not create the utopia you hope for. There are two kinds of common questions:

• Easy, highly specific to the OP, questions. Answers to these tend to get lots of upvotes in the short term then none ever again.

• Hard, specific but generally applicable questions. Answers to these tend to get very few initial upvotes because people just browsing the site aren't familiar with these topics. However, these questions are much more useful and people find them on Google. They have a long tail of upvotes.

By cutting off the long tail you penalise the really valuable Q&As; the ones people search for and find answers to.

Of course there's also the super rare first basic question on a particular topic. These are super rare so not worth worrying too much about but they help a lot of people so I'm okay with this as a situation

Expire users' points after a certain time period

This is quite unkind to members who've taken time to help people; you spent years helping people but you had a few busy weeks at work so back to the start for you.

Answer 2

I totally agree with your basic premise: it seems super unfair that anyone should earn thousands of rep points for a cheap answer from 2009.

However, your remedy has been suggested before and the math on what the effects would be has been done by Kevin Montrose here. It shows interesting results.

1. We don't know how widespread the actual problem - users gaining tons of reputation points from trivial answers - is. But in any case, it is going to be much smaller than one might anticipate:

   A total of 4,843 users have at least one post that has earned them >= 500 rep.

   That's 4,843 users out of about 60,000 users with more than 500 reputation points. Out of these 4,843, some have earned some significant undeserved points from a single post. (Keep in mind that there are some legitimately highly upvoted answers on SO - answers which took the authors days or weeks to research.)

2. Correcting this by putting a limit on both the maximum rep from a post, and the time frame, would affect a significant number of users across the reputation spectrum.

   The premise in the other suggestion is to remove the daily rep cap, and to limit the amount of earnable rep to 100 per question instead. Also, rep can be earned only 6 months after the answer was posted.. The results are such:

   Reputation Band    # Users    Average Change In Reputation
   [1000, 5000)        22,723                            -486
   [5000, 10000)        3,070                          -1,670
   [10000, 20000)       1,549                          -3,034
   [20000, 30000)         488                          -4,681
   [30000, 40000)         178                          -6,986
   [40000, 50000)          81                          -7,133
   [50000, 60000)          52                          -8,079
   [60000, 70000)          39                         -11,072
   [70000+]               125                          -6,511
   

   Lower-reputation users would be disproportionally affected by your suggested change, some possibly seeing half their reputation go.

   These numbers are also an indicator that the rep gain from these kinds of posts has not gone only to the "super rich", so to speak, and that at least the top users made it to the top despite the current system, not because of it.

   If the suggestion above were implemented, the top ten users of Stack Overflow would gain between 12% and 219% additional reputation points, while a total of 101,727 users would lose reputation from the change.

3. Your suggestion would not actually create a fair playing field; it would simply disadvantage different groups of posts.

   Because it makes a post's rep gain entirely dependent on the traffic it sees after it is posted, "timeboxing" would disadvantage posts in low-traffic tags, while favouring posts in high-traffic ones. That is an imbalance that is already a problem anyway. Older posts would also be at a massive disadvantage because they stem from times when SO's traffic levels were a fraction of what they are today.

While perhaps creating some improved fairness in some cases - but probably much fewer than you and I think - "timeboxing" would very likely cause great distress among users on all reputation levels. Previous, much smaller changes in the reputation system have provoked protests and wailing on such an epic scale than no one wants to remember them. The expected backlash from this would be way worse, which makes this unlikely to ever happen.

Also, it wouldn't really create actual fairness, either.

Answer 3

We already have a daily upvote cap which circumvents outrageous reputation gain.

It's life I'm afraid. Just like the founders of a firm have the majority of the equity. But you're in the tent now and, by your own arguments, you have it easier than future contributors.

Programming is a rapidly developing field so there will always be opportunities to make an impact.

But you're holding reputation in too high a regard: and drawing comparisons with other folk is not constructive for anyone. We're trying to build a knowledge base here. Too bad that the "what's the difference between a struct and a class" has already been asked and answered!

Why not concentrate on answering questions and watch the points trickle in?