What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 128 Stack Exchange communities.

Here is an interesting list of questions:

All top voted answers on non Wiki questions:

select top 20 '[' + q.Title + '](' + 'http://stackoverflow.com/questions/' + cast(q.Id as varchar(max)) + ' "link")', 
cast(a.Score as varchar) + ' <br/>' From Answers a 
join Questions q on a.ParentId = q.Id and q.IsWiki = 0
order by a.Score desc

(note, I have no IsWiki flag for answers so I can not filter them out - sorry Jon)

Help me remember a quote from Alan Kay 241
"Hello World" in less than 20 bytes 200
Is DateTime.Now the best way to measure a function's performance? 164
How do you set, clear and toggle a single bit in C? 158
Is it wrong to go to interviews while employed? 135
In C arrays why is this true? a[5] == 5[a] 129
Is Mono ready for prime time? 126
How do I check if an integer is even or odd? 122
"Hello World" in less than 20 bytes 120
What are the barriers to understanding pointers and what can be done to overcome them? 117
What is the most EVIL code you have ever seen in a production enterprise environment? 117
Why does everyone like jQuery more than prototype/script.aclo.us or mootools or whatever? 116
Why does C++ compilation take so long? 116
How do I move the turtle in LOGO? 112
What are MVP and MVC and what is the difference? 110
What's the difference between JavaScript and Java? 108
How do emulators work and how are they written? 107
How do I move the turtle in LOGO? 102
Why are we using i as a counter in loops 102
How to get attention for your old, unanswered questions 99

Now, I completely agree that Alan Kay should have a gazzilion rep. Still, something does not sit right for me with this list.

Should there be an upper limit on how much rep a person can get out of an answer?

share|improve this question
    
Someone just mentioned this idea today, and I discovered this ancient and almost entirely ignored post. Was it ever decided this is a bad idea, or was it just forgotten? –  Michael Mrozek Mar 6 '12 at 3:59
add comment

4 Answers 4

up vote 5 down vote accepted

This has been talked about before, at least a little bit. Back when he tried it (around Christmas?) that changing the time rep limit to a per-post rep limit made little difference. Personally I still think it would be a good idea. You should probably read all the arguments for and against the current time-based limit in this feature request (declined).

I'll keep being in favour of this, but at the same time I don't expect the situation to change any time soon.

EDIT: (On phone - please excuse typos.) I think it's worth thinking about the purpose of rep limits in general. One point is to limit how quickly/easily someone can get edit rights etc. I would argue that someone who posts 20 answers each of which gains 10 upvotes in a single day has proved themselves more invested in the community than someone who posts a single popular answer which gains 200 upvotes over the course of three months - during which time they may not even come back. To me, the first person deserves all the rep from their answers, but the second person's rep could reasonably be capped at (say) 500 for that one answer. In particular, positive feedback makes is a lot easier for a post to go from e.g. 40 votes to 45 than from 0 to 5.

Having said all this, I'm not as bothered as all of this discussion makes it sound. Rep limits are somewhat important in terms of early access to privs, but they probably don't deserve quite the attention some of us pay them... I would like to hear Jeff and Joel discuss this a bit more on the podcast though.

share|improve this answer
    
I think I may have misunderstood you comment to my answer up there. I think you were referring to the daily rep limit when you mentioned time-base limit. I wasn't even thinking of the daily rep limit as something needing to be changed in that answer (that's an entirely different debate) –  TheTXI Jul 18 '09 at 6:33
add comment

See

http://blog.stackoverflow.com/2008/12/daily-dose-of-daily-reputation-ca/

Based on my analysis of Stack Overflow voting patterns in 2008:

Implementing a logarithmic decline in the value of upvotes would make almost no difference to most people’s reputation scores.

Capping all posts at 10 votes (for scoring purposes only) would make almost no difference to most people’s reputation scores.

We'd have to run this test again to see what the effects would be, but knowing what I know about human behavior and statistics, I expect the results would be nearly identical.

edit: tests run and ... in 2012, most accounts would lose rep from this change.

share|improve this answer
    
Good. For the majority of users it makes no appreciable difference. Perfect. What about the ones where it would "level things out" and not just give amazing boosts? I think that a test implementation should be worked out against a backup of the database and see what is actually generated, along with a complete recalc for the entire site. And yes, I'm willing to work with a few members of the community to do that, rather than just dump it on someone's desk, altho I daresay that the SE team would be much faster. –  jcolebrand Jun 14 '12 at 20:03
2  
I suspect the effects would be different to 2008 just because now regular users (not just me!) have so many more posts. I'd love to see the results of running the tests again. –  Jon Skeet Jun 14 '12 at 20:11
    
"Most accounts would lose rep." Probably so...and I think it solves the wrong problem. Longer answer below, and I still think the leagues are a better solution for the "fairness" issue. –  CodeGnome Jun 18 '12 at 8:21
add comment

You're not taking into account the fact that for items which gain a lot of votes incredibly fast, the user is probably going to be stuck only earning as much as their daily rep limit would allow.

But for items that have been open a long time and earning rep that entire time, I still say you do not want to limit the rep gained from it. What you are basically saying is that as time goes on (or the popularity goes on) that item is obsolete and should no longer be rewarded.

That's like saying that if I write a book, I should only get paid for the first 100,000 copies and the rest goes to nobody in particular.

share|improve this answer
    
While I see your point, I'd say that a per-post rep limit makes a lot more sense than a time-based limit. We should discuss why a limit makes sense at all and then work out the most appropriate form. –  Jon Skeet Jul 18 '09 at 6:27
    
Jon Skeet: I am definitely on the side of a limit not making sense in any real form. I was just using time frame as an example that you can't just look at the score and say "Ah! they earned at least X amount of rep!" I think you probably know about this more than anybody. –  TheTXI Jul 18 '09 at 6:31
    
I mean a rep limit per post in any real form such as by score or by time-frame. This was not meant to be a jab at the overall daily rep limit. That's a different argument altogether. –  TheTXI Jul 18 '09 at 6:35
    
I think that when a post becomes old, kaizen should be rewarded not the original old and potentially out of date information. I'm saying "Java and Javascript are similar like Car and Carpet are similar." is funny, but it's not one of the top 20 most important bits of info on SO. –  waffles Jul 18 '09 at 7:27
    
TheTXI: I think we can discuss the two types of limit together. The most important question IMO is what the purpose of having a limit is in the first place. Once we know that, we can move on to work out what kind of limit accomplishes that goal in the fairest fashion. –  Jon Skeet Jul 18 '09 at 7:54
add comment

TL;DR

A canonical answer should continue to be rewarded, and past 10k I'm not sure that this causes any real-world problems other than a skewing of the bell curve.

Analysis

I think that this is a question that begs for a better definition of why anyone cares. There tend to be certain classes of post that should be considered:

  1. Posts which get a ridiculous number of upvotes like this one. Was the answer really worth 2,450 rep? I don't know, but a lot of people voted for it.
  2. Posts that get a lot of rep very quickly.
  3. Posts with legs that generate a little rep over a long period of time.

On the one hand, I sympathize with the idea that a single answer should never give someone 25% of 10k privileges before they've put their time in. On the other hand, in the example used the poster currently has ~79k rep; the votes were probably earned over time, and begs the question of whether 2.5k rep (as a percentage of the person's overall rep) is even relevant at that point.

A canonical answer should continue to be rewarded, and I'm not sure that this causes any real-world problems other than a skewing of the bell curve.

Unless the chosen example is typical, and all these high-rep questions get upvoted like that in a few days or weeks, it probably just reflects longevity of the posts. Note that the post in question is almost two years old.

The current system clearly rewards time-in-grade and eyeballs attracted. With that in mind, I'm honestly not sure how per-post caps do anything but incentivize people to post more answers rather than great answers.

I wouldn't change the current system unless one can point to some valid way to differentiate "deserved" from "undeserved" rep from these sorts of posts. Fairness is a terrific goal, but I'm not sure you can create it out of statistical crowd-sourcing.

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 .