What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 130 Stack Exchange communities.

There are over 350 thousand users on stack overflow. While it's nice to be considered part of the top 1%, it's a big group at over 3,500 users, and its too dilute to be a useful measure. You only need about four thousand reputation to be part of this club right now. And while I don't mean to disrespect the hard work people put in to get to 4k rep, with users holding a quarter million rep I can't exactly say that 4k should be considered the top tier.

Please consider defining some function that removes the bottom outliers from the calculation. Just removing all the users with only one reputation nearly halves the total user base, though I think the bar should be higher, at least fifty or so reputation. This would mean at least five up votes on answers, or ten up votes on questions, and that the user has most of the basic site functionality. Currently there are just under 80,000 users with above 50 reputation, which means only the top 800 users would be considered part of the top 1%. This happens to be at right around eleven thousand reputation for now.

I mean, we don't want just anyone to be able to say, "Hey babe, I've got two nines of reliability at stack overflow. Can I buy you an upvote?"

Alternately, introduce the top 0.1% and top 0.01% tiers.

Note: I'm speaking of the All time list, although it might profitably be applied to any long term list. Applying it to the short term lists needs to be carefully considered, as we want new users who are trying to get onto one of the league pages to be able to see their progress, even when they've only received a few votes. It shouldn't matter for the long term lists.

share|improve this question
4  
In the interest of full disclosure, the cutoff needs to be around six hundred rep to kick me out of the club, and quite frankly [insert Groucho Marx quote here] –  Adam Davis Mar 1 '11 at 23:08
1  
How did you get 260k users? The new /users page has 4 x 9 = 36 per page, and 9825 pages => 353k.. –  RichardTheKiwi Mar 1 '11 at 23:09
1  
I fully agree that 1 rep users with no activity in 30 days aren't users for the purpose of calculating this. –  Tim Post Mar 1 '11 at 23:10
1  
I better get to work on my SO rep so that I can one day be privileged enough to use that pick up line. –  Tim Stone Mar 1 '11 at 23:12
    
@Richard Thanks for catching that, fixed. –  Adam Davis Mar 1 '11 at 23:14
1  
I think there already is some sort of cut-off. My profile says 2%, and I am ranked at #780 on SO. –  ire_and_curses Mar 1 '11 at 23:26
    
@ire did you notice that it is top 2% "overall"? It appears to show you your "best statistics". For Jon Skeet and you, it's "overall". For me (new), it's "this month" –  RichardTheKiwi Mar 1 '11 at 23:32

4 Answers 4

up vote 5 down vote accepted

With a bit of investigation, I found that this is based directly off leagues. I don't read every single question on meta, so it is probably detailed somewhere.

http://stackexchange.com/leagues/1/month/stackoverflow
Search: Brock Adams
Result: #1713 month rank

User page http://stackoverflow.com/users/331508/brock-adams
shows: top 6% this month

Looking closely at the league page, there are 616 pages (50 each) of users reported, therefore #1713 / ~30800 => 6% (1dp)

So "non-users" are already excluded.

Given the stats (~30k users), 1% would include 300 people. A 0.01% tier asked for (in question) would end up with just 3 users, or effectively the medal band (gold/silver/bronze).


Original answer below

To second this, consider that at page 98 (1%), we have users at 27 rep for the week!

http://stackoverflow.com/users?page=98&tab=reputation

So by roughly making one answer a week, and getting 3 upvotes + 1 downvote (e.g. for copying another users's answer), congrats mate, you made top 1%!

share|improve this answer
    
If the weekly reputation was a moving window, rather than the start of the week at 0 GMT, this wouldn't be a problem. See meta.stackexchange.com/questions/81242/… –  Adam Davis Mar 1 '11 at 23:17
1  
But, more importantly, I think that new users should be explicitly included in the short term list calculation, as they should certainly show up on the page if they manage to get very high for a one week period. Being able to participate in the short term leagues is very important to encourage user participation. –  Adam Davis Mar 1 '11 at 23:19
    
@Polly - the question is not about the users page. Anyone should be able to see themselves. It's about the top x percent in the profile page (new change) –  RichardTheKiwi Mar 1 '11 at 23:23
    
@Polly I updated the answer - you ask for cutoff of 50 rep to get 80k users, well, it's already less than 80k users –  RichardTheKiwi Mar 1 '11 at 23:56
    
Thanks for looking into this more deeply than I had. Guess they already have a heuristic. –  Adam Davis Mar 2 '11 at 0:00

First @Richard is correct, it's the top X% of ranked users in the leagues, so not everyone. Only users with 200+ rep are even tracked, so we do remove most of the long tail here.

Also....

Alternately, introduce the top 0.1% and top 0.01% tiers.

That's a fine idea, it'll happen in a build today, here's what the new behavior will be:

  • Top 1% - 100%, show a whole number, e.g "top 40% this week"
  • Top 0% - 1%, show two decimal places, e.g. "top .23% this week"

We agree this would more strongly highlight the top-top users' accomplishments, it shall be done!

share|improve this answer

The cutoff mark should be higher than 50. My knowledge of programming is abysmal but, if I logged on Stack Overflow, I would have 101 reputation because of my activity on other SE sites.

share|improve this answer
1  
Good question - is "bonus" (100) considered in the weekly rep? See question here meta.stackexchange.com/questions/81096/… Jeff has fixed it bonus is out –  RichardTheKiwi Mar 1 '11 at 23:20
    
@Richardakacyberkiwi: Yes, but the base one reputation isn't. –  Borror0 Mar 1 '11 at 23:22

Percentage for this application is just the wrong tool. Use percentile instead.

share|improve this answer
1  
Lies, darn lies, and percentages! –  Adam Davis Mar 8 '11 at 21:48

You must log in to answer this question.

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