What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 127 Stack Exchange communities.

We are considering adding a new bronze tag-based badge.

Initially when we looked at the simple 100 upvotes on a tag to get the tag based bronze badge some ugly edge cases popped up:

For example: Thomas would get the [whycantyou] badge. And cletus already has the [big-o], [computer-science] and [complexity] badges from one answer.

I have nothing against those answers, I am not calling to make them wiki or anything, but it does show the system is wide open to gaming. Re-tag any of those questions and kaboom, you have a new badge.

We would like tag experts to get the tag based badges; we would like it to show long term commitment to questions with a particular tag.

Proposed changes, to protect against gaming and make the badge fairer

  1. Automatically take away tag based badges if the criteria is no longer met.
  2. Require a minimum number of non-wiki answers in the tag to get the tag-based badge (around 20 answers for bronze, 80 for silver, 200 for gold) AS well as a total score.
  3. Change it so we look at "sum score" as opposed to counting upvotes, so its in line with the tag "stats" page Fix the description on the website, to match the implementation that looks at sum(Score) where Score > 0 (though I think the where Score > 0 can probably be safely dropped without any adverse impact)

Thoughts?

share|improve this question
    
I didn't even consider that wiki answers could count for a tag-badge. Talk about FGITW in action to produce some easy-earned bragging rights! –  MPelletier Sep 2 '10 at 1:43
    
I think we exclude wiki now anyway. –  waffles Sep 2 '10 at 1:44
    
When you say "take away", do you mean a user's badge count would go down, or that they would have (possibly multiple) "no longer available" tag badges in their profile? –  Niall C. Sep 2 '10 at 1:51
1  
@waffles: Confirmation from Jeff et al. would be nice regarding wiki answers. –  MPelletier Sep 2 '10 at 1:54
    
@Niall: I think they must mean the count would go down. Because the current policy is to leave the count as is, but you have to "make up" the deficit before your count will climb any. –  dmckee Sep 2 '10 at 1:55
    
@Niall I'm suggesting deleting badges in this edge case, say you have a "car" badge and we retag it to "cars" your "car" badge would go away and be replaced with "cars". To stay consistent we would take away 10-20 badges from some real high rep users. But they all stand to get 30 or so bronze badges in their place, so I don't think it should be missed that much. –  waffles Sep 2 '10 at 1:56
    
@dmckee what deficit? tag based badges are only given once per tag per level. –  waffles Sep 2 '10 at 1:57
    
@MPelletier: question.PostTypeId = 1 and question.DeletionDate is null and answer.PostTypeId = 2 and answer.DeletionDate is null and answer.Score > 0 and answer.OwnerUserId is not null and answer.CommunityOwnedDate is null" –  waffles Sep 2 '10 at 2:03
    
@waffles: Isn't the current behavior 1) qualify for (say) bronze badge X making you count N 2) lose qualification for X: badge count is not reduced 3) qualify for another bronze badge Y: count does not go up (because you were one down) 4) qualify for bronze badge Z: count goes up (because you had squared away the deficit). Or am I confused? –  dmckee Sep 2 '10 at 2:08
    
@dmckee, can you pop in to the chat room: chat.meta.stackoverflow.com/questions/63148/… –  waffles Sep 2 '10 at 2:11
    
@waffles: So, question votes count too? I'm sorry, I don't know the SX schema by heart (yet!) –  MPelletier Sep 2 '10 at 2:42
    
@MPelletier question votes do not count for generalist and tag based badges. :) –  waffles Sep 2 '10 at 3:13
    
@waffles: thankfully. So, what was that bit of SQL? Do wiki answers count? –  MPelletier Sep 2 '10 at 3:21
    
@MPelletier its the "answer.CommunityOwnedDate is null" part, wiki answers do not count –  waffles Sep 2 '10 at 3:29
1  
Well, I finally got team confirmation that the tag stats score is the sum score of non-CW posts with positive score value. I feel content, now. –  Grace Note Sep 2 '10 at 12:55
show 2 more comments

4 Answers

Proposed changes - to protect against gaming and make the badge fairer.

  1. Automatically take away tag based badges if the criteria is no longer met.
  2. Require a minimum number of non-wiki answers in the tag to get the tag based badge (around 20 answers for bronze, 80 for silver, 200 for gold) AS well as a total score.
  3. Change it so we look at "sum score" as opposed to counting upvotes, so its in line with the tag "stats" page.

Feedback:

  1. 100% agree. But, don't remove the badge the second the criteria is not met. If you combine this with #3 someone could easily gain and lose a badge a few times with a slightly controversial answer. Give it a day or so before you recalculate.

  2. I like this idea, and I feel it may be the only new criteria that needs to change. This would make the badge very difficult to game without doing a whole bunch of retagging.

  3. I don't see a reason to make this change. The tag based badge system works using total upvotes (ignroing downvotes). I don't see how adding downvotes would help that much. If the only reason for the change is to add consistency with the stats page, then that makes sense. But I don't think this adds anything to the tag based badges. Well, it seems I have been lied to. #3 works for me.

share|improve this answer
    
Re: 3 see my edit :) –  waffles Sep 2 '10 at 2:23
add comment

Proposed changes - to protect against gaming and make the badge fairer.

  1. Automatically take away tag based badges if the criteria is no longer met.
  2. Require a minimum number of non-wiki answers in the tag to get the tag based badge (around 20 answers for bronze, 80 for silver, 200 for gold) AS well as a total score.
  3. Change it so we look at "sum score" as opposed to counting upvotes, so its in line with the tag "stats" page. Fix the description on the website, to match the implementation that looks at sum(Score) where Score > 0 (though I think the where Score > 0 can probably be safely dropped without any adverse impact)

This is pretty rare, but I agree completely with everything quoted above.

Going by the current tag badge ratios (votes per answer), the new bronze vote threshold will be 100, which I think is a good level as well.

With respect to #3, I would absolutely drop the Score > 0 clause. What's the point of counting downvotes if you're not going to count negative posts as well? It won't make a huge difference, no, but those posts should be included in the calculation, IMO.

share|improve this answer
    
So count all downvotes, essentially? –  MPelletier Sep 2 '10 at 3:18
add comment

I'd certainly like to see bronze tag badges. Currently the silver one (at 400 upvotes) is pretty hard to get for somebody who's a light participant in the site as a whole (yet focuses on specific tags).

The ratio between gold (1000) and silver (400) tag badges is only 2.5. If you keep the same ratio, then a bronze badge would be 160 upvotes. I would suggest increasing the ratio, to perhaps 5:

  • gold (1000)
  • silver (200)
  • bronze (40)

Maybe adjust the values a bit (say 2000/400/80), but keep the ratio consistent.

To address your specific points:

  1. Taking away badges is fine. Tags are a bit dynamic over time anyway, no problem there.
  2. Agree. A minimum threshold prevents the outliers from distributing spurious badges.
  3. I don't see much problem with only counting upvotes. Consistency would be good, but it doesn't bother me. Great answers get very few downvotes anyway.
share|improve this answer
2  
Adjusting from 1,000 to 2,000 is not "a bit." –  Jon Seigel Sep 2 '10 at 3:01
    
Ok, a "bit" might be 1200/300/75 with a ratio of 4. –  Greg Hewgill Sep 2 '10 at 3:14
5  
Actually: 1000 = 0b01111101000, and 2000 = 0b11111010000, so it really is adjusting a bit (the first one)! You're all programmers, you should know. –  MPelletier Sep 2 '10 at 3:16
1  
Exactly. It's a bit shift! –  Greg Hewgill Sep 2 '10 at 3:29
1  
I agree with everything you have here, except the presumption that the progression from bronze, silver, and gold should be linear. For silver and gold, 400 and 1000 seem about right (and I only have two silver and no gold tag badges). Most bronze badges are easy to get, and 100 seems fine to me: the idea is to reward early participation to encourage prolonged participation. –  Gnome Sep 2 '10 at 5:48
    
What I'm proposing is geometric progression, not linear. Linear would be 600/400/200 for example. –  Greg Hewgill Sep 2 '10 at 7:52
add comment

I think that tag based badges should also reflect accepted answers. Currently these badges reflect only number of upvotes (and some minimum number of answers). My proposal is to create second type of tag badges (like "activity" badges) which would reflect how many accepted answers per tag did you have.

Other alternate can be reworking current tag badges and instead of upvotes use reputation.

  • +10 per upvote
  • +15 per accepted answer

Where bronze requires 1000 rep., silver requires 4000 rep. and gold requires 10000 rep. But imo this alternative is worse.

Or maybe instead of badge some statistics could be nice.

share|improve this answer
    
Votes received after the rep cap still count to the badge totals. So if you get an answer than gets > 20 votes in a day you'll be that much closer to the badge. –  ChrisF Jan 8 '11 at 15:37
    
Yes but if you are active in a tag which is not visited by so many users you will get much less upvotes. My first proposal is about showing who provides solutions (accepted answers) for each tag. –  Ladislav Mrnka Jan 8 '11 at 16:30
add comment

You must log in to answer this question.

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