168

When you get reputation, you get a +X green spot on your reputation feed telling you that you gained reputation and if you drop down the menu, you get more detail.

How about we make it that when you lose more than you got since the last time you checked, you see it red with a little -X.

For example, let's say you've just looked at the reputation bar so any old rep notification is gone and you ask a new question. 3 users downvote your question and 1 upvotes it. You will see it green because you gained 5 rep but you should see it red because you are at -1 rep.

This would help people improve their questions / answer by seeing, as soon as they get on the site, that they got negative feedback and that something needs to be checked / changed.


It shows here so it could very well be applied to the notification count:

11
  • 1
    I think this is on purpose. Losing rep is not in the live updates included.
    – juergen d
    Commented Nov 25, 2013 at 14:28
  • 1
    @juergend It is on purpose yes.It's a feature request. Since it shows in the feed (see update) It should be added to the notification. Commented Nov 25, 2013 at 14:30
  • 6
    The live update feature is there to show you the good stuff that happened to your account (rep gains, badges and notifications).
    – Oded StaffMod
    Commented Nov 25, 2013 at 14:32
  • 84
    @Oded - I would be far more concerned about being notified immediately about downvotes then I would about an upvote. If downvotes are supposed to signal something is wrong, shouldn't that take priority?
    – JDB
    Commented Nov 25, 2013 at 14:43
  • 21
    @JDB Exactly I didn't see that section as a reward section but more as a "Ok this is what happened recently in your account" Good or bad. Commented Nov 25, 2013 at 14:53
  • 17
    @Oded I just confirmed today that the new live update i.e. top bar achievement is reducing: it was +5, and when I downvoted an answer it was reduced to +4. I'm 99% sure that if I would have downvoted 5 more answers, it would just cause the number to disappear rather than show negative. Bottom line: it's already showing reduce in points so why not let it show negative as well? Commented Nov 28, 2013 at 14:17
  • 2
    @ShaWizDowArd Yes I've noticed this to ! It's partly implemented so we should really move forward with this. Commented Nov 28, 2013 at 14:18
  • 9
    I think one exception to this should be the -1 when I downvote an answer. Yeah, my rep is changing, but I don't need to be notified about something I initiated. This is in contrast to when other people downvote me, which is an action that I otherwise might not notice.
    – Troyen
    Commented Feb 6, 2014 at 9:57
  • 3
    @Troyen but it tells you about the +2 for accepting an answer already, so it should tell you either both or neither if the guideline is "things I initiated" (also bounties). I'd rather it just tell me everything instead of trying to decide for me; I can just ignore it until the number changes to something I didn't do, or click it if it's bugging me. Commented Apr 14, 2014 at 21:04
  • 1
    Buying yourself a gold Great Question badge, eh? ;) (I'm not really complaining. It's gotten me 2 votes today already)
    – mhlester
    Commented Jun 18, 2014 at 22:13
  • @Equality In Tech It definitely should be. Downvotes are meant to warn that your answer is either poorly formatted or (dangerously) wrong.
    – user387578
    Commented May 4, 2018 at 4:50

5 Answers 5

104
+50

I too would like to see this implemented. @Oded's response to a different question lays out the SE reasoning:

We want to let you know about good things as soon as they happen - that's what our real-time notifications are for.

Negative reputation changes are... well... negative. So we are not in a hurry to let you know about them.


I strongly disagree, and all the comments and upvotes there and here appear to as well. I would much rather see notification of downvotes than upvotes. I get it - the drop-down is called achievements. But reputation losses do show up inside the "achievements" drop-down, so it wouldn't be unprecedented to highlight them where we can see them without having to click.

  • An upvote is (should be) the status quo. It's a little pat on the back.
  • A downvote means something's wrong. Maybe I wasn't clear in my post. Or maybe I misread the question. Often, this should be an action item.

Unless someone comments (which we all know isn't mandatory), I may leave an innocent questioner out in the cold, or otherwise miss an opportunity to help.

Is the purpose of the achievements display to stoke egos, or to make the site better?

If this feature request would be a major shift in the definition of "achievements", then so be it. A response of "things are the way they are because that's how they are" isn't a compelling reason not to change it.


If, and this is a huge if, the fear is that new users will be crestfallen to see notification of downvotes, then enable this only for users with the "Established User", "Vote Down", or "View Close Votes" privilege or something. But I again feel this is valuable and pertinent information for everybody, regardless of what sentiment it brings. Besides, brand new users only start with 1 rep to begin with, can't lose rep, and as such won't see this display.

bar


Anecdote

This happened to me just yesterday, and I think it perfectly articulates the value to seeing lost reputation:

  • I answered a fairly standard question here. Nothing to write home about. There were no other answers, I don't have experience with the framework OP was asking about, but I knew what the error meant, so explained what was happening. OP asked for some clarification, I responded to the best of my ability, and walked away.
  • Presumably my explanation wasn't sufficient, so my answer was voted down. I didn't notice this for five hours.
  • On finally realizing my answer didn't yet help him, I did a couple minutes of research, found the solution, and edited my answer.
  • That was exactly what he needed, and the downvote was converted to an upvote.

If I'd seen a notification of the downvote, I could have helped him earlier. At the end of the day, that's what this is all about.


Questions get a mix of upvotes and downvotes all the time. But when an answer gets a downvote, it can often be improved, and will be improved sooner with this display.


The second highest voted of all time, shows clearly the possibility that downvotes prompt users to improve their posts.

4
  • Example how asking on meta may help is here: meta.stackexchange.com/q/197949/225745 question got fixed enough to go from -9 to +2 because asker noticed it is going wrong and went to meta for help.
    – Mołot
    Commented Feb 27, 2014 at 10:42
  • See my answer posted about 10 minutes ago. That's all I was able to figure so far, I'll expand it when/if I'll be able to.
    – Mołot
    Commented Mar 4, 2014 at 10:02
  • 8
    On enabling this only at certain rep levels, I think that's actually backwards -- a new user whose first post out of the gate gets downvoted should get told as soon as possible that he needs to fix something, before he comes back hours later to find it massively downvoted and gives up. A new user's first DV is a teaching moment, if he notices early and is willing to be taught. (If he's not willing to be taught he'll be gone soon anyway.) Commented Jun 18, 2014 at 16:01
  • 1
    I'd love for it to be made available for everyone. Just trying to placate the SE team's fear of showing negative events to users.
    – mhlester
    Commented Jun 18, 2014 at 16:03
27

Some statistical data:

There are 2152347 downvoted posts:

select count(distinct down.PostId) from Votes as down
where
  down.VoteTypeId = 3

downvoted -> upvoted

And there are 440241 posts that earned upvotes after they earned downvotes:

Select count(distinct down.PostId) from Votes as down
join Votes as up on
  up.PostId = down.PostId and
  up.CreationDate > down.CreationDate
where
  up.VoteTypeId = 2 and
  down.VoteTypeId = 3

It means a bit over 20% of downvoted posts got improved, explained in comments or was found not that bad after they got downvoted.

downvoted -> edited -> upvoted

More specifically, there are 148979 posts that got downvoted, edited and upvoted, in that order.

Select count(distinct down.PostId) from Votes as down
join PostHistory on
  down.PostId = PostHistory.PostId and
  PostHistory.CreationDate > down.CreationDate
join Votes as up on
  up.PostId = down.PostId and
  up.CreationDate > PostHistory.CreationDate
where
  up.VoteTypeId = 2 and
  down.VoteTypeId = 3 and
  PostHistory.PostHistoryTypeId IN (4, 5, 6)

This gives almost 7% of all downvoted posts.

downvoted -> edited, positive score

There are 168593 posts that was downvoted, then edited, and now have a positive score. Or 292718 with non-negative score (positive or 0).

Select count(distinct down.PostId) from Votes as down
join Posts on
  down.PostId = Posts.ID
join PostHistory on
  down.PostId = PostHistory.PostId and
  PostHistory.CreationDate > down.CreationDate
where
  Posts.Score >= 0 and
  down.VoteTypeId = 3 and
  PostHistory.PostHistoryTypeId IN (4, 5, 6)

This gives almost 8%, or near to 14% of posts that (probably) was bad but now are not so.

downvoted -> commented

There are 94875 posts that was downvoted -> commented -> edited -> updated.

But in order: 778279 posts was commented after downvote. 318912 of them was edited after that comment. This means, to the best I can guess, that on over 40% of posts that generated notification after downvote an attempt to make them better happened. On 12% of downvoted->commented ~= 30% of downvoted -> commented -> edited it succeeded, at least partially, giving follow-up upvote.

Conclusions

Now, it is a matter of interpretation if 7% or 20% means something, but for me it is a strong sign that downvoted posts really gets improved, in non-negligible amounts.

Links to queries:

7
  • Of course I'm open for query improvement. For now I can't find how to detect downvote -> edit -> un-downvote for example...
    – Mołot
    Commented Mar 4, 2014 at 9:59
  • This is awesome. Could you please do one more query? Downvote -> comment -> edit -> upvote? Since a comment creates a notification, this will show how many people normally edit, vs. how many edit after a comment (please compare with commented + downvoted posts that weren't edited to get an idea of how many bother after notification, which is what this request would most closely mimic)
    – jmac
    Commented Mar 4, 2014 at 13:11
  • @jmac here you go
    – Mołot
    Commented Mar 4, 2014 at 13:26
  • Thanks Molot! Give me some time to mull over the numbers tomorrow. You're an SQL genius!
    – jmac
    Commented Mar 4, 2014 at 14:01
  • @jmac naah, not a genius, these queries are pretty simple - and probably could use some fine-tuning, too. By someone who is more familiar with data structure of SO.
    – Mołot
    Commented Mar 4, 2014 at 14:10
  • Bravo! This is very informative.
    – mhlester
    Commented Mar 4, 2014 at 16:24
  • I edited my answer based on your queries (slightly modified to include several cases). I may have goofed in my SQL, so you may want to take a look at it if you think I made some horrible error I didn't realize.
    – jmac
    Commented Mar 6, 2014 at 1:50
4

Based on Molot's excellent queries I took a stab at looking at the potential benefit of this change.

I used Molot's queries to look at all posts made in 2014 to look more closely at how downvotes and being informed of downvotes affects behavior. I combined his queries to look at:

  1. How many posts are downvoted
  2. How many of those posts are then edited by the user
  3. How many of those edits are successful
  4. Whether commenting (notifying the user) impacts edit frequency or edit effectiveness

26.2% of Downvoted Posts are Edited by the Owner

There were 75.2k downvoted posts in 2014. Of those 75.2k, 19.7k (26.2%) were edited by the owner after the downvote.

Only 10.6% of Edits are Successful

Of the 19.7k posts that were edited by the owner, only 2,097 (10.6%) were upvoted after the edit.

Commented Posts are Edited More Often

Of the 75.2k posts, 55.2k received comments, 20.0k didn't. Of the 55.2k that received comments, 15.1k (27.2%) were edited. Of the 20.0k that didn't receive comments, 4.6k (23.0%) were edited.

Non-Commented Posts had More Successful Edits

Of the 15.1k posts which received comments and were edited by the owner, 1.5k (9.9%) received upvotes after the edit. Of the 4.6k which didn't receive comments and were edited by the owner, 602 (13.1%) received upvotes after the edit.

Summary

People learning about downvotes seems to increase the probability they will try to do something about it, but decrease the probability that it actually works. This pattern holds for other time periods as well (I checked some dates in 2013 as well, similar pattern).

My assumption is that people who care about quality will edit their posts regardless of if they get a distinct notification or not. There isn't much sense in bringing people's attention to the problem if we can't become more successful at teaching them how to improve it. We are definitely commenting on posts with downvotes at a good clip, but we aren't turning those comments in to successful edits better than people who take it in to their own hands to edit after a downvote in the first place.

Adding a notification of negative reputation doesn't look like it will solve the problem.


Absolutely not.

What benefit does this provide? There isn't anything you can do about downvotes. Sometimes they just happen. Making it more visible when it happens is only going to increase discontent, without increasing the quality of content.

We already get loads of complaints on meta about 'undeserved' downvotes. This is not limited to low-rep users. If people start getting real-time updates every time they get downvoted, it will only compound the problem.

21
  • 21
    "There isn't anything you can do about downvotes"?? I don't think that's true at all, by a long shot. Downvotes usually signal that something needs to be fixed. Not always, but often.
    – Kerrek SB
    Commented Feb 26, 2014 at 9:32
  • 11
    If someone gets banned, we tell them to edit to improve quality, to get some downwotes retracted or earn some upvotes. So either we are consistently lying to q-banned and a-banned people, or "There isn't anything you can do about downvotes" is simply not true.
    – Mołot
    Commented Feb 26, 2014 at 10:15
  • 1
    I cannot think of a single time I noticed downvotes in my little rep area and thought, "Golly gee, that's something I could fix", because generally I put thought in to what I'm posting and there isn't much I can do about it. What exactly makes you think that having a -2 up there will actually make others change their content exactly?
    – jmac
    Commented Feb 26, 2014 at 13:12
  • 10
    By the same reasoning, what benefit does displaying positive reputation provide? There isn't anything you can do about upvotes. Sometimes they just happen.
    – Stijn
    Commented Feb 26, 2014 at 13:18
  • 4
    The benefit for me personally is a downvote usually indicates there's something wrong with my answer, and when there's no comment I might be able to look at competing answers to see where my mistake is and then fix or remove my answer.
    – Stijn
    Commented Feb 26, 2014 at 13:20
  • 1
    @Stijn, how often do you see people complaining about getting upvotes on meta? Want to remove it? Make a separate feature-request. Don't think it will help anything, but it may not hurt that much either. But there will certainly be negative implications for showing all downvotes as even when it isn't shown people still complain about them. If you are that concerned about getting downvotes, just double-check to make sure your answers are right.
    – jmac
    Commented Feb 26, 2014 at 14:02
  • @mhlester, the meta post has quoted comments (that have been deleted) by the asker, who has a rep of over 7k. Here is one: "@AaronBrager I have been on this site for a while. It used to be fun, now it is turning into a police state. I have looked at existing answers and have question how many different ways this problem can be solved, that's why I asked the question. I don't see why it's important for me to post my solution when that can potentially influence others. The rampant downvoting on this site is total nonsense, SO should penalize people for doing it."
    – jmac
    Commented Feb 26, 2014 at 23:43
  • If someone is angry for getting a lot of downvotes, hopefully he would be less angry if he could notice it after the first one, and ask people why, providing his rationalization. Next reader then could (and should) try to explain instead of downvote, and if anything goes well, question should be fixed before it's drowned. But on the other hand people complaining on meta is a good thing, it gives us a better tool to explain them what's wrong with their approach and their questions, or use "meta effect" to offset downvotes with our upvotes if we think their question is downvoted unfairly.
    – Mołot
    Commented Feb 27, 2014 at 10:08
  • 1
    @Mołot, only we don't want to require people to give a reason for their downvotes. So giving people a reason to make angry comments demanding explanation doesn't seem to be desired behavior. Nor does complaining on meta. There are a billion resources available explaining why downvotes happen, and how to do your best to prevent them, and pointing out that quality will virtually always win out in the long run. What does pointing out short-term negative votes provide as a benefit?
    – jmac
    Commented Feb 27, 2014 at 10:27
  • @jmac "we don't want to require people to give a reason for their downvotes" - that's exactly why we want questions on meta, where anyone can help unfortunate user, without being in any way connected to that downvote.
    – Mołot
    Commented Feb 27, 2014 at 10:28
  • @Mołot, if the person posted a quality question/answer that got downvoted, the answer is only, "It just happens some times. Long term you'll end up in the positive". If the person isn't posting quality content, I haven't seen a meta post end up with a positive result (though obviously I don't track every post complaining about downvotes). Can you provide some evidence that posting on meta leads to higher quality questions? That's the only reason this should be implemented.
    – jmac
    Commented Feb 27, 2014 at 10:36
  • meta.stackexchange.com/q/197949/225745
    – Mołot
    Commented Feb 27, 2014 at 10:39
  • @Mołot, you're saying that because a post that gets -8 can't be seen, we should inform people when they get -1? This seems like a poor solution if that's the problem you're trying to solve...
    – jmac
    Commented Feb 27, 2014 at 12:38
  • I'm saying that because asking for help on meta helped to turn -9 question into +2 one, your point that asking on meta is bad thing is not always valid. You asked for the proof that asking on meta may help to make content better. You got it. That's all.
    – Mołot
    Commented Feb 27, 2014 at 12:41
  • 1
    @jmac, thanks for the extra analysis! I'm curious though: deleted comments don't show up in this query. That almost certainly distorts the statistics towards fixes coming "without notification" don't you think?
    – mhlester
    Commented Mar 6, 2014 at 5:13
1
+50

I don't really have anything new to add here but am posting an answer since I'm formally declining this; jmac and Shog cover all the reasoning against doing this pretty well. It's just too hard to see the number of people motivated to improve their posts outweighing the number of people who will be upset by every downvote or want to start an argument about each one.

2
  • 5
    Not the answer I wanted to read, but thanks nonetheless.
    – Stijn
    Commented Sep 26, 2014 at 7:57
  • 1
    People are going to find out about downvotes anyway. They may as well find out sooner so they can improve quicker.
    – Picachieu
    Commented Jan 15, 2019 at 4:30
-2

This would help people improve their questions / answer by seeing, as soon as they get on the site, that they got negative feedback and that something needs to be checked / changed.

And how exactly does this help them improve their work? Sufficiently self-critical folks presumably already know when they post something half-assed; they don't need the reminder. Insufficiently critical authors are probably not going to have an epiphany in response to seeing "-2" pop up. I think it's only fair to show these rep changes in the dropdown itself - folks don't like seeing weird inconsistencies in their reputation - but there's no particular need to rub folks' faces in it.

Wanna help someone improve their work? Leave a constructive comment. They'll get a top bar notification for that already, and almost certainly have an easier time figuring out what to do in response to it.

I agree with jmac - this stands a much better chance of increasing pointless whining than it does improving the quality of posts.

That said, if you personally know how to respond constructively to downvotes and are obsessive enough to want to do so as quickly as possible after they come in... There is a work-around: just get upvotes on a regular basis, such that when you get a downvote you see a +8 / +3 rather than +10 / +5.

11
  • 25
    Misunderstanding a question does not mean my answer is "half-assed." Improvement shouldn't rely on the chance observation of odd upvote scores. Is it really so uncommon to treat downvotes like an adult and try to fix one's post instead of whining? For every complaint on meta, I bet there are countless cases like the anecdote from my post here. You just don't see them.
    – mhlester
    Commented Mar 4, 2014 at 2:41
  • 1
    @mhlester, you seem to believe, quite strongly, that there is this magical group of people who, upon receiving a downvote (and recognizing it) are rushing to improve the quality of their posts. You have the data explorer, why not put together some stats rather than a single piece of anecdotal evidence? My issue is that you want to display reputation with the goal of quality improvement, but haven't actually shown this is an effective way to do it. It would take me about 5 minutes to find 20 posts complaining about downvotes on meta. Convince us of the upside!
    – jmac
    Commented Mar 4, 2014 at 3:36
  • @jmac, you're right. More evidence is needed. It felt strong to me having witnessed it firsthand (I'm magical!), but I accept your skepticism. Would downvote, edit, undownvote numbers be convincing? I'm not the right person to put together the SEDE query, but I'll give it my best shot.
    – mhlester
    Commented Mar 4, 2014 at 4:20
  • @mhlester, if it were me, I would want to show: 1) with notification, people are more likely to edit, 2) that the ratio of edits is significant compared to quantity downvoted posts, 3) post-edit, the downvotes are actively removed, 4) based on that data, if showing negative rep gains is the best way of accomplishing it. My hunch is that if you want people to edit downvoted posts, showing negative rep is much less useful than a comment. And even with a comment, my hunch is that less than 20% of posts get edited. Negative rep indicator is going to be weaker than a comment, so the effect is small
    – jmac
    Commented Mar 4, 2014 at 4:58
  • So query 1) checks for ratio of edits on downvoted posts vs. downvoted and commented, 2) ratio of downvoted posts to downvoted and edited posts, 3) ratio of downvotes removed post-edit. I would limit to recent lower-voted posts to minimize noise (getting 8 downvotes and 10 upvotes will make -rep far less noticeable than getting 2 downvotes and 2 upvotes, for instance).
    – jmac
    Commented Mar 4, 2014 at 5:00
  • Ok, I'm throwing in the towel. Those queries are over my head. If you feel my anecdote is such the outlier, I don't imagine I'll be able to convince you otherwise. Just know it happened to me by chance only two days after my post here. It's really quite the coincidence...
    – mhlester
    Commented Mar 4, 2014 at 5:42
  • I'd also like to apologize if my comments were overly caustic. I do understand your points and take them to heart. As highly requested as this feature is, I do appreciate the concerns you've raised, and realize motives vary greatly between individuals, on both sides of this aisle.
    – mhlester
    Commented Mar 4, 2014 at 6:01
  • 1
    @jmac under your own answer I posted a proof that there are people who do just that - asks how to recover. So maybe you would care to use data explorer and prove this group is small enough to ignore them? Because maybe this is a group of two, but it exists, and you seem to argue it does not...
    – Mołot
    Commented Mar 4, 2014 at 7:22
  • @Molot, let me be blunt. Shog's opinion has a fair bit of weight. Currently he is leaning toward not supporting this feature-request. You need to convince him, and the anecdotal evidence provided so far hasn't managed to do that. Shifting the burden of proof to me isn't going to help you in the slightest.
    – jmac
    Commented Mar 4, 2014 at 8:09
  • @mhlester, I've got a pretty open mind if you can make a convincing argument that shows me the benefit outweighs the cost (and I am quite familiar with the cost as I spend quite a bit of time on meta and see 'why was my post downvoted?' questions come up very regularly). Explaining why you want it is fine, but we have hundreds of thousands of people using this site, and without proving this is a significant problem, and that this will solve that problem, chances are that people won't act. Hence the request for data. (Don't worry about your comments, I didn't find them harsh)
    – jmac
    Commented Mar 4, 2014 at 8:13
  • 2
    @MaskedMan The downvotes help other people with the same (or similar problem). They're an indication that what was written is or is not useful.
    – Zev Spitz
    Commented Nov 12, 2019 at 14:08

You must log in to answer this question.

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