What information should we show with "usercards"?

We’ve had lots of discussions pro/con about a single reputation number. This post is not about that.

We’ve also talked about showing more contextual information about users. For example, for answers, we could indicate the user’s “tag depth”, some indication of applicable expertise. Maybe we’ll have and show tag badges; haven’t gotten that far. The idea seems interesting but vague. (Yes, I can admit to my own vagueness. :slight_smile: )

For MVP, I was thinking we might do something more informative than “rep” and less complicated than contextual stats. How do y’all feel about showing, with the user name/avatar, something like this?

12 questions, 94 votes
28 answers, 542 votes

(This could be abbreviated to “Q” and “A” with full text in a tooltip.)

Vote totals would be net, a straight upvotes minus downvotes. That’s not the same as answer score, but that’s ok. We’re not trying to show score or reputation; we’re trying to give a sense of the user’s contributions.



I presume these are net votes.

My pet peeve on Stack Exchange in the last couple of years was users with highly negative net vote count who none the less accumulate significant net rep, so I don’t want to see only positive votes represented there. I am open to putting a floor at zero, however.

But basically nice. It not too much to read and interpret but it carries a lot more information than a single rep number.


Net votes, yes. See my last paragraph.

Good point. We don’t need to embarrass people with “-79 votes” or the like; just omit it if it’s not positive. Regulars will know what the omission means, but we won’t announce it to visitors.

1 Like

That is, IMHO, a fault of an inconsistent Reputation system. The SE reputation number credits upvotes and downvotes differently - +10 for upvote, -2 for downvote. 5 upvotes, 10 downvotes = net 30 points. With a straight net vote shown, it would be -5 votes. Very different.


No! If someone really ends up with -79 net votes then either they are being targeted by others (which is a moderation issue) or they are truly providing poor quality questions & answers. That’s life.

0 or not shown implies “new” or “no activity of any significance”. -79 is significant.


Well, I imagine that regulars seeing

132 questions
16 answer, 2 votes

or something similar would have a pretty keen idea there is a significant problem with this poster’s question history without having to show negative values for a few days in the case of newbies who get off to a rocky start.

Not that I’m welded to the notion of putting a floor on there, but it’s in keeping with my notion that new comers are put off in part by the accumulation of little annoyances.


The flip side is that for some people who might otherwise cause trouble, knowing that “everyone” will see their big negative score will be a deterrent. On the other hand, if they are really intent on causing trouble, that really won’t matter - they’ll post nasty stuff that gets downvoted until they get banned.

1 Like

I like the idea of having more contextual information. If I’m reading an answer, it can be good to know if whoever’s written the answer knows what they’re talking about on subjects relevant to the answer and this is considerably more relevant on an individual q/a level than ‘total number of questions, answers and votes’, which doesn’t help me in determining if they’re good at [this tag]. If I’m new to the site or tag, I don’t even know how frequently people vote (there are some tags where users vote a lot and others where they barely vote at all), so even showing ‘tag score’ isn’t useful unless you have enough experience with that tag to know what a ‘good score’ would be.

What I think would be a good solution would be to show that x% of [user’s] questions/answers are above [score threshold], or something similar (like ‘so many standard deviations above/below average’) as this pretty quickly tells me if they know what they’re on about. As I wouldn’t want to bias voting based on this, I’d suggest putting this info in a ‘hovercard’ (or whatever I should call it) so that users don’t just go ‘oh they must be right, so I’ll just upvote without reading’. Perhaps alongside the number of questions/answers they’ve written in the interest of fairness. I’d further suggest that the score threshold be user dependent so that we can differentiate between people who e.g. consistently get ~50%, ~70% and ~90% score thresholds as saying ‘90% of posts above 90% score’ and ‘90% of posts above 70% score’ are different things.

Perhaps an alternative to the above would be the same, only instead of using score thresholds, use the user’s average score.

Outside of the context of questions and answers, it’s useful to know other things (are they good at flagging?, are they good at editing? etc.) but the only way I can think of having this sort of information without having an overly big usercard (that takes up too much space) would be to have badges.

So, my (current) solution to this would be to have a usercard with user-selected badges (and rank level?) displayed, that on hovering-over brings up a bigger version of the usercard that displays the number of questions and answers that user has posted both on that site and in that tag, alongside either the percentage of posts they have on the site and tag above some (user-dependent) score threshold or the average score they have on the site and tag.


This sounds like a job for an adaptation of academia’s h-index. Someone has an h-index of 5 if they have at least 5 papers that have been cited at least 5 times each; the h-index is the highest such number. They might have 12 papers that have been cited 5 times each, or 5 papers that have been cited 8 times each, but if they don’t have 6 papers with 6 citations each then they don’t qualify for an h-index of 6.

In this case, you could define a tag p-score as being the largest percentage such that at least p% of their answers have a Wilson score that is higher than at least p% of answers in that tag. Someone with 70% of their posts being in the top 30% would have a p-score of 70, if less than 71% of their posts were in the top 29%.

Typically, within a given field h-indexes are fairly consistent, but between fields they aren’t really comparable. Here that would be less of a problem, since we’re normalizing to tag-specific distributions anyway.


The basic concept is OK, but a few tweaks would help:

  1. The "xxx votes" is misleading. At first glance that means the total number of votes cast, not the accumulated ± score. Something like:
    28 answers, +542

    By deliberately putting the “+” sign there, it becomes obvious that this is a combined score, and also points out that the value could go negative.

    If someone posts enough bad posts and ends up with a negative score, then it should be shown. It is useful information to others. If you only have two or three posts, then a negative vote total isn’t that meaningful, and people will understand that. If you’re at 20 posts with a vote total of -5, that’s significant, and everyone else has a right to know that.

  2. Only one of those lines should be displayed, depending on context. For a question, show the first line. For an answer show the second line.

Like it not, these things are linked. There is limited space and clutter-tolerance next to a user’s name, so it needs to be used carefully and serve multiple purposes. One purpose is to give a general idea of the content trustworthiness of a user. Another is to give an idea of the level of contributions. The third that you haven’t mentioned is to give some reward and public recognition to the core active users providing lots of content. And yes, that’s basically a “score”.

These things need not be at odds with each other. The vote total on all answers of a user can serve the score purpose too.


These are interesting ideas that we should explore, probably with some user testing. We have to trade off completeness and complexity – what’s helpful, what’s confusing, what’s too hard to explain, what’s intuitive? I’d like to do something straightforward for MVP and keep options open for richer “user stats” in later versions.


Please keep in mind that within young communities users will not have a lot of interesting things to show for in their user cards. The taxonomies in young sites are also very under-developed. (SE actually only hands out tag batches when a tag becomes viable (which IIRC is at 100 questions).)

Therefore this might not be a very helpful MVP feature at all.
So I’d advocate for not having any such information there in MVP.

  • Regarding the actual feature I highly advocate against using any tag based mechanism in the beginning but start with an overall question/answer-based mechanism. That would remove the “tag viability” issue from the first iteration of the feature.
  • I also caution about aggregating votes in a user card. If we do that we have introduced reputation via a backdoor. We also then implicitly set the value of a vote to one. (Which would be even worse than setting it to ten like SE because people like their rep to be high. :wink: )

Thus, I’d suggest for the user card a mechanism like @tuggyne suggests. I’m not quite sure how well this will play out for small sites and young sites. I guess one could dump a smaller SE site into a DB and check what those values would be and how that would look. (Or just user script it simply over SE; the data is all publicly available.)


I created a user script for SE.
It displays a user’s question & answer count instead or rep or badges.

Yes it’s ugly. The goal is to verify the idea of this thread vs. a real data foundation.
If you want me to add any easily reachable data from SE’s user just let me know or create a PR.

The second one queries also the API to get some more data. It will run into the API rate-limiting at some point though.

Neither of those are meant as a great SE user script.
They are meant that we can verify theories of things we might do differently.

Scraping user profile only


Scraping user profile + API usage


P.S. There’s no paging on the API in the second one. So it queries only up to 100 posts of the author which is inevitably wrong for 100+ post people. This affects the lower two lines of stats.


We need some system that provides the advantages of rep on SE. It draws out more contributions from top users by fostering some competition, it gives new users something to aspire to, and most importantly, it provides public recognition for those that spend a lot of volunteer time adding value to the site that is appreciated by others. I don’t know why “rep” seems to have gotten a bad rep around here. Call it what you want, but we still need something like it.

It seems that displaying the total vote tally would serve multiple purposes, with “rep” being one of them.


Not trying to discuss this again. I’m just saying as long as the decision regarding “not showing rep” stands we shouldn’t backdoor the decision ourselves. :wink:


I have not seen any such decision. Perhaps this was something that happened before I started participating here. A rep-like score is going to be important for some types of sites. If this was discussed early on, then quite possibly representatives of those kinds of sites were not here. It would be a shame to not support the kind of sites where rep is important just because nobody was here early on that participated in such a site.

1 Like

It could have something like the “top 2% this quarter” that SE shows in profiles.

That doesn’t show that you’ve built up a long-term history of significant contributions. While the last quarter is one metric to possibly show on a user’s profile, it isn’t the one parameter that should be shown about a user when that’s all you get.

For the record, this is the discussion (and some links in it are also interesting for that topic):

I like this idea, but wonder if it would worth incorporating an “age” factor. Old posts tend to have higher vote totals than new posts, but that doesn’t indicate that old posts are better, since (a) they have had more time to collect votes and (b) quality standards and voting patterns may have shifted. An obvious example of this is someone who posts a number of decent questions and answers early in a beta when standards are relatively low, collects a lot of votes, and then stops posting.

So if we go this direction, I’d at least suggest that the default, publicly visible number should weight newer posts more heavily than older ones, or even, for simplicity, exclude posts older than a year or two.