It seems that if you look at the column on the right of the page, you can see upvotes and downvotes separately for recent posts. The same [n, m] format is displayed for recent comments, but it doesn't seem to actually sync with the score displaying on the comment. This feature only seems available on the sidebar: looking at the actual comment or post doesn't give you this information.
Thanks, whoever did this!
Awesome! That was one of the most asked-for feature!
And it is implemented on comments too, as a title text: Mousover the karma score of a post or comment and a little popup will give you the percentage of positive votes!
(a side effect of this is that if we see a comment at 0, we can't immediatly differentiate 1 upvote and 1 downvote from 50 upvotes and 50 downvotes, but that's pretty minor)
Many thanks to the trike apps people, and whatever volunteers helped implement, test and deploy this!
Edit: This specific fix appears to be thanks to Wesley Moore.
And for comparison - whatever you do, don't vote on this comment.
(note: this has been voted on. Go find your own non-voted comment somewhere. Sheesh)
This is a great new feature. Thanks!
I will note, though, that the implementation of this feature seems to have artificially inflated my karma score. It has gone up by about 200 points over the last couple of days, and I don't see any sudden increase in actual upvotes to account for this. Also, my karma over the last 30 days doesn't display a commensurate increase.
Here's my hypothesis about what happened: I had a post that I moved from Discussion to Main after it had already accrued a number of upvotes. The upvotes it got while in Discussion only gave me 1 karma point each, but I suspect something about this new feature has retroactively scored those upvotes as if they were for an article in Main, so I got an additional 9 points for each of them.
I'm not complaining, of course. Just a heads up about what seems like an unintentional side effect.
On the other hand, my 30-day karma shot up by over 50, independent of my total karma. I'd been puzzling over this, because there's never been a point in time when I had enough downvotes all at once to account for that sort of leap when they passed beyond the 30-day threshold.
I'd noticed a ~70 karma uptick and didn't know where it came from. Thanks for shedding light on that.
Aha. That explains why Eliezer now has ~4200 more "karma in the past 30 days" than he did.
Edit: Never mind. The effect is now gone, which means it was actually this change.
Same thing happened to me, and I also had moved an article from Discussion to Main after it had gotten a lot of upvotes. So that's almost certainly the explanation.
I noticed today that hovering over a user's total karma score displays the percentage of positive votes; didn't realize it was in fact a new feature until I saw this post. It came as something of a surprise to me to learn that my karma of 53 is in fact 56% positive, which implies a total of around 440 votes. I wasn't aware of being this controversial.
I wonder if displaying the total positive and negative votes instead wouldn't be more intuitive, though; especially with comments. As Emile pointed out, it would also convey more information.
Edit: I thought there might be something fishy about this...
It's very useful feedback. I have 82% positive. Going through my old comments, I found that a lot of comments I've made that I thought would be controversial actually had 100%. The comments that had a low percentage tended to be the ones where I hadn't expressed myself well. Given that I have a lot of unorthodox views, I found this reassuring.
It appears to me that 30 day karma has switched back to votes received in the past 30 days from votes on material posted in the past 30 days. Why? The original switch was very soon after the introduction of the feature.
I noticed this because people who haven't posted in months have 30 day karma. I then tested it by voting on an old comment.
You're correct, John deliberately changed this as he thought it was a bug. I'll chat to some people to work out what the desired behaviour is.
Previous behavior was chosen intentionally, see this thread for the discussion. I think the change should be reverted.
Trike Apps is also working on a solution for the LW wiki spam, since captchas have not been sufficient.
FWIW, I just checked what the RationalWiki IP-edit captcha does: we're using QuestyCaptcha, which asks "What is the nth letter of our name under the logo?" (said name being of course "RationalWiki") - and this (a) utterly confuses the spambots (b) doesn't hamper accessibility.
Edit: Although of late we have a vandalbot written specially for us, which hops from open proxy to open proxy. So an arms race over the questions would be laborious - thinking of good new quesitions is slower than telling your bot how to deal with them - and our work is to find what proxy list he's using and inform the blackhole list operators of it.
MediaWiki captchas are terrible and literally worse than useless (keep many humans out, fail to keep spambots out). reCaptcha is a little less worse. But if they have something that works for wiki spam, please be sure to post it to mediawiki-l! Thank you :-)
Protecting one specific Wiki from spam and protecting Wiki's in general from spam are two different problems. Spam security by obscurity works for a single Wiki.
This may have been a minor feature on your end, but it's certainly not a minor feature to the community!
Woot.
A thought: Now one might be able to find the most-controversial LW posts, (measured as max(up+down-abs(up-down), perhaps) in the same manner that Top gives us the most-popular. It might be interesting to see empirically what topics LW treats as controversial.
Remark: a+b-|a-b| = 2 min(a,b) so your proposal interprets "most controversial" as "largest absolute number of minority votes". (Not a bad interpretation.)
Your math-fu is simpler than mine.
That's the condition I was aiming to describe, yes. I was looking for a formula that would pick up posts that had a large proportion of downvotes, but were popular enough to imply a well-fought argument rather than a post that was simply bad.
It has always been explicitly possible to sort comments by controversy by clicking on "sort by." I believe that in the beginning there was a link to sort posts by controversy (discussion). At some point the link was removed, but the URL kept working. At some point I recall someone discovering it and people being suspicious that the sorting was buggy. But currently it lists posts that have received lots of upvotes and lots of downvotes. Added: read the rest of this thread for more detail about these links.
I haven't figured out what the definition of controversial is. There used to be a paragraph here speculating about it. I previously linked to the open thread, which was not a good example to exhibit sorting. The politics thread contains actually controversial comments and thus sorting actually does something.
Thanks for the correction.
By default, it only lists the past three months of posts, which weren't very controversial. Under the current system, controversial posts probably get demoted to discussion. So long ago that I forgot, I had turned off that filter and so I saw a list of posts from before the current system with scores close to 0 that had received about 50 votes total. I saw the same on discussion. The three month filter for discussion does have scores close to zero, but usually not quite so many votes.
Thanks; I hadn't noticed the filter. When I set it to “All time” it does show posts I'd expect to get both lots of upvotes and lots of downvotes.
This is actually a bug, not a feature. Some major changes to how karma is stored and passed around the application were made in order to implement the percentage tooltips. Scores are now passed around as an array containing the number of up votes and down votes, where previously it was the difference of the two. It appears the this array is being incorrectly rendered in the side bar, instead of the calculated score.
Given that people are already coming up with crude hacks for extracting some of that information, it would be nice if the percentage tooltip displayed the raw array as well.
I'm sitting at 87% positive karma, and looking at recent comments and posts, apparently that's actually pretty low!
Almost all the users I looked at had at least that much positive; I saw one or two out of a dozen with less.
I'm a little surprised but I'm not sure how surprised I should actually be, since I already knew that we have a great community that keeps people around who post good things.
Trike Apps is also working on a solution for the LW wiki spam, since captchas have not been sufficient.
FWIW, I just checked what the RationalWiki IP-edit captcha does: we're using QuestyCaptcha, which asks "What is the nth letter of our name under the logo?" (said name being of course "RationalWiki") - and this (a) utterly confuses the spambots (b) doesn't hamper accessibility.
Edit: Although of late we have a vandalbot written specially for us, which hops from open proxy to open proxy. So an arms race over the questions would be laborious - thinking of good new quesitions is slower than telling your bot how to deal with them - and our w... (read more)