Voting still has all the same effects (on karma, and on position on the page), so I encourage users to keep doing it. The only difference is that comment scores aren't displayed in threads.
* If a comment is sitting at 1 or 0, I'll try to avoid downvoting it unless I really think the comment subtracts value from the site.
* If a child is attracting more votes than the parent, and I think this is because the child commenter didn't comprehend what the parent was saying, I vote to level out the comments.
* If I see a comment sitting in negative territory that I feel should only be a 0 or 1 because it wasn't that bad, I'll upvote it even if I otherwise wouldn't have.
* If a thread is already dominated by a couple of high-scoring comments, relative to its age, I'm less likely to post a new top level comment. My feeling is that if a good comment just sits at the bottom because it arrived slightly too late, I've only added noise to the thread. The lack of visible points makes it much harder to gauge how a new comment will 'compete' on the thread.
* If the scores indicate that something is being widely misunderstood, I might comment where I otherwise would not have to try to restore sanity.
Of course, if this change were to stick, it would alter the character of the site, and we would all develop new voting and commenting systems. I can already tell that this change magnifies my tendency to vote with an eye toward reordering comments rather than voting on each comment in relative isolation.
I do think that people make too much of karma and comments scores. But I don't understand why pg thinks this particular aspect is an improvement.
exactly
After all, the biggest function by far of the votes is to influence the sort order, so the good stuff gets read by the people that skim. A vote is like a little bit of editorial power, you get to have a say in what goes 'above the fold'.
I don't want to be the asshat to drop a comment below 1, just because there's a better comment that's currently below it (but probably with more karma). At least from my usage, you'll see a net gain in karma inflation.
I can see ways hidden scores may work for or against extreme totals. On the one hand, without the satisfaction of seeing a vote take immediate effect, overall voting may go down. But on the other, without the indication that a comment is already 'far enough' in a desired direction, charged comments may rack up unprecedented positive or negative values. I hope you can share info about which effect predominates.
I will miss the chance to quickly 'rescue' slightly-downvoted comments I don't particularly like or agree with, but feel have been unfairly squelched into nether-ratings by hair-trigger down-voters.
My prediction: comment voting will drop precipitously.
On the same note, if an article is waaaay down at the bottom, and someone's taken the time to read it and likes it, then it should get a good solid bump. This will create a "bubble effect" and allow those treasures that are buried deep down in the comment land, to rise to the top and even accelerate as more users find out and rank them.
This essentially is a formula to equalize the comments and give them a fair playing field as they battle for the top most position.
There's tons of good gems at the bottom of the page, and they need to compete at the same level as those at the top.
You could tweek the "weighting equation" and try linear, quadratic, or exponential type formulas.
You could also apply this technique to articles, so those at the top don't get overly hyped and those at the bottom have a fair chance to making some movement to the top.
Also, one thought in terms of the scoring value is to have it with a few digits of decimal precision. So you could still see how competing articles are doing at the very top and see your vote make a difference (be it very small).
Example two competing top articles:
Article Exhibit A: (score 47.25) Article Exhibit B: (score 46.33)
Now when I click to bump exhibit "B" it moves to say 46.38 a boost of "0.05"
This should do the trick (driving the long tail).
Often when I notice a bump in my karma I look at my comment threads to see what was so popular.
BTW Cool experiment, think it might get rid of some of the group-think
EDIT: Fixed. Wow that was fast!
I feel kind of lost.
It can be when the site is new and only an overall smart population lives on it, but yc is (I guess) growing, and as it grows the quality of the voting (and therefore the score) decreases: insightful comments running against the general opinion get mercilessly downvoted, useless comments matching the general opinion rise.
That's one man's opinion. I read all the comments on every thread here, and find that not all the high-quality comments rise to the top. It's hit-and-miss. In a way that invalidates the theory of upvote-judging, no?
It's a shitty system that forces you to adapt to biases. Right now PG's experimenting with removing a bias and improving the system. You're arguing that it would be better if he left the system flawed.
It's true that PG is experimenting with the system, but it's a value judgment to say the old way is more flawed than the new way. In other words, your last two sentences beg the question.
Instead, let me just say that this common notion -- that because good content is the ultimate signal of quality, we should read all the content before making a choice -- is like saying that, because the way your spouse gets along with children is vitally important to your future happiness, you should try having a child with all the eligible spouses you can find before you choose one.
Since most people don't have infinite reading time, people use a heuristic to work out what's worth reading and what isn't, and upmods are a reasonable, if fallable, heuristic to use. I've just come back from an 8 hour shift at a warehouse, and I'm not going to bother reading all the comments - it's simply not going to happen. Sorry if that offends people, but I'm tired. I'm going to miss out on the the good comments because I don't have time to read all the comments.
Secondly, upmods smooth out discussion by preventing redundant comments and people repeating the same point ad nauseum. While this might be perceived as "groupthink", being able to separate a fringe viewpoint (even if they're not necessary wrong) from a popular view is an extremely useful heuristic, because the prevailing opinion is often (but not always) more likely to be a reasonable and reliable answer.
I have a feeling that most people actually practice these heuristics in real life too. Given finite reading time and the contents of all books in history, most people read books that others people have recommended or are popular.
My first comment was http://news.ycombinator.com/item?id=739112 Being my first intervention on HN, I thought I would be able to post anonymously so I included my name at the bottom of the post. The first reply (which, I admit is polite, well written and from a relatively famous person in this community, although not completely on-topic with respect to the article originally linked) got upvoted above ten. My comment, which revealed a lesser-know but true fact strongly related to the original article, with reference to an experiment that you can reproduce at home, was upvoted twice or something like that.
Fuck that. Are you saying that you're too lazy to read what people are saying and form your own opinions? That groupthink is just going to naturally turn out better than forcing every member to think about what they're reading?
For the longest time, HN has become more and more "accessible", at the cost of a lot of intellectual discussion. I've seen a lot of conversations where one side of the argument is downvoted to hell and it makes the other side look "correct". I just saw a thread two days ago where somebody in the debate responded to a guy who was at -4 by saying "Look at your comment score, looks like I'm right." That's shit. That's utter shit and it hurts Hacker News.
I'm fine with spending more time reading a thread if it means I'm legitimately thinking about what people are saying. This is a terrific improvement to the system, and it feels a hell of a lot less claustrophobic than it did when every single thing people said was being judged as if it were an objective statement capable of being "good" and "bad."
Hopefully this also stops people from downvoting statements based on disagreement. Feels less vindictive when you can't see what your vote's done.
That's assuming that the majority is right. If somebody gets voted down there is a pretty good chance that they were wrong, or even if they were right, they said in the wrong tone/manner.
I also see a curious downvoting pattern on those frequent incidents when somebody mass downvotes all my comments for some reason or other. I'll assume for argument that on average I make useful, relevant comments. When new comments that I make get downvoted, there's about a 50-50 chance that the next person to come along also downvotes the comment, even when it's a good one. They see the other person's downvote and it reenforces in their mind that there was a good reason the comment was being downvoted. The ones that get a reverse upvote, however, then soar upwards.
I had made one comment a few days ago that go downvoted such, then for a while hovered between -1 and -2, with very little differentiation. Then it hit 1 again, and within a few hours had hit up to something like 16. That's a pattern that I see a lot. People are less willing to upvote a comment that's grey than they are to upvote something solidly in the black. That's a flaw.
The use of pejorative language obscures your content. Like "fuck that". Also consider pg's comment recently about "bogus" research, and how so many people then jumped on him. Actually, someone got sued in the UK for calling research "bogus", instead of factually and objectively calling out the problem with it. It's also in the HN guidelines to use neutral and factual language without connotations.
By observing myself, I've noticed that my interpretation of a comment is affected by the upvotes/downvotes on it. It's not surprising, given the experimentally confirmed tendency to conform in humans. http://en.wikipedia.org/wiki/Asch_conformity_experiments
Finally: negative scores on a story are still shown... not numerally, but by how gray it is. Thus, the current experiment does not apply to comments with negative scores. I think it would be worthwhile extending it to include those, at least for -1 or -2 comments (for example, by not graying them out til they hit -3).
There's no such thing as a democratic theory of truth. The truth isn't something that can be voted on. Does God exist because more people believe in Him than believe in a secular universe?
The use of pejorative language obscures your content. Like "fuck that".
"Fuck that" obscures nothing. It's a concise way to say what I wanted to say. I could have said "I disagree", which takes twice as long to say and sounds garbled. I think there's a place for the word "fuck": It's for saying something very clearly and quickly when there's no need for delay. So, "Fuck that" doesn't have connotations. Not in the way that "That's shit" does, or even "Fuck you".
Finally: negative scores on a story are still shown... not numerally, but by how gray it is. Thus, the current experiment does not apply to comments with negative scores. I think it would be worthwhile extending it to include those, at least for -1 or -2 comments (for example, by not graying them out til they hit -3).
I agree completely. As an aside, it's interesting right now watching my comments just in this thread. Once they hit 0 and turn grey, there's a lot of turbulence and they hover around there until they go black again, but the comments here that are in the black have quite high scores. That reinforces my belief that reading a grey comment biases you to think it's a bad comment. So in my mind, the threshold now where a single voter can grey a comment is too low. Setting it to -2 means there'd have to be a three-vote discrepancy between ups and downs, which I think sounds about right.
I disagree with that. Of course I could have said "Fuck that." However, just like you probably wouldn't want me just saying "Fuck that" in response to your viewpoint other people aren't going to appreciate a comment that says "Fuck that."
I will vote down any comment such as that, and I'm sure that many other members will as well.
I swear a lot, I grew up in a fairly rough area of a big town and it took me some self control to get rid of it.
It's not that I mind, it is that other people mind, and the use of it in 'polite discourse' is therefore discouraged, there are other ways to express exactly the same thing without chancing stepping on someones toes. In general, if every third comment on HN would contain swear words we'd have to start calling it 4chan, and I'm sure plenty of people - including me - would leave.
That says noting about your freedom to choose whatever words you want, but you have to be aware that you are changing the atmosphere in a non subtle way by your choice of language.
Don't just go along with the pack.
Don't you think that reaction is a little bit extreme? Unlike you, apparently, I don't have time to read all 800 comments on all the threads on this site.
Unlike you, apparently, I don't have time to read all 800 comments on all the threads on this site.
Really? When you're spending 30 days flying around on a plane and sleeping in airports, I'd imagine you're the one with more time to waste.
The top comments on HN are still the highest-voted ones. Read however many you want before the discussion tires you. Also, I can't think of many HN threads that ended up with more than 300 comments. 300 is at most an hour's reading; an hour isn't very long.
I pointed out an observation. Your response used the word fuck and then called me lazy, which is extremely rude. Would you have said that to my face? This isn't personal, dude. I just get the same feeling now when scanning comments that I get when I am looking at products on Amazon and can't see the price.
In person, I'd probably be doubly critical. Don't think of it as rude; I think criticizing what somebody's saying is a sign of respecting them. To quote a guy I'm in the middle of reading: "The more we love our friends, the less we flatter them; it is by excusing nothing that pure love shows itself."
If I were being rude, I'd ignore your comment and not call out what I thought was stupid/wrong about it. As it is, I'm debating what you said because I think there's a discussion to be had. For instance: Why would you feel the same reading comments as you would buying things? There's no harm to reading a few extra comments. You lose nothing but your time, which you're already spending on Hacker News.
I don't have time to read all the comments so I'll look at the greatest hits. You can find the conversation based on the kharma scores, both positive and negative. If someone says something -4, it might be worth reading because it got such a reaction out of people. But a bunch of 1s and 2s on an old thread are noise that aren't worth reading.
What, that I'm a student? Reading comments doesn't take that long. As I said above: Fifteen minutes at most? And that's if it's a seriously long thread. Most threads are five minutes' skimming at most.
If someone says something -4, it might be worth reading because it got such a reaction out of people. But a bunch of 1s and 2s on an old thread are noise that aren't worth reading.
Not necessarily. Perhaps somebody made a brilliant comment, but only after people were done reading. In any case, that hasn't changed now: Top-rated comments appear at the top of the thread, same as always. So if you want the most "valued" conversations, read at the top.
So if I pick a thread I read all of it, sometimes again at a later point in time (or via the 'threads' page).
This also frees me of any need to complain about offtopic stuff, as long as a few on-topic posts remain.
True, but often times the top comment will not be the highest rated, but the reply or subsequent replies below the comment will be rated higher than the orginal post, but the entire thread gets moved to the top because of that.
Now I have no way of knowing that.
[edit] I also have no way of knowing when the MAJORITY of readers are disagreeing with me.
Up votes/down votes has always been a way to express yourself without commenting. If it's not visible, it's not as expressive, nor is it as rewarding to the commentor or the voter. I officially dislike this.
Someone may not down vote my comment, which, say, has 7 points. But they may leave a reply that is argumentative to my comment. With this new system, I can only see that my comment has 7 points, but I can't see that the counter-point reply to it may have 34.
The fact that many people are agreeing with the counter-point to my comment is significant, as is my inability to see that.
Why would that matter ? The majority isn't always right.
HN is not a popularity contest, but when you look at the comments that way it certainly starts to look like it.
Let's see what I can say that many people will agree with so I can be voted up ?
As unalone already remarked, now you have to make up your own mind about what you are reading, which is a huge improvement over the groupthink mentality. It forces you to think for yourself before casting your vote, it takes care of the huge feedback loop that was in HN before this change.
Even now, there is still some of that left, after all the sort order still tells you what the majority voted for, but that's subtle enough that it probably won't matter too much, and the new sorting algorithm seems to do a very good job of bringing the good stuff to the top.
What the old system did was to provide a numeric indicator of other readers' emotive participation in the discussion. It wasn't a popularity contest (although it could be for some) and it wasn't an agree/disagree rank (although it could be for some)
These "bad" uses of the score caused the system to get out of whack as more and more people participated. The problem is, there are "good" uses of the score, which is to use it as a way to filter the site. After all, there are thousands of comments added per hour and it's crazy to think that I'm going to have time to read every single one. I need an indicator of those comments that seem to affect the most number of readers because that's where I want to add my limited amount of effort.
I am not on a treasure hunt for good comments. I am not here to please PG or the other readers. Some might be but I am not. I am not jumping onboard to upvote things already upvoted -- the sad fact is that I can't read everything, so some things will have to have higher priority than others. If you take away the numeric system, I'll find some other way to filter -- perhaps by comment author or by scanning the front page for buzzwords that interest me.
I, the user, have a need to skim. Now the system can either easily provide that for me, or I'll probably go somewhere else. I don't mean that as a threat or anything -- I'm sure nobody gives a hoot where I read on the net or not -- but as an indication of the type of reaction the general public is going to have.
The sort order is what gives you exactly that, and that is still there, so the 'good' stuff is still at the top (and even better than before now with the improved sorting algorithm). It's just 'relative' now, instead of 'absolute'.
But right now I look at this thread and all I see is a bunch of jumbled-up comments. I don't know if the one at the top is worth-while for me to read or just matches some arcane criteria established by the programmer.
I'm forced to use "threads" to see where I've commented before and to use the new comments page to see where people are currently commenting. Any semblance of being able to apply my own sort criteria is gone.
And I think it gets to my ability to sort by my own criteria and the nested nature of the comments. Do I want to find highly-ranked comments under a lowly ranked parent? Sometimes those are the best ones.
I don't know. Let's see how it goes.
Yes it is.
"Who are you? Why does it seem like you're angry all the time?"
Which is an angry, personal remark.
The word "fuck", not directed at anyone, is not important. Personal attacks are.
If you don't like unalone's style, ignore him, don't psychologize him. Personally I appreciate the fact that he wrote substantive on-topic comments.
I think Are you saying that you're too lazy to read what people are saying and form your own opinions? is a rather personal remark. A psychologizing one at that. Pot, meet kettle.
"Are you saying that" is an opener that means, "How is it that you're viewing this, because I see it like this". I wasn't trying to be snide and put him down. I was letting him know that's how I was viewing the situation, and asking him his interpretation. It was also a teeny bit tongue-in-cheek, considering, again, he's currently spending a month of his life flying in planes.
More importantly, though, the discussion itself was about the poster's demeanor. In all other situations, downvoting isn't an accurate measure of sentiment toward the post, because voting mixes such things as agreement, clarity, obnoxiousness, and humor ratings, weighted differently for each person, into one opaque integer. However, in this case, when a person is voted down for [obnoxiously] arguing that they aren't obnoxious, and the parent is voted up for arguing that they are, in fact, obnoxious, the score really only has one meaning, in both cases: how much people dislike the poster. This is an exceptional situation; I never would have argued "by comment score" otherwise.
You mean in the same way a book's price is an indicator of its quality? ;)
The nice thing about comments is that they are short and quick to read. Unlike books or films there's no need for reviewers or critics to help you find the better ones. Just read and decide for yourself.
This is great. So now I can see which of my comments are getting voted up/down and so have feedback for my writing, but I have to vote other comments up or down based on content (and author sometimes I guess).
Oh i was just referring to the "vote pg up anyway" trend.
If the author is someone you know (and I've made many friends from YC) there is a minor incentive to upvote a comment even if it doesn't read well because you know the person and can guess at what he was trying to say. So instead of judge(content), the subroutine becomes judge(content, person). It sort of feels like "voice tone" is added to a comment if you know the person. The "tone" and (perceived) intent with which something is said makes a huge difference
This could lead to a known troll getting a downvote and a known "good guy" getting an upvote for the same comment.
And that is probably a good thing.
By the way, the kind of skimming you're describing is what leads to mob voting :) People skim for the highest rated stuff, read it, then rate it up more. Plenty of good comments never get rated up, and people who say the most inflammatory stuff (or just make stupid jokes) get the votes early on and stay in the "lead" as it were, the system having been made into a game.
The entire point of using karma is that it means something -- it has value. Heck, I bet to some folks it has a lot of value. I'd further bet that it has so much value to some folks that you could monetize it. But that's a discussion for another day.
I wonder if you could just have three buttons/widgets: insightful, funny, and troll.
Then just click what you think of the article, and insightful articles rise to the top.
Heck, you could even make it a user preference whether to sort insightful first, funny first, or troll first (why you'd want to do that I don't know, but some might)
It would involve just one more graphic element than the current up/down arrows, and would give people release to say the funny things when they feel like they have to instead of having the site self-police.
At least on slashdot, the "troll"s are sometimes, in their own way, funnier than the "funny"s (if only because of not involving overdone /. in-jokes).
But that's only sometimes, not all of the time. If there could be a button for turning jokes off or on (or sorting by funny) then it'd be a hoot to see the site in that asepct.,
In short, I don't think it's a human behavior problem. Computer systems should let people act like people naturally act. This is a display problem. Some folks are seeing things in spots they would not like them to be in.
Instead of viewing the site as some huge function to take limited input and provide maximum output, perhaps it's better viewed as a big bucket of various randomized data. Then the purpose of the site is to sort and organize that random data in such a way as to provide maximum utility.
I agree. HN, while of course useful in other ways, seems so humorless to me. I understand that some HNers are paranoid about HN turning into Slashdot or Reddit, but it would probably be nice sometimes to have a more community-like feel, even if it were totally segregated from the "real" HN.
Exactly! A lot of games just got killed, and now we're hearing some protesting about it.
The entire point of using karma is that it means something -- it has value. Heck, I bet to some folks it has a lot of value. I'd further bet that it has so much value to some folks that you could monetize it. But that's a discussion for another day.
You don't find it slightly pathetic that people would be judging themselves by the score of their comments on a web site? And I say this as an upstanding member of the Karma Kommunity.
There is, actually, one system of points that works brilliantly. Metafilter's favoriting system, which allows upvotes but not downvotes, lets top comments get notified, but doesn't penalize any other comments, and because MF uses a flat commenting system the conversation isn't at all affected by what people like and dislike.
I think this change is for the better.
I'm used to scanning comments as a way to determine which articles to read. Postings with a lot of comment action usually indicate something emotionally appealing.
Once I go to the comments section, then I filter by score, only looking at comments above a certain threshold (depending on how much time I have). I usually try not to comment, unless some comment has been upvoted to a high level and doesn't make sense to me.
Alas, there are a huge number of comments that I disagree with or fail to understand, and only by knowing the score am I able to determine whether or not my adding to the discussion will help the other readers.
That don't work no more. Now I'm just blindly poking around in the comments section, not really feeling connected to the discussion at all...
The top-level comments presumably rise and sink based roughly on their karma (though the algorithm also seems to have other considerations, like momentum). But what of the second-tier comments and below? Ripostes are often where the action is.
I might find that I'm reduced to skimming for names that I recognize. That's how the rest of the web works. The web in general has no karma system, so the way to "upvote" something is to repeat it with your byline: You Twitter it or reblog it (or retweet it, or re-reblog it) with a link and your signature, implying a certain degree of implicit endorsement.
I expect that if this system holds we'll start seeing more "amen" comments. I kind of hope so, frankly: Now the only way I will spot an incredibly insightful but short comment buried downthread is if I see a bunch of comments from people I vaguely know saying "amen".
I'm not sure this trend will represent progress, however.
If we want to tinker, consider this: relative rankings on a page. Find a way to distinguish the top-voted 20% of comments from the second quintile, and the second quintile from everything else. (My instinct is not to bother distinguishing things below the top 40%, and of course negative-karma comments presumably are still doing their disappearing act, my favorite cute HN feature.)
At the risk of repeating the legendary eye-searing Orange Name experiment... we could turn the byline of comments in the top 20% a different color from the second 20%, which in turn is a different color from everything else.
Red is probably the wrong color for any of this -- too loud. Think "green". I'm thinking shades of green: Comments in the first quintile have very green bylines, comments in the second quintile have slightly green bylines, and all the other comments have their usual light grey bylines.
PG wrote an essay which discussed the 'fluff principle' at work in other voting-based aggregator communities, in which the most superficially sensational content usually rose to the top, beating out longer-form but more intellectually substantial content: http://www.paulgraham.com/hackernews.html
I think this move by PG is aimed at reducing the herd mentality of voting, and reducing the ability of readers to engage in voting without having engaged the content first.
However, I think your scheme of highlighting comments by quintile is probably a good one. It gets us away from thinking about raw scores, but also serves to reward and reinforce good commenting contributions/behavior by setting an example for others to follow. In order to stave off the herd mentality of 'piling on' upvotes or downvotes, perhaps votes made subsequent to highlighting (or greying out in the case of negative comments) should have half the weight.
If you think that the secret to encouraging engagement is to flatten out the structure of the information, and ask everyone to read everything in order to discover the influential, popular, or insightful bits...
... you desperately need to read The Paradox of Choice:
http://www.amazon.com/Paradox-Choice-Why-More-Less/dp/006000...
There is nothing engaging about thirty screenfuls of undifferentiated choices. When presented with that, I'll just leave. If I wanted a firehose of undifferentiated, recent, quality content, presented in a way which made it very difficult to nucleate a conversation or form a community, I'd be using Google Reader.
[EDIT: Incidentally, if not illustratively, I should point out that I haven't actually read the entirety of The Paradox of Choice myself. I listened to the author lecture about it for an hour in a podcast, and I started to read it, and I appreciated the concept, but I felt that the book kept repeating the same point too many times and I had other uses for my time... ;]
Yes. It's like "we want the voting system for our own purposes. You schmucks can just read every comment under an article because your time is not as valuable to us as our desire to prevent mobbing"
I'm sure that's not the case, but that's the way the new system feels.
20 Points for a comment 'supporting' my position and 20 for a comment arguing 'against' my position indicate that I shouldn't be too certain, but that my position is (probably) at least rational.
20 Points for a comment 'supporting' my position and 1 for a comment arguing 'against' my position is an additional sign that I am probably on the right track.
1 Point for a comment 'supporting' my position and 20 for a comment arguing 'against' my position is a clear sign that I may have missed something and should read that comment carefully.
I really dislike this change.
If somebody refutes you elegantly, then you should look into their refutation regardless of whether or not the refutation has many points.
Meanwhile, I think if somebody replies to you with something like that, it's worth investigating regardless of if they're the top answer or not. I'll draw a parallel to Reddit's "What's your favorite album?" threads. If you were to go by comment points alone, you'd think Radiohead's OK Computer was the best album of all time and that NIN took the four spots under that; meanwhile, albums out of the hive mind don't get as many upvotes, even if the commenter is more passionately in love with them.
Similarly, just because a lot of people think one thing on HN doesn't mean that one thing is the best answer. Looking up multiple solutions may take more time, but that leads to learning and growth, which I think should be the point of any community.
You obviously participate in HN a lot. Don't fault the rest of us for only wanting to do it part time.
Sure, it's a noisy signal because scores don't just indicate agreement, but the context usually makes that clear. If a factual claim is up- or downvoted, it usually indicates that people believe the claim is true or false. For analysis or opinion, it may have to more with whether the comment is insightful or thought-provoking.
I'll stipulate that the voting might be more "fair", if that is a goal. I'm not sure it should be. It might make the "game" of getting high karma more fair, but I'm not bothering with or bothered by that game anyway.
On the other hand, I can no longer use the results of this voting to help me skim a conversation. I want to know what everyone else found most valuable, without having to read through every comment on my own. I can't do that anymore. This completely breaks how I use HN.
1) Keep the absolute scores secret
2) Continue graying out down-voted comments
3) Do not show any numeric indicator for conversations with less than N comments (I'd experiment with N near 10)
4) When the number of comments is greater than the threshold, show which percentile the comment is in for the set of positive comment scores in that thread.
This way, when you look at a comment you don't think "Does this guy deserve more points?", instead you think "Is this comment better than 75% of the other comments here?"
I think this would improve voting AND increase the value of that voting: people would have the right frame of mind when voting and the UI will present an easier value to determine how worthwhile it is to keep reading.
The problem could of course be solved by showing the points...
The other side of that is that threads with very little traffic don't need as much meta-information, but there's an in-between.
And I really don't like the red dot. It doesn't fit into the flow of the layout and it distracts me from the content.
Could this dot be changed to green? Also, could you put an alt or label on that dot with something like "over 5 votes" or some other descriptive text? A bit more accessible.
It might be worthwhile to try and map the color code onto the arrows. Red up arrow instead of red dot could serve both as highlight and as "stop, don't click me".
I also suspect that you haven't thought this through. Try mocking up a page full of red arrows instead of red dots. Even without doing so myself, I predict that because the arrows have twice the area of the dots you'll find that it looks far more cluttered.
Meanwhile, my bikeshedding alarm is flashing. ;)
Although I guess I never saw the karma thing as a huge problem to begin with, so I suppose I'm a bad person to opine about this.
But the more I ponder what you're trying to solve, the more I think you should provide scores weighted within the discussion. Perhaps, the quintile a comment is in. Use a bar that adds segments and shifts from one non-judgmental color to another. I wouldn't bother showing anything below a threshold of ten comments on a single link.
Otherwise, your piling-on problem returns, just with a little less fidelity.
On second thought, the problem with the quintile idea is that it presumes that every thread has an equal distribution of valuable contributions, middling contributions, and lousy contributions. But we don't want to rule out (e.g.) the possibility that an entire thread might be great and that every post deserves a dot.
Moreover, the quintile idea retains aspects of the piling-on problem: A sufficient quantity of super-upvoted comments will push other worthy comments out of the top quintiles.
Finally, quintiles are just way too complicated.
So I'm prepared to believe that I was wrong twelve hours ago. I like the threshold idea. Especially because the threshold can be tuned. If 5 isn't enough it can be turned up to 11 by a well-meaning moderator, acting at the appropriate time.
How hard would it be to allow people to implement their own version of the ranking function and then have the comments passed through that. Store the function in each person's profile, make them visible, and more importantly, make them share-able (like an app store?) so you can see which functions turn out to be most popular...
There's a giant can of worms there safety-wise (people running their own code on your server ?!?!) and unknown loading effects (nothing to stop people writing infinitely recursive loops...), although that said, it didn't stop you with viaweb did it?
Also haven't looked at the news.arc code recently so not entirely sure which part could be customised...
Harder than you think, I expect. You're proposing to do a custom query, of arbitrary, user-chosen complexity, for every single page view.
Moreover, the result will be of dubious merit. Individualized filter functions screw up conversation. That's what's wrong with conversation on the rest of the web -- people all read blogs, but nobody reads the same blogs, and people encounter the blogs in arbitrary order, and some blog entries are more out of date than others, and the result is like trying to carry on a conversation with a bright but slightly deaf time traveler who is probably a troll.
We have a distributed, personalized version of HN comments: They're called blogs and RSS feeds. The quality tends to be higher. Your personal filtering options are much more numerous and powerful. And they're just not the same thing. Otherwise I, for one, would be writing for my blog instead of posting on HN. Lord knows it would be better for my career.
The difficulty in all this would be policing the kind of code that can be written - perhaps a small DSL that compiles into arc code, or possibly direct arc code that gets vetted by a/some moderators.
The actual sort function seems to be (from arc3.tar, but may well have been changed here... nothing is safe from a REPL!) EDIT: indeed! see http://news.ycombinator.com/item?id=845932
(def display-subcomments (c user whence (o indent 0))
(each k (sort (compare > frontpage-rank:item) c!kids)
(display-comment-tree (item k) user whence indent)))
The conversation threading could indeed become a bit fragmented, but then again, each time you look at a story on HN the comments will be in a different order due to the weighting and voting.
No, it's about the same.
It's not a problem with hiding/showing scores/hints, it is a fundamental problem with the voting model. As it stands, the process of voting is meant to re-balance the discussion tree to make it more interesting to an average reader.
IMO what makes more sense is to use votes to determine the preferences of an individual specific reader. If I keep voting up posts by a user X, then in all likelihood I will be more interesting in X's posts in the future. The same goes for downvotes. Additionally if X votes up Y's posts, then I may also be interested in Y's posts as well.
Lastly, the up/down voting score for a specific post is going to be equivalent to the score under existing model, and that can be used to sort posts for anonymous or uninitiated users.
PS. This is not a trivial change and it is a lot of work. I realize that.
I am wondering was there actually conclusive data (other than negative feedback, which might have been enough?) during when the score was removed and before the dot was implemented. As in, without a way to see the high scoring comments, did the quality of child comments go down? And if so what? Voting behavior changed?
Or was it an issue of being able to easily see popular comments? (And if it was the later, then maybe there are other solutions of visualizing highly ranked comments although likely more complicated, like a sort)
People can check Lists, for interesting stories/comments. http://news.ycombinator.com/lists
Personally, I like seeing my karma go up when I submit useful comments or stories, so they work as a reward for me (and possibly everyone else here). However, I think the more important function of up/downvotes is bringing the 'best' discussions to the top and burying the garbage.
My dictatorial solution would be to hide comment count (and all visual indicators other than thread position), from everyone but the user they belong to. I would still get my 'reward', the mob-voting influence would be diminished, and comment quality would go up.
I believe pg has in the past just used "this seems better/worse" as a metric, but it's hard to say how noticeable any change from this will be.
Perhaps my sense of dislocation will pass in a while.
I say bring back the ability to see one's own karma, perhaps with a preferences option to turn it off, regardless of what happens with the visibility of others' comment scores.
In fact, if anything about one's own karma would be hidden, I think the overall score would be a better choice to hide than the scores of (one's own) individual comments.
Do you really care so much? Anyway, If I were pg I would play around actually: half users see votes, the other half orange dot or whatever. I bet the numerical scores create a sheep effect and people who see high numbers are more likely to rate comments up.
Indeed. Interesting how fast the brain adapts. I just caught myself looking at the total score and trying to predict which comment is getting upvoted (since I see the total increasing). And I don't even care about comment score!
(I have also noticed that people don't stop upmodding even if the score is visible. I got 80 points yesterday on a one-line comment poking fun at Joel Spolsky. WTF?)
Just like when there is the delayed "reply" link that isn't showing, you can still reply to a comment by clicking the "link" link.
Edit: why is this being downmodded?
I didn't see the comment you mention, but that is exactly the sort of thing I meant about voting starting to have a mob feel.
Although I don't have any data, I think that if the same comment was posted today, it would end up with more points. If I see a comment with 80 upvotes, I usually let it stay there rather than bumping it to 81.
One other comment; I find it hard to navigate my "threads" page now. I used to focus on highly-voted comments to improve and reply to; now that I don't know what other people like, I have nothing to go on.
I guess spending less time on HN might not be a bad thing, however :)
Edit: one more thing. I still have a reflex-like urge to click the upvote button when the author name says "pg". With no number, it is even harder to suppress the urge.
So far I have managed, but I can't guarantee that I won't change my mind :)
Interesting to experiment.
PS: please don't ask me how to quantify such data :)
Over all, I like the comments on HN much more than say reddit or digg by far. Plus, I've never seen a community that responds so well to criticism as this one does.
+5 -3 +4 -2 +4 0
That means someone is being generally rude or uninsightful, but the replies are good, so I'll just read the replies. If there's a long thread of a bunch of 1's, then two people are talking at each other both uninsightfully. If both people are voted up a bit, that means it's a good discussion and the whole thing is worth reading.
I'll give this a good trying out before forming an opinion either way, but the loss of skim-by-score ability is slowing me down just a touch right now. It'll be interesting to see how this turns out.
Edit: Sorry, it looks like this was explained after I had loaded this page. Still, this seems like a partial reversal. Also, this story is at #16 on the front page and really should be at #1 considering the magnitude of the change.
Oh and just one more question: I'm seeing this comment at the top of the page. Is that just how it looks for me, or is it that way for everyone because of the new ranking algorithm?
Another edit: From PG's explanation, I thought the gray dots were all turned red, but I still see gray dots for some comments (including one with a negative score).
Thanks. The comments here are getting to be so much that they are hard to follow. ;)
At one point I tried giving all comments initial dots, and turned them red when the comment had a high score.
We are having the same problem on how to show it correctly on GraffitiGeo. We ended up using "by ivinkirigin, points: 10", which in my opinion is still not perfect.
Any idea on this?
If the numbers are changing you get accustomed to it quickly.
By the way, my original comment has an absurdly high karma count. I think that is a direct result of hiding the count.
I don't believe I've seen a site where any user can give more than one vote, so I don't think this is misleading. But I've not thought about it before, and I do like "by ivinkirigin, points: 10" better.
Kudos on putting so much thought into your UI. Definitely bodes well for GraffitiGeo.
A suggestion: Any chance the numbers could show later as the article gets older? I'd like to see how the votes resolved after some time. Or perhaps hide them only on front page articles?
heh! so did I . I actually asked a friend to look at it and tell me if he saw what I was seeing! (Long day, too much code, off to bed)
EDIT: I tried to vote up the comment above and got a "Can't make that vote" message. Then I tried to vote down a comment and got the same message. Bug?
EDIT2: Now I get a blank page on upvoting. I guess PG has a REPL open and is changing the site "live"
EDIT3: Blank page on voting (firefox/linux). This acts as a subtle disincentive to vote. "Oh I am going to see that blank page again and then need to click the browser back button. I'd rather continue reading(vs voting)". Is this by design? I hope not.
EDIT4: Fixed.
A 'good but late' comment won't be seen by anyone anyway, due to your front page ranking algorithm that prevents even the most active threads from spending long enough on the front pages.
If my personal interest + the comment score < arbitrary threshold, i quit
Won't this lead us towards a bimodal distribution? Without the crucial knowledge of where the comment's score is, the votes will tend towards extremes: we can't collectively decide when the target comment is near the range that we find appropriate.
reddit_user 622 points
Glenn Beck is a wanker! He sniffs arses!
Bad as inaccurate quantification is worse than none.
Bad as rough, but roughly accurate is better than none.
Then there's just a list of comments and one can vote them up or down on the content of the comment. Ok I did say it was weird.
EDIT: Some great counterpoints below. I agree with them. This is a bad idea.
cperciva said "Knowing who wrote a comment often provides very useful context"
You are both completely right. My suggestion doesn't make sense.
It was just a "spur of the moment" suggestion to provoke thought (de Bono's PO mental operator). I dind't think it through at all.
For instance, the ggp is a security professional. I give more credence to his opinions than others.
On the other hand, when somebody who is deeply respected (be it pg, or any prolific high-karma commenter) posts on something where they have no particular expertise the comment shouldn't get special consideration due to the author.
I can't think of a way to reconcile these, and there's enough people who post here that are actual authorities in relevant areas that I wouldn't want to give up the first. Being able to know that, say, a comment about security was written by tptacek matters more to me than "oh look, pg made a two-word post that got 9001 points".
Whoa! Nice :-). Any concrete examples?
Some controversial threads with large numbers of votes were starting to feel like fights that attracted a crowd of onlookers. Those are very engaging, certainly, but it's probably better for both fighters and onlookers to have less of them.
Removing information helps nobody.
Out of the total comment Karma alloted to all comments for an item, what proportion of this amount is contained within each comment?
The issue here is that upon voting, you couldn't change all displayed comment percentages at once, it'd have to change on page reload.
Also, it could be a rounded 1-10 scale.
-- edit: would be good to reveal any comment score upon voting: gives an incentive to vote and remain critical. It's also like a marker as to what comments interest you for future recall -- that's something new.
- either cap underlying comment scores at 50 points
OR
- instead of giving a percentage score for a comment's proportion of total comment points, rank comments in terms of lowest positive score to highest positive score: then give each comment a score between 1-10 by segmenting the long tail into 10 even parts. (the initial ranking may not even need to be used)
having no visible scores seems surreal.
That would balance the elimination of mob voting with being able to ascertain (historical) credibility.
Also, like some others , I like know if something I've posted was particularly liked (or not).
And I've been skimming pages looking for the high-karma comments to try to dig out what are likely to be the most worthwhile comments. I suppose there's some feedback effect in place there, as popular comments simply get more attention (and so likely garner more up-votes votes), but there's pragmatic value in having a metric to selectively peruse a post with so much to read. (And, no, ZapRead isn't the answer :) )
Simply choose half of the UIDs at random and always show them the comment score, and don't show it to the other half. Track which comments are upvoted most when their score is visible versus when not (there are many other metrics to use as well). It should be pretty easy to determine how much groupthink plays into voting.
I suspect that most people only read (and therefore only vote on) already popular comments, so I suspect groupthink is fairly high here.
On the other hand, showing different HNs to different people doesn't exactly strike me as community-building behavior. I think it would make the site feel even more like Temple Grandin's barnyard, with us in the role of the cows. Will my species behave better if we design the floor like this? Will that tweak to the algorithm prevent us from starting fights among ourselves? Oh, no, a gathering of more than twelve people -- we'd better build a robot to break those up.
What you won't be able to do is aim things toward a desired positive score, such as downvoting a one-liner joke with 100+ points that you wouldn't downvote if it were at 20 points or less.
That is, comments that (intentionally or not) pander to the audience will get more upvotes, but good comments that happen to get several early upvotes won't get the same pile-on effect. That's just a WAG, though.
Similarly, now dcurtis and I are arguing upthread, and I'm not anxiously wondering if he's been voted up more than me. We're just in a normal argument where neither of us will be proved by the community to be "more" right than the other. That's good.
Leaving karma off somehow gives me confidence to contribute more.
Another idea is to remove the (time 'ago). Good comments are timeless :) and old comments seem like nobody is watching anymore, so the dialog seems over.
Thank you Paul
Good idea.
The problem with that is that anonymity leads to trolling.
Also, credibility matters. Some people are known for their expertise in certain areas, like Matt Maroon and poker.
Give it a couple of days worth of trying before casting your 'vote' on the changes, it's a pretty disruptive change because it breaks the routine, but I really think this sort of experimenting is what will bring out the best in HN, if hackers are starting to be conservatives then we really are in trouble.
Give it the benefit of the doubt and an honest chance.
http://news.ycombinator.com/item?id=217822
So I'm really glad to see this and hopeful it works out. I think hiding the votes is better than a hard upper vote limit, as it is more organic. I'm really curious to know after the experiment how this affects the voting behavior (more votes, less votes, more for lead, less, etc)
I often read comments with highest scores first while often there are good comments forgotten and sandwiched between a few übercomments. Or the übercomment is so conclusive that there's nowhere to take the discussion to. Hiding scores allows me to approach the discussion more objectively.
I agree with PG, HN started to have mob feel in the comments section.
* the reader wants some filtering, so they can just read the good stuff
* the filtering is done by the readers, which requires they read more than just the good stuff
So the filtering can never be extremely effective, i.e. showing everyone only the good stuff, because then the filtering would not get done at all. I suppose it is a weakness in all public-contribution-based systems.
The work has to be done by someone. Perhaps there are other, better ways of allocating that work. A first thought: lower scored items appear stochastically -- lower score, less frequent random appearance. That might help by making it seem like there is less 'work' to do . . .
It seems the whole subject would warrant a fair amount of pondering and experiment (although it looks basically economic) . . .
How about allowing users to set two thresholds for themselves: (1) show only comments above X; (2) begin greying out below Y
What precisely X and Y are or could be I'll leave to the experts.
But this would let everyone decide their own absolute or relative threshold of comments they want to see. Those of you with plenty of time to spend on HN can have fairly low X and Y, whereas those of us that only want to spend a little time on HN each day can have just the popular presented to us. Voluntary herd membership (or rejection), if you will.
Variant along the same lines: An absolute number account setting indicating "I want to see only the N highest-scoring comments/parents".
But, it will stop them from finding those highly rated comments hidden by a low rated parent comment.
Is there any chance we could have two style sheets, one in which they are hidden and the other where they are visible?
And, to rant a teeny bit, that's what downvoting is used for. You don't downvote people that disagree with you or that say something you don't like. Downvoting is for removing comments that aren't relevant to the discussion. That's why they go grey. When I saw your comment, you'd been downvoted and greyed out, and you made a comment with a relevant or interesting point. You should not have been downvoted.
I hear there is this thing called a "dev server"...
I don't trust myself not to kill my app with a live REPL :)
Now you don't really know if people like your work or not, so why bother contributing?
(I know why: "because someone is wrong on the Internet!" But it is nice to get some karma too, :)
If they are turned back on, I'd suggest making it an option, and possibly hidden by default.
Although that being said, I tend to gloss over the scores anyway for the most part...
This feature will be interesting.
(This one is going to go down, but that's all I have to say.)
Hacker News -- now with less information?
I would imagine that the more readers become engaged with the site the more information about the site they would want to consume, no?
It just clashes with the idea that sites should be very simple to use, but engaging in detail over time -- that as users spend more time on a site they become interested in more details, whatever those details are. I think you see this in the number of HN mining apps posted. People who are engaged want more stuff from what they're engaged in.
I'm definitely not for information for information's sake, just find your tweaking counter-intuitive. Seems like shooting all of the animals to prevent them from escaping the barn.
I look forward to somebody telling us how the experiment went.
I don't see this as odd at all.
Greater amounts of Information widely dispersed == "good/better" and less/restricted information == "bad/worse" is a very dominant underlying assumption - sometimes so embedded in our intuition that we see restricting information as counter-intuitive.
If this experiment succeeds, then it might show that in some situations, the opposite is true?
As a thought experiment, suppose you could know the exact moment of your death. This is more information than you have today. Is this good or bad? (I can see arguments for both sides)
Suppose everyone knew. Would this be good or bad? (I can see arguments for both sides here too)
Yes -- it's a very interesting experiment and I'm really curious as to how it will play out. There's obviously a bit of mob rule mentality on HN and something should be done about it. Perhaps this is it?
I've always thought that in making a web site you make it completely intuitive and engaging -- then you incrementally add more detail and complexity to satisfy power users. Sort of like the qualities of a good game where it's easy to learn but hard to master.
But I'm probably off-base. As you said, there are a lot of underlying assumptions here -- on both sides. The beauty of experimenting is that you can move from talking about things to actually seeing what works.
I'd say, "Hacker News -- now with less data"
Less data could = less information.
Less data could also = more information.
We've all dealt with users who become more empowered to do their jobs once we "adjust their view". Now we're the guinea pigs.
Maybe it is good idea to display author's karma instead.
People love to see magic numbers. =)
Having you just take this away on a whim because you felt it "might" be better is pretty poor taste in my opinion.
It all most feels like papa Apple just rejected an iPhone app because they didn't like the "feel" of it.
Please bring back the comment scores ASAP. I can't envision myself staying round here without them ;-)
At a minimum, give me the option to turn it on for my own viewing pleasure.