1. 25
  1.  

  2. 14

    (For reference: ‘Fresh but boring’ means that a comment is less than 24h old, and its score is in the range [-2 … 4], inclusive.)

    Comment votes are to some extent a self-selected sample: I don’t think any of us vote on every comment we like or dislike.

    In my case, one of the things that gets me from indifference to participation is ‘score correction’: if a comment is at -1 and I don’t think it’s bad, I’ll vote it from -1 to 0, even when I might not vote that same comment from 1 to 2. This far into the experiment, I find myself making less votes on those ‘boring’ comments. Which is a pity: I rather enjoyed this low-stakes jockeying for a comment’s position.

    I assume the aim of this experiment is getting more votes on comments that might otherwise get few. Is that right, or are there other things you are interested in?

    1. 4

      I suspect this is in response to the comment voting which took place here: https://lobste.rs/s/tjp74a/xmenu_is_fork_suckless_dmenu_with_open

      1. 5

        Before the experiment started, I saw a lot of downvotes in that thread, plus complaints about downvotes. If this experiment is intended to prevent the conversation becoming about how people vote, then I am not sure it will work:

        • Even when the comment score is hidden, you can still estimate votes from your reputation change. As I write this, I think my comment above is at +4, because my reputation has gone from 85 to 88 and I have no other recent posts or comments. If I were downvoted, too, I could still notice.

        • Comment score becomes visible again at -3 and lower. Egregious comments get downvoted way beyond -3, in my observation.


        [I wanted to place this separate thought in a separate reply, but the commenting system won’t let me.]

        If the intention is to prevent piling-on, how about this idea? Comment scores could be clamped at the lower end to -5. Further downvotes are still counted, and balance out future upvotes; but any tally that comes out to -5 or lower would be displayed as -5, and have a reputation impact limited to (-5 - 1 =) -6. That’s enough to make the point that a comment is disliked; it is hopefully not low enough to incite passions; and feedback beyond ‘this is not a good comment’ generally happens through responses, dragonizing, or moderation.

        Some furthere reasoning on why to clamp negative scores, but not positive scores. +5 and +15 communicate the same thing, but +15 makes the author feel even better, which has no disadvantages that I can think of. (For a system which does clamp scores at the top end, see Slashdot.) Similarly, -5 and -15 communicate the same thing, but -15 makes the negative feeling stronger. Getting negative feelings are much less useful than getting specific feedback. If we limited negative scores to -5, there is a limit to negative feedback via downvoting; beyond -5, further feedback shall have to be in the form of responses that explain what is disliked.

        So if there is a perceived problem that this would solve, perhaps try this? If there is no problem, no need to try it, natch.

        1. 10

          By hiding the volatile score of a comment early on, I hope it will allow a discussion to play out without the parties getting distracted and whining about the voting of individual comments. When a comment gets 2 upvotes and 1 downvote, the user may never notice. If the comment gets 1 downvote before 2 upvotes, the poster can see it at 0 and get all bent out of shape about it, becoming accusatory and derailing the conversation.

          Egregious comments get downvoted way beyond -3, in my observation.

          Comments have been capped at -5 for a while. Now a comment will only grey out when it is between -3 and -5, collapsing at -5.

          1. 4

            “By hiding the volatile score of a comment early on, I hope it will allow a discussion to play out without the parties getting distracted and whining about the voting of individual comments. “

            Im 100% with you on that. Counters some of the psychological impulses that lead some (or many?) to try to improve votes more than content.

            1. 1

              By hiding the volatile score of a comment early on, I hope it will allow a discussion to play out without the parties getting distracted and whining about the voting of individual comments.

              That makes sense, and sounds like it could work for that.

              Egregious comments get downvoted way beyond -3, in my observation.

              Comments have been capped at -5 for a while. Now a comment will only grey out when it is between -3 and -5, collapsing at -5.

              Well. That’s very much my mistake, then. Sorry about that.

        2. 2

          (For reference: ‘Fresh but boring’ means that a comment is less than 24h old, and its score is in the range [-2 … 4], inclusive.)

          I’m seeing it on comments older than 24h. For example this one is currently 31h old, and has its score hidden.

        3. 10

          Another potential experiment that I noticed in use on YC News: you’re not able to downvote comments replying to yours (maybe also grandchild comments, I didn’t check). I could see that being useful for keeping things cool.

          I was also dismayed to see downvoting on comments where someone admitted a mistake. That should be encouraged rather than taken as a justification to pile on someone, and I’m glad it was corrected.

          A similar YC pet peeve is downvotes on a well-intentioned and non-fake-authoritative commenter who could not have known they were wrong based on the story and then-current discussion. I see them flood in after someone posts a correction.

          I don’t have ideas for tempering these latter two behaviors, but it’s regularly in my thoughts.

          1. 4

            In my experience, most downvotes in a big thread happen from lurkers, not the people involved in the back-and-forth.

            1. 2

              “Another potential experiment that I noticed in use on YC News: you’re not able to downvote comments replying to yours (maybe also grandchild comments, I didn’t check).”

              I like this feature. People arguing with each other are likely to downvote on reflex. The HN model essentially lets rest of audience peer review for reply downvotes. It seems to work fine on technical threads so long as nothing too divisive is said. If that happens, the effect is the same regardless: people throw votes back and forth in mass with audiencd mostly ignoring the flamewar.

              1. 2

                I’ll often explain my disagreement with someone and get a single downvote. I’d be quite surprised if it were anyone but the person I replied to.

                1. 3

                  I wouldn’t assume that. I’ve seen people reply to me, and get one downvote, and I know it’s not me. Inevitably get accused, though. I downvote very rarely. I think I’ve been accused of malicious downvoting more times than I’ve downvoted anybody for any reason. I’ve never seen a thread improved by assumptions about voting patterns. Just pretend it’s cosmic rays.

                  1. 1

                    “Just pretend it’s cosmic rays.”

                    New paper proposal for Black Hat: “Vote flipping for fun and proft. Or why not to store votes as a single bit.”

                    1. 1

                      Interesting. This is a strong argument for disallowing downvotes on responses to you though. The commenter could then know it was not you that voted it down, and it could help keep conversations civil. (Or at least stop accusations like these.)

                      1. 2

                        You’d think that, but on sites that implement the rule like HN, it just means the accusation adjusts to incorporate sock puppets into the plot. In the end, I find it’s a useful signal that somebody would rather argue about downvotes than the topic. Tells me when to take a break.

                        1. 1

                          You’re thinking in the right direction but not this example. Cosmic rays are one of the most ignored causes of problems in existence. The admins will look for everything but cosmic rays if they already used ECC RAM (checklist item) in one or most systems. It’s why the DNS “bitsquatting” attack was discovered decades after the known, mitigatable effects of bitflips on servers. ;)

                2. 5

                  A piece of feedback on the UI: seeing a - that is hiding a number makes me slightly restless, it feels ‘unfinished’ in a way. It feels the same when I see those comments with a central dot on Teddit. Does anybody else find their brain is bothered by hiddenness of the number?

                  I’ve tried replacing the dashes with a parenthesized (1). Here’s a screenshot. The parentheses are ugly, but seeing a number does help. Something like a grayed-out 1 might be nicer, but the numbers are gray already.

                  1. 5

                    Totally bikeshedding here, but I find the - slightly distracting too, especially as it is so close (spatially) to the collapse thread [-] element. I wonder if a nbsp, a question mark, or something like (unicode black circle), would work better visually.

                    1. [Comment removed by author]

                      1. 2

                        Or perhaps a question mark?

                      2. 2

                        best of all, if you’re not going to show the score anyway, just leave it as the default “1”; i.e. also hide the information that there might or might not be votes that have not been rolled into the score yet.

                      3. 1

                        Hah! I thought I was the only one who was slightly uncomfortable. I’d also like something other than a -. :)

                      4. 2

                        I just think the old way was good. This new thing, hiding karma really sucks.

                        1. 1

                          I also like that it short circuits the social media like pattern of people making content free but likely splash inducing responses just for the immediate gratification of a zillion upvotes.

                          1. 1

                            you still see if you’re downvoted, incidentally: a current comment of mine has a score of - but an annotation of

                            | +1, -1 incorrect