The basic philosophy underlying user-specific feeds on Hubski is that common pages often suffer from group dynamics as they become increasingly popular. That is not to say that everything about community spaces is bad and should be avoided (tags create community spaces), but they bring significant challenges, and can lead to a lowest-common-denominator effect.
Feeds are customizable and individualized on Hubski. Comments are not, with one exception; currently, if you ignore or mute a user, their comments will tend to sort to the bottom of a thread when you are viewing that thread.
As Hubski grows, the shared space of comments may be strained. Easily accessible comments may accrue more votes than more thoughtful comments, and a lack of customization might result in less interesting comment sections overall.
We are considering an experiment that would bring individual customization to comments. Here is the general idea:
When viewing a user's profile, you will see a simple rating device, with settings that represent: 'poor', 'neutral', and 'good'. The neutral setting is selected by default. You can opt to change the setting to 'poor' or 'good' if you like.
Selecting 'poor' will tend to sort that user's comments to the bottom of a thread. Selecting 'good' will tend to sort that user's comments to the top of a thread.
Rating would be individualized, and anonymous. For example, I would not know how insomniasexx rated me, and how she rated me would affect her comment sort alone.
This is the seed of an idea, so we decided to share it with you, and to get some feedback before we ran a trial of it. It's likely that some alternate implementation might be even more useful or interesting.
Your thoughts are appreciated.
Two things: 1. I like this a lot, especially as the beginnings of a function that may change with use. 2. My favorite thing about hubski is that this hasn't happened yet. The comments with the most circle-shares still generally correlate well with the longest comments. Length isn't an indicator of quality by itself but it is an indicator of effort, so I'm happy to see this pattern as hubski grows.As Hubski grows, the shared space of comments may be strained. Easily accessible comments may accrue more votes than more thoughtful comments, and a lack of customization might result in less interesting comment sections overall.
I'm gonna be the nay sayer here. I'm skeptical that this is a value added feature. While I like individualization, I don't like things that are inherently negative. I'm not sure if this fits that bill, but it seems awfully close. I would prefer some kind of system where people whose comments you consistently upvote are biased higher in the sort, for example. Although I will say, I was against mute at first, and that seems to have worked out well.
I think this hits on an important point. Individualization leads to decentralization -- which gives every user more power to control his/her online experience -- which is good! However, having an aspect of the site that may make users fearful to share ideas, post an article, get into an argument... is something Hubski has always tried to avoid... and I think this is wise as well. Maybe there is a middle ground? Maybe we could have a system where you can (privately) "star" commenters you appreciate. This would sort of mimic the philosophy of the Hubwheel. The Hubwheel is only positive (there is no equivalent to the "downvote"). So maybe we should have the same thing for an individualized commenting system?While I like individualization, I don't like things that are inherently negative.
b_b and I were talking about it, and functionality-wise the most useful signal would actually be to indicate the commenters you definitely didn't appreciate. That is, if I had to pick the people I did appreciate, it could be either a very narrow, or very broad selection, and drawing the line might seem arbitrary. However, it's always pretty clear to me when I don't appreciate a certain commenter. Although I do worry that it could create a negative vibe by having that decision to make.
It's a tough call. I'm not sure if any comment platforms already have individualized comment sections so there may be no template to base this off of. Could you instead just design algorithms that "know" what commenters you frequently "hubwheel" or respond to?
I'm not sure it will either, but I think it'll be a valuable experiment to see what it does. I raged at mk about this earlier, primarily as blowback against the Reddit admins that aren't even there anymore, but experimentation in social networks is crucial because there's no data whatsoever. I'm going to predict it's an interim method that won't end up being adopted long-term, but I will also predict that if properly studied, it will lead to discoveries of methodologies that will help.
My only concern with this is that it's another feature that I have to maintain. I really like the concept though and I think it will have an interesting effect on the experience but I don't know if I or future new users will choose to maintain the feature. Like b_b noted, if there were some way to have this done automatically based on past votes on comments, that would be infinitely better. I have no idea how feasible this is but perhaps if I vote on someones comment 5 times in one month, it elevates them to good status. If I don't vote on them at all for a couple weeks it would demote them back to neutral status. I could still change the settings manually of course and I would always have to manually switch the position to "poor." Very ideally, it would be cool if the sliders had the same sort of gravity the feed had. Probable problems: 1. If someone else or I simply aren't around Hubski for a period of time it could ruin the ratings. 2. Processing/server/coding issues of implementing this feature 3. Getting the proper number of votes per time period that fits a wide range of people's hubski habits. Something that may mitigate these problems would be a "lock" button. I could manually lock kleinbl00 and _refugee_'s comment reputation in place regardless of my interaction with them while others float around. I also really like kleinbl00's point: I agree wholeheartedly. Further, I believe that if done properly, this feature could really make Hubski stand out. It really changes the way a comment section works. It's not chronological or most voted on: it's particular to each person.I will also predict that if properly studied, it will lead to discoveries of methodologies that will help.
It seems like there is some desire to see an automatic functionality, perhaps one that you could turn on and off. The difficulty is that although it is not too difficult to discern what commenters you appreciate, it is not as easy to distinguish the ones you are neutral towards, and those that you don't care for. At some point it is helpful to know what commenters you do not appreciate. However, only identifying the ones that you do would still likely be an improvement. I also agree with kleinbl00 on that point. This might reveal other possibilities. doesntgolf makes the point that you might eventually have to rate too many commenters to make this worthwhile. However, it is possible that we could turn this into a functionality that combined the signals of multiple users. For example, if you and I rate a commenters similarly, part of your bias could be used for my sort. Still, I am not sure if that would be a good thing. It does however, create a signal that we don't currently have. I do think that it might be easy to under appreciate the feeling of being able to do something to improve your comment sort. I'm not sure if it would be the case, but I suspect that it might be an alternate outlet for what might have been a negative comment.
And I've been thinking about this and I realize I might be wrong. I keep harping about scalability but lil's discussion here has made me realize that all my favorite communities are small. The idea might not be to make Hubski infinitely scalable so that it never gets that big feeling. The idea might be to say "enough is enough, we have a waiting list for participation." Spitballing at this point, but I'm slowly realizing that if what you have works right now, it might be smarter and simpler to lock it down than to buttress against it not working a year from now.
We've kicked around the idea of a very small subscription fee, something on the order of a dollar per month, with the thought that only the people who really want to be here will pay. I'm against it, however, because I think that we stand to lose rather than gain in that scenario. People really don't like to pay for things, and I think rather than making a small barrier to entry, a fee would just drive everyone away. We also talked about an invite only model, whereby members would earn invites based on participation. That idea is cool, but there's also some problems with it. The thing is, we want to encourage more people to join up, as the whole point is lively discussion, but we really do want the right people. It's a difficult problem.
I know what you are saying. It's funny, just prior to the syncretic influx, we had almost talked ourselves into an invite system. I recall that in the wee hours while driving to launch that snail in a weather balloon, we were talking about the merits of it. I was inspired by lobste.rs invitation tree approach. It would be interesting if rather than just earning badges as your hubwheel turned, you earned invites. I think that ideally, rather than an aggressive front end filter, we have a situation where the right people stick around, and the wrong people don't. I think that we have been fairly successful there, but it's the slow creep that we all worry about.
Designer news has it set up that you can only sign up between noon and 1pm every day. This makes it so the people who really want to join can join without jumping through invite hoops while still keeping out those who want to make a one off comment or spam.
No it's 12-1 EST. It took me like 5 days to remember to do it when I got to work. Kind of annoying but I'm sure it helps.
I don't know if it's a good idea or not. I guess the only way to find out is to test it out! I'm all up to give it a trial run for a couple of months and re-discuss it later. I've actually been wondering how comments are sorted in Hubski. Is it some kind of mix between how many votes the comment has and how recent it is? Are the number of replies taken into account? I'm pretty curious about how the current system is set up, is it explained in an older post somewhere? Are you thinking of implementing it as an option or will everyone's comments change to that? If I'll have to rank the people by myself, i'm guessing if I want to opt out I just put everyone on neutral. If it's some automatic feature like b_b suggested, the question whether to make this an option or implement it on everybody comes up. Anyways, thanks for asking for our opinion :) I think everybody wiil agree it's nice to feel we have a say in the matter!
Currently comments sort based on time, and the number of votes. This would not change that. Using this rating mechanism would only bias the sort for the users that you rated something other than neutral. So, if you didn't use the mechanism, you would see no change from the current sort. I've been considering that the bias won't change the score of the comment, but the gravity on it. All things being equal, newer comments sort higher. This would cause the comments of a user that you rated 'poor' to fall faster, and the comments of a user you rated 'good' to fall more slowly.
I think just make sure that it is well understood at this point of selection how this pertains specifically to comments alone and that they don't see your rating. Maybe the name of the toggle itself will reinforce this? Could be called something like "X's comment reputation" or something like that? With all of the sharing and voting in different areas, I think the biggest hurdle could be just mitigating confusion over the feature's parameters. Definitely for it though.When viewing a user's profile, you will see a simple rating device, with settings that represent: 'poor', 'neutral', and 'good'. The neutral setting is selected by default. You can opt to change the setting to 'poor' or 'good' if you like.
I absolutely agree with this, and I'm up for suggestions. HTML5 has input sliders, and I am thinking about using that. Regardless of how we label it, I think it will have to have a '?' next to it which brings up a popup explaining the functionality in detail.With all of the sharing and voting in different areas, I think the biggest hurdle could be just mitigating confusion over the feature's parameters.
I'm a fan of this idea. As a general principle I'm a fan of individualized features. thenewgreen always states that everyone makes their own Hubski. It would makes sense to allow people to do this with Hubski's most important feature, and most marketable asset: the discussion.
Alright, I've done some more thinking about this today. How about something along the lines of: implementing a reputation/karma system (which, IMO, wouldn't need to be displayed publicly on profiles, but that's not too relevant.) The higher a user's reputation/karma, the more their upvotes count for in comments (and in posts? IDK). I feel like this is a half-baked idea, but is a potential solution for combating the lowest-common-denominator content from rising. The 'Old Guard' so-to-speak would have more sway in their sharing/voting.
Sounds like a "power user" mechanic.The higher a user's reputation/karma, the more their upvotes count for in comments (and in posts? IDK).
FYI, we implemented what we are calling "Hush", any feedback is most appreciated. Also, on a completely unrelated note, you and I have talked before about how much we loved the show the West Wing. I recently started watching The Newsroom and while I'm only two episodes in, it does have some of the familiar pacing that the West Wing had. -Sorkin's famous walking fast dialog thing. I'm still not sure if I'll like the show or not but so far it's decent. Have you seen it?
Yes, I've watched the first season, it's great. I was hooked from the very first episode where he gives his "America is NOT the greatest country" speech. Very preachy Sorkin stuff, loved it. The love triangle stuff on the show is a little departure from normal West Wing stuff, but it's tolerable. It's a good show, only seen the first season when we had a free HBO preview weekend here. It's pretty good, no West Wing though, but fills that same preachy intelligent ranting niche. :)
This seems like a feature that would be a good fit for Hubski in it's current state. But if Hubski is to keep growing at a pretty rapid rate it seems like both this and the mute/ignore features would be more and more useless. If I have to continually go through and mute/ignore/rate-poorly each user that posts low effort crap then it's going to be a lot of work and at some point it just won't be worth the effort. I admire the 'ideal' that everybody can make Hubski their own, but I also enjoy a good classic 'wild west' forum where everybody sees the same thing. I think that's partly what 'global' is right now. I guess what I'm trying to insinuate is that I wish there were some way to just inherently give better ranking to thoughtful, high-effort content, and discourage low-effort stuff. I don't know.
It seems that the idea of something like this is, so far, universally appealing. Some like the idea if it being automated based on previous decisions while some welcome the ability to custom curate the feature. This distinction seems to be the question. Ideally, I think having it as a feature that Hubski sorts out for you would be ideal. Also, being able to turn it on/off would be good. I like it mk.