Collapsed comments?
I'm starting to see occasional collapsed comments when I open threads. The first time I saw one, I thought I'd accidentally collapsed the comment myself, but this one was definitely already collapsed when I opened the thread.
Is this related to the "tagging" system that Deimos discussed the other day? At the time, he said that "tags" wouldn't have any effect. Are these effects now working? What are the effects? Also, there was no actual decision about what the different tags would be, and what they would mean. As I said in that thread, I decided not to use these tags until: a) they were agreed and defined; b) they actually did something. I thought they were just placeholders for now. I'm confused.
Or is there some other feature operating here? For example, people keep talking about "whisper" comments, and I'm not sure if they're just fantasising about a feature they would like, or if it's something that's actually planned. Are these those "whisper" comments?
So, the "noise" report option is working? Good to know.
Thanks.
Well, it's not a tag option, either, because I don't see a tag saying "noise" next to the collapsed comments. They're just collapsed without any explanation. They're not tagged as "noise".
I'm going to keep saying this until you're all sick of it (if you aren't sick of it already): you can call this a "tag" feature all you want, but it looks like a report feature, swims like a report feature, and quacks like a report feature. It reports comments to the system for the system to perform actions on.
I've been trying really hard to avoid engaging you on this because I know it's just going to be a waste of time (on both our parts), but no, this has almost nothing in common with a reporting feature. I understand the argument that "tag" might not be the best name for it (even though all "tag" actually means is "label" or "classify", and that's basically what this is), but "report" is a far worse name.
Reports are sent, reviewed somehow (usually by a human), and potentially acted on or not. The comment-tagging system has very little resemblance to that, and calling it "reporting" would just be ridiculously confusing for everyone.
Yeah my biggest issue with calling it "report" is that it implies manual review of said report before any action is taken but that isn't what's happening. If you call it report, when people "report" something and nothing happens to the comment they "reported" when they check back later then they may assume their "report" was ignored, which simply isn't the case. Tag isn't perfect, but until someone suggests a better word for the mechanism it's the best/closest we got.
There were a couple of suggestions made in the previous thread. The most common one was "flag": we users are flagging these comments to the system as potential problems.
Yeah I saw that but if/when any positive "flags" are added, that kind of loses what little accuracy it has over "tags". It's basically going to be a hybrid tag/flag/report/sorting/filtering/crowd-sourced moderation system so no one word is likely going to be perfectly applicable, IMO, but at least tags is generic enough that it mostly works in all the cases and generally understood enough that it isn't going to cause too much confusion.
It does have the advantage of differentiating this comment option from topic tags, which operate differently.
Sure but flags generally has a negative connotation at least in terms of its common usage on the web, which isn't necessarily going to be the case with this system. And as I said, tags isn't perfect either but I suspect no one word is ever going to be. But I would be happy to be proven wrong if someone can come up with a word that is. ;)
Could maybe have two separate buttons for negative and positive tagging actions.
Flags: Noise, Flame, Bad Faith Argument, Idiot.
Tags: Humour, Kudos, Delta, Charitable
(I am super just spitballin here)
Actually, no, it wouldn't be. I've been hoping that I might start convincing people, and ultimately change your mind while this feature is still in its embryonic stages (because I'm certainly not going to change your mind when it's been operating for months!). However, if you tell me "they're 'tags' and that's not negotiable", then I'll know that my campaign is pointless. You're the boss. I can persuade, argue, and debate, but I don't get to make the final decision.
It also has no resemblance to any tagging system. The comments are not tagged with anything. I click on "noise", and it doesn't get tagged for other users to see.
But... isn't the website reviewing these comments and then, most importantly, acting on them? The whole point is that reporting a comment as "noise" will now get the system to act on that comment. I honestly don't see how that is not a reporting feature!
I've been avoiding saying this, but I think you're all approaching this from the point of view of computer developers, rather than from the point of view of users. You see that you're adding a field/tag to a record in a database, so you call it a tagging feature. However, functionally, it acts as a reporting system from the users' point of view: they mark a comment for review and action, and the system reviews the status of the comment ("Does this comment have enough reports to reach the threshold for action?") and then acts on it ("Hide the comment.").
Alright: The name may change from "tag". The name will not change to "report".
I'm not opposed to changing it, and I think I'd even welcome it since using a different term to more easily distinguish them from topic tags would probably be a good thing. The trick is finding a term that doesn't have its own implications or other baggage that would be even more confusing than the ones "tag" has.
It's similar to the debates that have come up before about whether using a term other than "vote" would be a good idea—I agree that "vote" isn't a perfect term, but I think every alternative people have suggested has been even more confusing or inaccurate. Both "comment tag" and "vote" are definitely closer to "the least bad option" than "the best option".
After browsing through this thread a bit, I'll suggest "mark" as an alternative. A bunch of people are already using the verb "mark" when taking about them. It's not quite neutral like tag is, being slightly negative (at least to me). I don't think there would be any confusion with markup/markdown and its still probably less confusing than comment tags and topic tags.
"Mark" isn't bad. I think my concerns toward it would be ambiguity about what it means, and potential confusion with "bookmark". That is, when someone sees a button that says "Mark" on a comment, what do they assume that clicking it will do? My general feeling is that they'll be more likely to believe "that bookmarks the comment" than "that lets me add some information to it".
It's definitely the sort of thing that they'd learn quickly anyway, but I'd be mildly concerned about it being common for people to be initially confused by it.
I'm still conceptualizing with "filters", as in tagged criteria for delimiting search or view, versus "flags", as in indicators for action. Noise gets filtered, trolling or other flagged malicious behaviour has human intervention.
"Mark" seems okay: "Users can mark a comment as 'noise' to get the system to hide it."
What would be the noun version? "I'm applying a 'noise' marker to this comment." Would that work?
Mark is a noun too!
I don't think the noun use would come up too much(outside of meta discussion at least), but If I had to I would say
In reality, I would just go with
I think it works fine as a noun in all sorts of situations
Thank you for making that clear enough even for a duffer like me. :)
If you're looking for another option, there were a couple of suggestions made in the previous thread. The most common one was "flag": we users are flagging these comments to the system as potential problems.
I figure once privileged users are able to filter by "noise" or "troll" for moderation purposes, and regular users will be able to filter by "joke" for entertainment purposes it will indeed become a tagging system.
Meta tags, used in HTML documents to attach meta data and search terms to pages, are not present to the user. Granted they are not completely inaccessible, but there is some precedent for the word "tag" referring to information used by machines for sorting and kept out of sight of the end user.
Ah you're right.
But I suppose that it still means that HTML "tags" refer to typically-unseen, machine-informing information?
Is there a way to see what you've been tagged with?
Yeah I'm curious about this too.
I see a lot of people flapping about saying its not a report system and trying to figure out how to word it, but I can't find anyone mentioning what the actual issue with listing what tags a comment has. Especially after the comment has been actioned upon.
I'd really like to see what things were moderated for. or reported as, or tagged flagged marked booped or whatever the fuck it ends up being called. Because otherwise it'll be easy to just assume it was a collapsed comment from a previous visit.
The reason that they haven't done this is to prevent bandwagoning and bias. Somebody sees a comment marked as Noise and it causes a bias when reading the comment, when those tags could have just been somebody disagreeing
If the comment is already collapsed when you open the page then the bias is already there..
Yes, I'm starting to experiment with applying some effects based on the tags, but nothing's solid at all yet and I'll probably keep adjusting (or reverting) it based on how I see things getting tagged in practice. The "noise" effect is the only one that's currently enabled since that's the only tag I've seen get used very consistently so far, but I have some other adjustments that I'll likely apply over the next day or two based on what I see.
Is there a way to mark it not noise, or does a certain number of votes override the Noise tags' effect? If not, all it takes is a couple people to disagree and the comment is collapsed for good.
There's no way to "counter" tags right now, but that will probably be something that's added eventually and used as part of determining how trustworthy certain users' tags are. If their tags get opposed often, that most likely means that we should put less weight on their tagging.
"Noise" doesn't mean "I disagree", and if people are repeatedly using it that way, they'll probably lose access to tags or have their tags weakened. I'm monitoring how people are using the tags pretty closely for now.
I frequently discuss things on Tweakers.net, a Dutch tech website, which has a fairly unusual but pretty cool comment vote system (that I think this place should consider using, at least for comments!). Here is the faq section in Dutch (if you click on 'moderatiesysteem' at the top), but basically what it comes down to is that you can vote with a certain amount, depending on what you think a comment adds to the discussion, which is pretty much what we're going for here. Here's a rough translation/summary of what it does:
I don't know if that helps, but it makes the tag feature only applicable if someone scores something a 0 or -1, I suppose. And I think since points/karma/whatever isn't counted here it might make an interesting alternative for comments.
fwiw they have some system of ranking posts based on this, where comment threads with +3 will go to the top, followed by high-average discussions (like a thread with a bunch of +2's). Furthermore they have a toggle in the discussions at the top that allow you to see a certain score and up. -1 is collapsed by default, but that way you can click +3 and everything except +3 comments are collapsed for that comment section.
Thanks, that's really interesting. I'm reading through the page now (with Google Translate) and they're doing some neat things. Using an average for the post's score instead of a total, having "tiers" for the users as well where their votes have more weight on the average, and "boost points" that let people increase their weight for a limited number of votes.
I do think we definitely need some ways to vote more positively on posts. Right now there's no ability to express any difference between "thanks for replying, here's a vote to acknowledge that I read it" and "this is one of the best comments on the site".
Looks like it's time to dust off my old two-click voting suggestion image once again... You had to know that was coming. ;)
Of course updated to the new "text based design" goal for Tildes it would be:
vote
-> click once to open list of vote reasons (similar to how comment tag works now) -> second click to register vote + selected vote reasonI think the 4 options in your image are very limited. But I quite like your proposed implementation of the system.
Yeah that image is from over a year ago, before the site was even up and running yet and before Deimos had decided on text instead of icons as a design philosophy. Those 4 options were also basically just "inspired" from Slashdot and were just meant to be rough examples. And I think by it being pure text now too that actually opens up a lot more possibilities since you can have more than 4 options unlike with that radial menu style where space is limited.
Thanks! Do you want me to post this as a separate topic to get discussion going?
I'm slightly curious what your ideas are for the other tags. Like if there are so many flames what are you thinking the outcome for that would be? Also is there a doc related to when certain tags are expected to be used. Guidelines I guess. If not are there plans to make something like that?
I don't know for sure yet, but there are lots of possibilities. Getting a comment tagged heavily as "flame" could do things like report it automatically to admins or higher-trust users, remove it automatically for review, rate-limit the author's commenting temporarily as a "calming down period", and so on.
I think a good default for the noise tag would be if whether the comment is collapsed or not depends on the ratio of tags to votes. Statistically, the more popular a comment, the more likely it will be to get tagged regardless of applicability.
Good point.
If there were, say, an "insightful" tag I might have tagged your comment with it.
I keep advocating for more positive tags(/flags/marks) in the style of Slashdot, but I've slowly grown more ambivalent.
The distinction is that we already have a general "vote" for expressing positive sentiments, and the tagging semantics should probably be kept orthogonal to vote/not-vote semantics. An alternative might be to drop "votes" entirely in favor of choosing one tag from a set like [Insightful, Informative, Funny, Noise, Troll, Flame, Spam] (this is closer to the Slashdot approach). This would probably move the sites design/ux a bit farther away from the reddit lineage though.
I'm not sure how much I mind that, to be honest. It would be nice to know when people appreciate my comments, of course, but if I got more specific feedback I'd be happy.
Not for good. If we spot misapplied tag actions we can report that to Deimos who can undo them and then warn/punish those who misapplied them. If someone consistently misapplies tags they can have their privileges revoked or in egregious cases they can be banned. Tags are a lovely little honeypot, IMO. I do agree that scaling the amount based on some metric (e.g. # of comments, user traffic, etc) is probably going to be best, but I imagine that is what Deimos had in mind anyways... this is just the first rudimentary step towards that to see if the idea of auto-actions on X number of tags is even viable.
Aren't you like 3-4 moderators (or is it only Deimos) for the whole site? Can you really moderate the activity of hundreds of active users and the corresponding comments?
It depends what you mean by "moderators". This site will have quite a decentralised moderation model.
For example, there are quite a few people who can edit tags on topics, and move topics to different groups, and likely some people who can edit titles. These could be considered moderation-type functions, and they're spread across a lot of users.
As of this week, most people can apply tags to comments, which may result in a comment being hidden - another moderation-type function.
However, for everything else, there's only Deimos: Deimos is currently the only person who can lock topics, remove comments, remove topics, or ban users. This will likely change as the site continues to develop, and Deimos continues to decentralise the various moderation-type functions. But, for the moment, it's just him.
If this site had some monitoring/analytics, we could set percentage - for example comment is collapsed if at least 20% of people marked it as noise.
But without monitoring, this won't work, as if there is 200 comments thread, and 5 out of 7 users who seen a comment mark it as noise, it still won't collapse because another 500 users seen the thread but not the comment. And users are more likely to not tag comments if it's already tagged.
I don't think NotNoise button would be the best thing possible, as 90% of the people would be neutral on it: they won't tag it either way, so it would turn into fight: this-is-noise against this-is-not-noise users. These voting users would have 100% influence on the post collapsing and I think, that the 90% of users who don't vote should affect the comment collapsed state as well.
Can't it be based on a ratio of score to noise tags? Maybe something like:
Exactly what I was thinking. If significantly more people approve of the comment than think it's noise, it shouldn't be collapsed.
Although, there should probably be a lower limit on Noise tags causing collapse, something like:
I agree. It was just that the condition then gets pretty long, so perhaps a different way to lay out that logic would be preferable.
Once there's enough data I think @Diemos should build a histogram of
tag_counts["noise"] / num_votes
and determine the appropriate split point.That's good idea! I think ratio should be something like 4:1 tags:upvotes (so
... > 4
), so it's harder to collapse something (or 2:1?), but otherwise I really like this method. I wonder what @Deimos thinks about it.I don't think anything would ever get hidden if you need 4x the amount of votes. The ratio should be backed up with data analytics, but I don't see it going above 1:1.
Sure, it's certainly possible. Trying to figure out the threshold is difficult though, and that's why I just want to start testing things and see how the tags start getting used in practice. So far even getting 2 noise tags on a comment is very rare, so it doesn't need to start getting fancy just yet.
At the moment, based on the code that @Bauke linked to in their comment, it's a very simple rule: if a comment gets tagged as "noise" by two users, the comment will be collapsed. That's it.
There may be more subtlety added over time, but that's the "first draft" version of this rule: "2x noise = collapse".
Hmm, having clicked on that link, perhaps a directly linked comment should always be expanded upon loading the page, regardless of the tag status?
I said in your earlier thread that I didn't see any point in clicking on "tags" that don't do anything, so I've never clicked on any of them. Could you please let us know which report options start becoming operational, so I know which one(s) to use?
We didn't even finish agreeing on the definitions of the various tags, or which tags should exist.
Perhaps we'll know better how to use them once we have a better idea of how users use them organically?
I kinda feel like it's overly prescriptive to define these things, decide how they should be used, and what they should be called, before we realize what the community is naturally going to do with them. I kinda like how this site is playing certain things by ear, to be honest. Throw things at the wall, see what sticks. Take feedback from the community, both through comments, but as well through data analysis.
If it doesn't work, we'll come back to it or just scrap it. If we find people are using it for a completely different purpose, let's roll with that instead.
As I said in the post, seeing how people are using them is part of how I'm deciding when to make them operational, and what exactly to make them do, since I can consider it from a perspective of "if this comment had been affected in a particular way because of these tags, would that have been a good thing?"
I'll probably end up enabling, disabling, and tweaking the effects quite a bit as I see what happens. If you want to help with that you can tag things when you think it's appropriate, but if you don't want to use them without knowing exactly what they'll do (or even if they'll do anything at all) that's fine too, it just gives me less data to help make those decisions with.
Okay. I'll start clicking on these "tags" so you can include my data.
My kneejerk reaction to this is a negative one, and I feel pretty confidently you have just re-implemented Reddit's downvote system only split in 2, comment order is whack, and downvotes get preference.
Consider a comment on Reddit gets voted up a fair bit, but downvoted more. If it hits a certain threshold of negativity it becomes collapsed for users who have that threshold set on their accounts (by default). This is obviously core Reddit vote functionality (the other being the ordering of comments and posts).
Now consider the combined effect of tilde votes and noise tag. Votes move a comment up on most sorts. But now, noise tagging acts as the downvote threshold and hides. All the old Reddit issues with downvotes suddenly appear, with the noise tag becoming the disagree button (which we already saw with a much smaller community with the flame tags and such, even when they did nothing). Only without the ability to be counteracted by votes, so things which are controversial and get loads of attention both shoot up the pile and get hidden. But even if there is eventually a counter-effect somewhere you end up with the same circlejerk effect of disagree voting and any dissenting opinion gets hidden from the echo chamber.
There is much more groundwork to be done before noise tagging should do anything to posts IMO.
The trust system has to start somewhere. As said elsewhere in the thread abusing this system will lead to your tags having less weight or your ability to tag being removed. In extreme cases I'm guessing it will also lead to bans.
In the future I'd assume trusted users will have to ability to unhide a comment that has been improperly tagged.
While I was in the same boat as you when I saw this and had a knee jerk reaction of 'oh so downvotes?' Just thinking about it more makes it make a lot more sense to me and is going to be a core feature of the site. Trust.
I have some suggestions as to tag visibility
Tag
link-button, the current tag counts are shown if non-zero, e.g.Noise (2)
instead of justNoise
Tag
)I've noticed it too. Don't disagree with the idea in general, but it's hard to tell if something is collapsed because I already read it or if it's marked as "noise". I'm pretty sure I saw a comment earlier that was new, wasn't marked as new, but collapsed. Haven't confirmed if that's what it was yet, or if I just missed it when I first visited the thread.
Yes, it would be very helpful if those comments were marked as 'noise' in some form @deimos.
The interactions between noise collapse & comment highlighting were pretty wonky as of last night.
Oh it's still real wonky. The whole concept of combining multiple different processes that want to collapse/uncollapse comments in different ways is kind of hurting my brain overall and it's hard to figure out good ways to handle it.
I think noise needs to be treated more harshly, in some ways quite a lot like delete. If it's noise (or whisper) it shouldn't trigger unviewed notifications & shouldn't get comment highlighted. Maybe worth having generic code for things which are there but in some ways not fully there?? Like Delete, Noise & Whisper, that could then be per-user optionally applied to other stuff so I could have a setting to opt-out of highlighting of humour comments I wanted?
Something's not right. I just clicked back into this thread and there are 10 new comments for me to read since my last visit. A full 7 of those are auto-collapsed:
None of these are flippant, or off-topic, or in any other ways objectionable. I've been noticing this happening in other threads too: Red-striped (meaning new, unread) comments that are auto-collapsed, but are actually substantive and good. In my opinion this should never happen. I'm legitimately really concerned about this turn of events and I'd like to know why it's happening.
Yeah, I'm working on it (and I think it should be fixed now). It's not related to them being tagged noise, I just screwed up some of my collapsing logic for old comments and it was also collapsing "leaf" comments (ones with no replies) even if they're new. @spit-evil-olive-tips
It would probably be good for you to edit out all those mentions though, you don't really need to notify 5 other users because I messed up the comment-collapsing logic for about 10 minutes.
And try to relax a bit, jumping to the conclusion that this was some kind of censorship issue because there was a minor temporary collapsing bug is completely over-the-top.
Duly noted, thanks for the quick response and sorry for the tenor of my write-up. I've removed the mentions, not sure if that recalls the notifications after-the-fact.
Didn't mean to imply it was deliberate censorship, I was just concerned that the recent tagging system changes were producing some unintended consequences, and without knowing whether that was "working as expected" or a simple bug I wanted to check in.
It should. I'll take a quick look to confirm that it removed them, because that would be a bug if editing out (or in) username mentions doesn't interact with the mention notifications. (Edit: yes, confirmed it's working)
And it's fine, but like I said: it's an alpha site where you're obviously aware that I'm currently working on collapsing-related changes. If something seems out of whack it's probably more reasonable to assume that it's a bug than go immediately to "legitimately concerned about this turn of events" levels of framing about it.
Editing in mentions works too. A few days ago I was pleasantly surprised by that feature. ;)
Definitely noticed this happening as of today.