Many updates to The Feature Formerly Known as Comment Tagging
A couple of weeks ago, I re-enabled the comment tagging feature. Since then, I've been keeping an eye on how it's being used, reading all the feedback people have posted, and have made a few other small adjustments in the meantime. Today, I'm implementing quite a few more significant changes to it.
First, to try to head off some confusion: if you're very new to Tildes, you won't have access to this feature yet. Currently, only accounts that are at least a week old can use it. Also, the docs haven't been updated yet, but I'll do that later today.
Here's what's changed:
-
The name has changed from "tag" to "label". I think it's better to use a different term to separate it more easily from topic tags since the features are very different, and "label" shouldn't have the implications that some people attach with "tagging".
-
As suggested by @patience_limited, "Troll" and "Flame" have now been replaced with a single label named "Malice". I don't think the distinction was important in most cases, and the meanings of them were a bit ambiguous, especially with how much the word "troll" has become over-used lately.
Basically, you should label a comment as Malice if you think it's inappropriate for Tildes for some reason - whether the poster is being an asshole, trolling, spamming, etc.
-
This new Malice label requires entering a reason when you apply it. The reason you enter is only visible to me.
-
Another new label named "Exemplary" has been added, which is the first clearly positive one. This label is intended for people to use on comments that they think are exceptionally good, and it effectively acts as a multiplier to the votes on that comment (and the multiplier increases if more people label the comment Exemplary). Like Malice, it requires entering a reason for why you consider that comment exemplary, but the reason is visible (anonymously) to the author of the comment.
Currently, you can only use this label once every 8 hours - don't randomly use it as a test, or you won't be able to use it again for 8 hours.
The interface for some of these changes is a bit janky still and will probably be updated/adjusted before long, but it should be good enough to start trying them out. And as always, beyond the interface, almost everything else is subject to change as well, depending on feedback/usage. Let me know what you think—comment labels have a lot of potential, so it's important to figure out how to make them work well.
There's a little (very, very small) UI thing I noticed with the
Exemplarytaglabel as well, so putting that here instead of a new parent comment:The One per day text isn't vertically centered.
Literally unusable.
That is, until I click Reply on a comment. The comment window then expands by just enough to vertically center the text. This happens whether the Label button/section has been opened or not.
As I said, very minor.
True, thanks. I think an outline like @arghdos suggested might work well, I'll try some things out.
Solarized's green color really is quite bad. I could switch it to blue (edit: did that), or the cyan that Joke is currently using. From Solarized's "accent colors", we're also not currently using magenta or yellow for anything significant, but most of the others are already in use on some level: https://ethanschoonover.com/solarized/
(@MetArtScroll and @balooga since you made a similar request too)
As long as green and red aren't next to one another (for us colorblind folks, it just looks annoying and the same) I'm good with the colors.
Looks like @deimos was trying to go from green->red in steps.
Perhaps he should just make it more of a traffic stop green to maintain the color palette and make it more appealing.
Any reason why it can't just be
Exemplary (1/day)so it's just one line? There's plenty of horizontal space.Mostly just that it looked pretty weird when I tried that. There's also a lot less horizontal space on a phone (though it could probably be made to wrap only on phones too). I'll tinker with it a bit more.
Thinking about it some more, why not even just
Exemplary(1)when you have one andExamplary(0)when you don't? I think this will accommodate allowing people to have more than one at some point, and they can dig through the docs if they're wondering why it hasn't refreshed yet.Wouldn't this read more as someone else has give it an
Exemplarymore? Maybe it needs to have a1 remainingor1 of 1. (Though I'm on mobile and not sure if there's room for either.)I mean specifically for the menu that pops up when you click "label" - not what's shown when a comment is flagged.
Ah my bad, I misread what you meant. Thanks for clarifying.
Ah tru, I should know better than to forget mobile
Not so much on mobile.
good point, I should know better
I think a light outline around it might be cleaner, e.g.
Very interesting, thank you. I especially love that the exemplary tag comes with a reason, so people can see feedback on why their comment is labelled especially good. Very pleased to see the progress Tildes is making! :)
I like the change to the two negative labels. It is a much better explanation of what constitutes bad behavior. I have probably used Troll in places that you didn't intend me to. Sorry for making your job harder.
I feel that Malice should probably also have a countdown timer, though that might not prove ideal as it could prevent people from reporting offenders who are spamming comments. Perhaps a much shorter one would work, maybe an hour or less.
I especially like that "Malice" is a judgement about a single purposeful action. Deciding that a comment is "malicious" (i.e., "harmful") is completely different from finding it "contentious" or even "provocative", which can happen through carelessness. The latter can sometimes contribute to healthy discussion if treated cautiously. The former cannot, by definition.
Thanks! They all sound like good changes. Looking forward to trying them out.
Hard 24 hours or is it reset every midnight?
The next one is available 24 hours after the time that you use it. (@Bauke)
Edit: this has been changed to 8 hours now.
Ugh. May I suggest a 23 hour timer so that it doesn't functionally float later and later in the day? Assuming you want it to be able to be used once a day, someone using ~ around the same time each day would eventually have the timer push the label outside of their normal ~ usage window.
Sure, I don't know if the one-per-day limit will even stay around overall, but bumping it down a little bit is easy. We could probably do a reset at a specific time as well if it makes more sense, though that can be weird in some other ways.
Honestly once per 8 hours would probably result in being functionally the same as people tend to browse during blocks of the day.
Sure, let's try 8 hours for now. Having people use them a little more freely probably isn't a bad thing, and we can always re-adjust later.
Eight fits pretty comfy in the 'one per browsing session' ballpark.
Congrats on another milestone, because today Tildes is now more sophisticated in moderation than its contemporaries. I wonder where the natural evolution of these systems is going to take us ten years from now.
Yeah only having one per day may lead to the too awesome to use dilemma. Perhaps having a stackable 2 or 3 of them that reset every day might alleviate that problem?
What may happen is that people get to save up a certain number. Something low, probably like 3 total, and with that bumping up slowly as the user rises through tiers of trust. Maybe the most trusted users could save up to seven or something like that.
That would flip the 'too awesome to use' on its head, since if you're already at your cap, you're going to 'lose' the next one anyway.
I sort of like the idea of one per half day or two per day. I understand we want to limit these, but looking though some comments since you've made this topic post, I feel like I'm hoarding mine. A "back up" would be nice.
English is weird.
No particular reason other than just a little convenience with having all of them a single word with no spaces/punctuation/etc. It could probably be changed, I think having the hyphen would be the most correct.
I just went to add my first "Exemplary" label to a comment - and stopped cold when I was asked for a reason. I felt like I had to write a bit more than a lame "it was good", but I didn't really feel like writing a critique to justify my motive for wanting to give someone a virtual pat on the back.
So I cancelled the action.
... and then I came here to explain why I probably won't use this feature.
Since it's just a message to the commenter maybe it shouldn't prompt for a "reason" but a "nice message" ?
Currently it's:
Maybe:
I'd think it was totally valid to say: "I really enjoyed reading this" or "This is a point I hadn't considered and I think it adds to the discussion".
I've just realised that I don't actually have to put anything meaningful in that box: I could just put an "x" in it to make the function work.
So? They got an 'Exemplary' label on their comment. Isn't that good enough?
I could write something meaningless like "this is a very good comment", but that's already implied by the fact that I clicked on 'Exemplary'.
Or maybe I just type "exemplary comment" in that box.
Or I could not use the feature at all, to save myself the hassle and stress.
I guess it's "good enough". But if you think a user has written something exceptional enough to warrant one of your limited supply of exemplary labels, taking a few seconds to say something meaningless like "that last point about aliens was really interesting!" shouldn't be overly difficult.
It's for the commenter's benefit, it's nice to come back to a message like "your comment was great!", however implied or meaningless, rather than a literal "x" or "savgfrrhe" or whatever else has been spammed into the box because someone decided it wasn't worth their valuable time explaining something implied. It lets the commenter know they/their comment's been appreciated. Which I get is still implied, but we aren't robots damnit some human recognition for a comment you've made is nice.
Or maybe have the message accompanying the label be optional but encouraged?
Why should I have to analyse the comment I'm labelling as 'Exemplary'? Can't I just label it as such based on a gut feeling that it deserves more than a mere vote? Do I really have to justify my decision to give someone else that extra recognition?
I've worked out my solution: I'll just type something inane and generic like "exemplary" in the message box. I don't have to provide an individual customised personalised hand-crafted review for every exemplary comment if I don't want to.
People complain that voting is too easy, so it becomes an almost meaningless action that favors comments that attract a lot of quick votes. So I add something that effectively gives you a couple super-votes a day, and all you have to do is put in a tiny bit more effort and write a sentence when you're doing it. But that's too much work, and you're already looking for a workaround to make it as easy as clicking Vote again. Even though the comments you just wrote complaining about it are more than you'd have to write for a week's worth of Exemplary reasons. There's really no winning.
It's just a message for the comment's author for you to say why you picked their comment to use one of your limited Exemplary labels. It doesn't have to be deep, just say "thanks" or tell them why you enjoyed reading it enough that you chose to use the label on that comment.
It's not about being too much work. It's about not knowing what to write: I got writer's block! I was faced with a blank text box and had no idea what to put in it. I just wanted to apply an 'Exemplary' label to a comment, not compose a personal message to its author. If I wanted to communicate with the author, I would have written my own comment to reply to theirs. But that's not what I wanted to do. I merely wanted to push the comment further up the page, and give it that special status that comes with being exemplary. But then suddenly I was being required to write a message to the comment author and I found that I had absolutely nothing to say beyond mere inanities.
Like I said, I found a solution. After some thought, and reading super_james' reply to my original comment here, I realised that I don't have to come up with some original and insightful message. I can just type something simple, like "exemplary", or apoctr's "great comment", or even your "thanks". I was applying too much pressure too myself. I don't have to compose something personal and special; the message in the box can be inane. I just need something (anything!) to get me over the hurdle of the text box to the functionality I'm trying to enable: applying an 'Exemplary' label to a comment.
I have since gone back and applied that 'Exemplary' label that I originally froze on. It's all good. I worked it out.
Do you think changing what the prompt says to something else would help with that? Does the current wording make it sound like it needs to be something formal?
I don't remember what the wording says. And, seeing as I've used up my 'Exemplary' label for this 8-hour period, I can't open the function to read the wording.
I'm not sure that it would have made any difference, though. It was the mere fact of being unexpectedly required to write something when that wasn't my intention which stopped me cold. I wasn't prepared for writing anything. Sometimes I want to be loquacious, but sometimes I merely want to vote/label something - and this was the latter. But now I'm prepared for it. I'm good. :)
Anyway, I'm the only one who's complained about this, and I've come to terms with it... so why change it?
If it's something that bothers you, I'm sure it'll bother others too. So if there are some simple changes we can make to improve it, it's definitely worth considering.
The current prompt is: "What makes this comment exemplary? (required, visible to the comment's author anonymously)"
I definitely think something more welcoming and less formal could probably help.
"What makes this comment exemplary?" sounds like asking someone to justify their exemplary vote, with an implication that if they cannot then it shouldn't be applied. Whereas something like "Leave a personal message for the comment creator" could perhaps be seen as a bit less judgemental and slightly more encouraging.
Well, now I know why I got writer's block! Sometimes it's hard to pin down what makes a comment exemplary. It might be a gut feel or overall impression which can't be explained. It might be a lot of little things which can't be summarised in a small text box. It might be something small and insignificant, but meaningful to you. It might be anything. But this prompt requires you to explain yourself. You can't just say something is exemplary, you have to justify it. It's like in maths: "It's not enough to get the right answer, you also have to show your working."
What's the point of this message? Is it to ensure that the voter isn't just performing "an almost meaningless action"? Is it to add a hurdle for the voter so they don't do this thoughtlessly? Is it for the voter to communicate with the author? Is it for the author to receive validation? What's the actual purpose of this text box? Why did you put it there? What do you want to achieve from it? That will guide the appropriate wording.
From my perspective, I think I want it to serve two major purposes:
I think @cfabbro is on the right track. It should look informal and welcoming.
Maybe something like "let the comment author know what you think of their comment" or "leave a brief note to the comment author about their comment".
A good analogy here might be to look at how App Store release notes are written.
Apple require them to allow users to understand why they should upgrade, but most developer release notes are now something like "Bug fixes and UI improvements" -- they may have spent weeks working on that release, but it's too much effort for them to spend a couple of seconds thinking about why they did so. Humans are lazy by nature...
Maybe one way to get thoughtful content is to have default text like: "This is by far one of the best contributions I have seen all day!" then everyone will furiously overtype that with something more to their liking.
It's not all laziness—this was a funny story from last week about why being too specific with app store release notes can be a bad idea: http://www.eod.com/blog/2018/09/rejected/
Yes I saw that. It’s a very well written account. I think that’s why I came up with this particular analogy.
You shouldn't need to do any "analysis" beyond reading their comment in the first place. You read a comment and thought a link they provided was really informative/useful? Just tell them that. I guess in some circumstances you might come across a great comment but not be sure why you like it so much, but honestly I can't think of any examples off of the top of my head and sending a simple "great comment :)" goes a long way for the human factor. Sending just "exemplary" also does that, if maybe sounding slightly detached, so it's a good solution.
As for "why should I"? Because you're doing it for them, not yourself. It's entirely for their benefit, not yours, because you really liked their comment and want to show your appreciation.
Again, it's not about you, it's about them. You're not justifying why you like their comment to some court, you're just telling the commenter "hey, I think that comment was neat" because it's a nice thing to do and because you think they deserve some recognition for it beyond a single vote.
Well, sure. I'm just trying to explain why it's nice to. Also I wouldn't say the message "that link was great :)" would reasonably be described as "an individual customised personalised hand-crafted review" regardless of whether it's technically accurate.
Are "Exemplary" labels visible to anyone? Seems like the kind of validation a commenter might like to see on something they wrote. Maybe less valuable to other users though.
It's very visible to the author of the comment, along with an expandable list of the reason(s) that people entered when labeling it (anonymously, doesn't show which users labeled).
An exemplary comment will also have a
greenblue stripe down the side that all users can see.Maybe another small idea: swap the colours of
ExemplaryandOfftopicSeconded, that pea-green doesn't look nearly as inviting or positive as the bright blue that matches other UI elements on the site does.
Sweet! Looking forward to seeing the effects of this kind of formal recognition of content quality.
Just tested and the green stripe for exemplary doesn't override the black OP bar. Is there any way they can both be shown?
I think it should override now, it was just that it wasn't "activating" unless at least 2 people labeled the same comment before.
Ah ok. Yeah it's working now. Although linking directly to the comment overrides the exemplary bar as well, but that one actually makes sense, I suppose.
Might this be used for anonymous trolling? I know you can intervene, but it's best to think in terms of scale.
Sure, it could. If people want to lose their ability to use labels or even their entire account just to send a short anonymous message to someone, that seems like a pretty good honeypot to me.
Almost everything can be abused, but we can't keep hobbling features because someone might abuse them. If it looks like it's going to become a real problem, we'll figure it out.
I guess eventually every private user->user message will require the ability for the receiver to notify someone that it's malicious & cause the message to be reviewed? Seems like a can of worms with the privacy implications. (I guess tildes can't guarantee that the Exemplary messages will maintain anonymity?)
Say I really want to find out who sent me this exemplary message, can I hit the report button and my buddy the mod says 'aha it came from such and such'.
I don't quite understand. People will need to PM Deimos to report the abuse so he can investigate by looking in the database, which falls well within the scope of the privacy policy. See "What information we collect - functionality-related"
And "how we use and share your information".
No special feature is required, so how does that have any privacy implications beyond the scope already covered in the privacy policy?
edit: Ah, didn't see your edit before replying. AFAIK Deimos hasn't said anything about trusted users/mods being able to see the exemplary messages and I doubt that will be the case. I imagine this sort of abuse investigation will be handled exclusively by admins for exactly the reasons you just pointed out.
Yeah, I think it's pretty unlikely that "regular" users will ever get access to any information like that because of the privacy implications. There's a lot of private data that shouldn't be accessible to non-admins. It's possible that someday the site will be so active that it's not feasible for me alone to review reports involving private data, but that's probably still a very long time away.
Makes sense. Thanks for the explanation.
The admins will still know who made the comment - it's just not shown to the author. If something nasty shows up in there, the user can pm about it and that'll be the end of that person's ability to use that feature at the least.
I think what you're describing would fall under the umbrella of brigading, since users only get one per day. It would have to be a coordinated mass trolling effort (or sock puppets). Personally I think the downsides of showing attribution for labels outweigh its benefits. True anti-brigading measures should be framed as such, and will need more potency than just displaying griefers' usernames to their victims in order to be effective.
Two points.
Could I suggest that "Exemplary" and "Malice" are changed to "Exemplary..." and "Malice..." to indicate that they require a further step before they will take effect?
Given that those two labels do bring up a dialog box, that would be an ideal place to note the specific restrictions on their use, rather trying to squeeze it into the labelling area.
(edited to add that I think this is a nice simple solution, looking forward to see how it works.)
Thanks, I'll try it out and see how it looks/works. Are you thinking that the time restriction on Exemplary should be inside the actual dialog box as well, instead of below it like it is now?
Yes, that would provide in one place all the information to allow the user to decide whether to proceed, as well as solving the issue of how to communicate that the limit only applies to those two labels.
Once the trust system is put in place, would it be beneficial to have more trusted users be able to use exemplary more often?
Probably. There's definitely things we can do with making Exemplary labels available more often and/or making them stronger for particular users (since there's an internal concept of "label weight" for each user).
Personally, I think everyone's
exemplarytag should weigh the same. Even though the more trusted users are going to know more about the site in general, it just seems like the main barrier to getting tagged as exemplary should be the number of users who agree.I do think that trust should definitely place weight on the negative tags, for sure, since these are more for moderation rather than making comments stand out.
Definitely like the idea of more frequent tags for more trusted users.
Ooh, at first I wasn’t sure about that but the idea of cooldown length being a function of user trust is growing on me. That makes a lot of sense.
Or cool down could be related to reputation, with higher-level users having either a shorter cool down, or more labels to use than a newer user. Would make sense for people who are here very often and contribute a lot.
Hmm, I think I might have inadvertently made it so that currently it will only show up if two users label the same comment as Exemplary. I'll fix that, I don't think that's how we'll want it to work since that'll probably be a somewhat rare event. Thanks for pointing it out.
Edit: should be fixed now.
Great changes all around! I like the 8 hour limitation for Exemplary, it's a smart and simple way of keeping them special.
I also generally like that there's now a positive tag/label. Just like forcing people to give a reason for "downvoting" stuff, this is forcing people to give a reason for "upvoting". As I said earlier, I can see labels replace voting altogether, it never hurts to make people think twice about voting (for example a "helpful" label or something similarly named that basically is "vote").
Good ideas here. I really like the fact that "exemplary" is limited in how often it can be used. That should make people think carefully before using it. Later on maybe the replenishment rate could be tied to level of trust.
Does the "malice" tag trigger an alert for you or do you have to directly view it to know about it?
I have a simple little tool that lets me know about them right now, and I'll build something more formal for it eventually.
I would still suggest slightly more label visibility.
Labellink, the current tag counts are displayed likeNoise (2)rather than justNoise, though probably not the reasons behindMaliceandExemplary(maybe in the future users with mod or semi-mod rights)Noise, then I might seeLabel (Noise)rather than justLabelwhen the labels pane is hiddenI still disagree that labels (other than Exemplary) need to be visible in general, since that will encourage a lot of negative behavior like pile-ons, mocking ("lol this comment has 10 Noise labels"), and so on. So far, I still think that making people decide independently when to label something is best, and just having the labels cause particular effects when they're applied "enough". It may still change though, a lot of it is based around seeing how people are using them in practice.
It should definitely be more obvious when you've already labeled a comment though, with some sort of indicator without having to re-open the actual label button menu.
Re: indicators, if I may make a sort of cutesy suggestion:
Label has five characters. There are five labels. Each label has a distinct color.
If one has previously labelled a comment, perhaps the corresponding letter in Label could change to the corresponding color of the corresponding label.
Of course, some folk would very much dislike that. And it breaks if there are ever more-or-less labels. Or the word Label gets changed. Or maybe that would look janky/better/worse on some themes. Or the coding could be overly complex. Or maybe that isn't great for the colorblind.
Thank you! :)
How much are votes multiplied by when a comment is labelled 'Exemplary'? Or are these up-multipliers for positive labels and down-multipliers for negative labels intended to be secret Tildes magic?
Nothing's very secret with it all being open-source. It might not always be this clear since it's possible to change individual users' "label weight" (and someday this should adjust automatically), but right now all users have a label weight of 0.5. The multiplier is
1.0 + total exemplary weight. So if one user labels a comment Exemplary, it basically has a 50% bonus to its votes for the purposes of the "relevance" sort. If two people label the same one, its votes are effectively doubled, and so on.I keep forgetting about that, considering that I don't know how to find or read the code. Anything in there is Secret High Magic to me. :)
Thanks for that.
I like it, the 24 hour limit with a report for exemplary seems like it'll hit the perfect sweet spot that promotes good comments but doesn't lead to overuse of a "super upvote" that might lead to hive mind like behavior.
All the options don't actually fit on my phone screen anymore—a little disappointing because it was perfectly centred before.
Hmm, that's a bit weird. It's the same number of options as before, so just having one be slightly longer was enough to push it off the screen?
Yeah I'm on mobile here and on parent comments I can see everything but the word malice is cut off at the 'e'. When I try and label a children comment the whole malice label potentially goes off screen depending on how far down the comment chain it is.
This is the old unsolved too-long-comment-chain problem. If you have long enough comment chain, even the normal buttons (vote, label, reply) overflow from the screen. I don't think the problem could be solved without something like the reddit "continue this thread" button.
Maybe there could be horizontal scroll bar, but that's more avoiding the problem rather than solving it.
The word Malice is entirely cut off, and most of Noise is, on my iPhone SE.
Bizarre; did the margins change or something?
I made a few minor margin/padding changes as part of adding the little note underneath Exemplary, but I didn't think they'd affect things much. I'm planning to tweak it a little more tomorrow and try to make it wrap nicely if there isn't enough horizontal space, so hopefully it should be fixed fairly soon.
That's great; thank you!
Nice.
I like especially the "exemplary" tag.
I'm thinking about how rate limiting will play out. I think it's useful, but specific mechanics might be changed, and possibly applied to other elements.
"Untrue' or "nonfactual" might be another classification. Though this could overlap with "noise".
The notion of a depleting, renewable stock might be ueful. A set of queues, each good for a specific period (or conversely, a set of rate limits). Applied to submissions, comments, votes, and lables. Say, n/minute, /hr., /dy., /wk, ...
Another is that actions are divided by utilisation per day. One upvote has a stronger weight than 10, 100, 1,000, etc. (I'm not sure how to weight these, though some nonlinear decay might be preferred.)
A feature for users to review both their issued andd received labels would be helpful.
Are there any flames for a link labelling system? As of right now, if someone posts something that falls into the "that is absolutely not ok here" category, the best course of action is to send you a DM. I think, at least in the site's smaller stages, labelling could be an excellent system in lieu of a report system. If a text post is being intentionally malicious, we could use the malice label, yet if it's incredibly insightful (as you'd see in a group like ~life), it could also be a recipient of the exemplary tag. Low effort content could get a tag like noise.
This would definitely need some tweaks, potentially with a
violates site ruleslabel, but I think it has the potential to be incredibly useful.It's a tag called malice, yeahhh