YouTube Moderation Bots Will Start Issuing Warnings, 24-Hour Bans (arstechnica.com) 59
An anonymous reader quotes a report from Ars Technica: YouTube has announced a plan to crack down on spam and abusive content in comments and livestream chats. Of course, YouTube will be doing this with bots, which will now have the power to issue timeouts to users and instantly remove comments that are deemed abusive. YouTube's post says, "We've been working on improving our automated detection systems and machine learning models to identify and remove spam. In fact, we've removed over 1.1 billion spammy comments in the first six months of 2022." It later adds, "We've improved our spambot detection to keep bots out of live chats."
When YouTube removes a message, the company says it will warn the poster that the message has been removed. The company adds, "If a user continues to leave multiple abusive comments, they may receive a timeout and be temporarily unable to comment for up to 24 hours." [...] It does not appear that YouTube is involving channel owners in any of these moderation decisions. Note that the post says YouTube will warn the poster (not the channel owner) of automated content removal and that if users disagree with the automated comment removal, they can "submit feedback" to YouTube. The "submit feedback" link on many Google products is a black hole suggestion box and not any kind of comment moderation queue, so it sounds like there will be no one that responds to a moderation dispute. YouTube says this automatic content moderation will only delete comments that violate the community guidelines—a list of pretty basic content bans—so hopefully it will stick to that.
When YouTube removes a message, the company says it will warn the poster that the message has been removed. The company adds, "If a user continues to leave multiple abusive comments, they may receive a timeout and be temporarily unable to comment for up to 24 hours." [...] It does not appear that YouTube is involving channel owners in any of these moderation decisions. Note that the post says YouTube will warn the poster (not the channel owner) of automated content removal and that if users disagree with the automated comment removal, they can "submit feedback" to YouTube. The "submit feedback" link on many Google products is a black hole suggestion box and not any kind of comment moderation queue, so it sounds like there will be no one that responds to a moderation dispute. YouTube says this automatic content moderation will only delete comments that violate the community guidelines—a list of pretty basic content bans—so hopefully it will stick to that.
Is this another case of ... (Score:3)
... you are being timed out. We won't tell you what you did "wrong" until after the fact?
Re:Is this another case of ... (Score:5, Informative)
Re: (Score:3)
> It's mostly spam
Perhaps mostly but of the billion comments that got deleted, probably 50 million were real people who took real time to write a thoughtful comment.
Complaints are everywhere about YouTube insta-deleting comments. I've seen it; no obvious spam triggers.
Re: (Score:2)
Back of the napkin, I figured out that if 1% of the blocks were conversation-worthy, it means 61 conversations per second.
It's not the justice system. "Better that 50 million good posts are culled than that the message boards fade to uselessness under spam and abuse."
Re: (Score:2)
I've witnessed this first hand.
Person replies to one of my comments. I see their comment show up in my email but if I go to YouTube it shows the proper total number of replies but expanding the thread shows half of the comments are being censored?!
The really strange thing is that it was a perfectly normal comment! I asked the person to split their 3 sentence reply into 3 separate posts. One of them was censored. Both of us couldn't tell why. It didn't use any "banned" words. Just normal conversation.
Re: (Score:2)
I don't know. I have been ghost blocked twice for 3 months each time, either for noticing a bunch of nazis being a bit nazi in the comment sections, and commenting on it, or alternatively for marking a lot of spam as spam. Those are the only two things in common it did prior to both bans. So I have stopped arguing with nazis and marking spam as spam, and haven't been blocked since..
Re: (Score:2)
YouTube will be doing this with bots,
Please remove this content. You have twenty-four hours to comply.
You now have fifteen hours to comply.
You are in direct violation of Youtube Code 1.13, Section 9. You now have five hours to comply
Four... three... two... one... I am now authorized to use a permanent ban.
Re: (Score:2)
... you are being timed out. We won't tell you what you did "wrong" until after the fact?
YouTube tells people what they did wrong?
Re: (Score:2)
Frequently NOT because they don't want you "technically" complying with the rules but still saying something they disagree with.
Re: (Score:2)
... you are being timed out. We won't tell you what you did "wrong" until after the fact?
"You know what you did. Now go to your room. And NO YouTube!"
Some channels are abusive, how do I work this? (Score:1)
Here's a good example:
https://youtu.be/wfKSOrofg8U?t... [youtu.be]
This guy live streams on Omegle using a Kermit the Frog Puppet, finds of age girls, shows them a zucchini (made to represent Kermit's penis) Now I've been on the live chat for this, and as you can expect it gets pretty raunchy. All kinds of comments about the "Victim" (who never sees the youtube chatter) go flying around. I don't think the bot can accurately define the context of the chat. It's directed towards someone not even aware or has visibili
Re: (Score:1)
Economies of scale (Score:3)
Like most tech companies, YouTube wants to maximize the number of customers than can be retained per each company employee. Using rough estimates from public data sources, we can guess that YouTube has about 3,000 employees and 3,000,000,000 users. That means each employee, on average, has to add value to the company equivalent to supporting 1,000,000 users (with the usual caveats about averages implied here.)
Because only automation can be used to support this magnitude of customers, customers typically suffer by having no recourse to appeal to the company when they are penalized in a way that merits per-case attention. To even exacerbate this problem further, because the product is "free", and thus the barriers to entry for using the product are extremely low, malcontents with bad intentions can outnumber legitimate users by 1,000 to 1, or even more. Thus, for every legitimate customer that would like special attention from YouTube, there are hundreds to thousands of bots that make it difficult for YouTube to even differentiate legitimate requests from illegitimate ones.
This is why I will never rely upon a big tech company that primarily provides free services for the majority of its revenue. There's no way such a company can support its customers ethically while simultaneously protecting them from all the malcontents attracted by free services. If you lose access to your Google Email address, Google will politely tell you, "So sorry, just create a new email address." Conversely, if you lose your ability to pay for products on Amazon, Amazon will go to great lengths to help you become a profitable customer again.
Re: (Score:1)
Re: (Score:3)
Some of it's creditors are very unsavoury characters who will not react well to being told they might not get repaid.
I'm enjoying it all very much.
Re:Economies of scale (Score:5, Insightful)
Amazon will go to great lengths to help you become a profitable customer again
Maybe at some point in history... but once they have destroyed all competition they will just tell you to go pound sand like every other monopoly. What are you going to do? Shop elsewhere? Remember when amazon started banning residential addresses for violations of their ambiguous free-returns policy a few years back? AFAIK they have since stopped this practice, but the same exact question came up then. What do you do if amazon bans you and then refuses to accept an appeal, review the case, or even let a human to talk to you. You can go back to brick-and-mortar and other online retailers, but what happens when those no longer exist or are otherwise inferior options due to amazon steamrolling the entire market with frightening inevitability?
The problem here are monopolies more than anything else. Youtube IS the place for online video. Especially medium-length, episodic, non-news video (i.e. not tiktoks, not facebook updates for your fam, not tweets about current-events, etc.). If the platform were a federated set of smaller players, then there would still be innovation and competition for your individual business (e.g. providers catering to higher-earning creators, providers serving niche categories, etc.) Hell, if they had the option to pay for human support, that would already solve a lot of problems (i.e. the bots can still play bot wars with the automoderation while humans are able to purchase another humans time and judgement). But there is ZERO incentive to do anything whatsoever right now, because youtube at this scale simply doesn't have to compete with anyone else for eyeballs in this particular media space. Making the product any better is a waste of money once you have climbed the mountain over the bones of everyone else.
Re: (Score:2)
Re: (Score:2)
Users of free services are not customers. They are the product. Paying advertisers are the customers.
Bots on bots (Score:2)
in a hot tub.
Slashdot Telegram (Score:2)
DM on telegram, you won $1000000 prize!
Yeah, enough of that schlock. Hopefully it's that kind of overprevalent junk comment that they are targeting with this.
when are they going to bring this to blogger.com (Score:1)
I'm tire of having to see ads for working from home from myscamwebsite.com
I don't have much hope in bots (Score:2)
YT has been flagging all of my comments as spam for several years now, and removing them within minutes. Needless to say they are not spam. And there is no way to appeal to that. So basically I'm not posting anymore on that site because it's hopeless.
Re: (Score:2)
They don't actually remove them—they never post them.
If your comment is detected as spam, it sends back all the acknowledgements that you sent it, it even shows up in the comment thread. But, if you refresh the page, it will not be there. This is even worse in the API, since it just gives you a 200 reply no matter what.
Where this will really get painful is on channels from non-US cultures. Channels like Garn (an Aussie comedian who does skits about Australian life) are filled with swearing Aussies. Ho
Re: (Score:1)
as far as I can tell this is not triggered by swearing
it seems to do a fairly good job of finding things that "look" like they are rude
but thats a far cry from detecting actual rudeness
its going to drive rudeness to the place between the lines
Re: (Score:2)
An improvement over today's secret deletions (Score:2)
I started to notice late last year that I am heavily censored on YouTube. YouTube tries to hide what it is doing using two methods, and if they only used the second method, almost no one would realize it is happening.
Often I edit my comments because I'm rarely quite happy with how I worded it the first time around. This started to fail. It would say that an error occurred. Then if I refreshed the page, my comment would be gone. It would be deleted from my comment history, too. After this happened twice,
Re: (Score:3)
I started to notice late last year that I am heavily censored on YouTube. YouTube tries to hide what it is doing using two methods, and if they only used the second method, almost no one would realize it is happening.
With Elon Musk releasing more and more data from Twitter we are finding out exactly how much collusion is going on with social media. I expect soon we will see laws coming out that will outlaw such activities as this. Maybe even a end to social media immunity to lawsuits.
Re: (Score:2)
I expect soon we will see laws coming out that will outlaw such activities as this.
You expect the first amendment to be repealed soon? I'm not sure they have the votes for that, but then we may have an actual dictator in a couple years, so it may not matter.
Re: (Score:2)
Oh please, this has nothing to do with the first amendment. This has to do with companies, like twitter for example, breaking their own terms of service to silence people they don't like. Take Trumps ban for example. The "twitter files" clearly show that nothing he did broke the terms of service. They batted his account back and forth till Vijaya Gadde decided at the behest of Michael Obama to ban the orange one.
Re: (Score:2)
Re: (Score:2)
I don't think you understand, but then I imagine that's a common occurrence for you.
Oh, I understand perfectly. You see, unlike you, I'm not a god damn moron. You are preaching first amendment; I'm discussing them violating their own Term of Service. You got that? I'm pretty sure you don't even though I typed this real slow for you. You need any help with the big words let me know.
Other than that, pay attention or shut the fuck up. I would advise you to do both.
Oh, but by the way. I don't' think corporation should have first amendment rights.
Re: (Score:2)
I replied to the portion of your comment that said "I expect soon we will see laws coming out that will outlaw such activities as this."
You will see those words in my original comment. There is a little gray bar to the left of them - that indicates that they are a quote of yo
Re: (Score:2)
moronic statements ignore and not read.
I'm sorry, I don't have the time or inclination to read anything posted by moron. This conversation is over. time to get your meds checked. Take care.
Re: (Score:2)
Re: (Score:2)
It is not done by staff, it is done by bots. So if three people mark your comment, it gets deleted, if they do that to a number of your comments, you get a three month shadow ban. It is used by certain groups to silence people calling them out.
Re: (Score:2)
Re: (Score:2)
I found I just had to stop figthing for basic human rights for everybody, and the nazis would stop getting me shadowbanned..
Re: (Score:1)
I'm guessing their anti-spam bot is really not sophisticated at all. It probably simply has a list of keywords that they think might appear in a spam message and it just deletes all comments that have those keywords in them. Maybe there are other rules too, like message has a link in it = spam, message is too long = spam. Maybe something even dumber than that.
Underrated (Score:4, Funny)
If they ban every commenter who uses refers to someone as "Underrated" or a "National Treasure" that would help a lot.
Re: (Score:2)
+1 Funny.
Why doesn't YouTube have "Funny" and other appellations like Facebook and Slashdot?
compromised user accounts (Score:1)
fighting comments, but not doing anything about compromised user accounts filled with political spam.
Priorities set straight.
Don't be rude about the Euro (Score:2)
I got censored for calling the Euro the currency of the 4th Reich.
When Will Youtube Stop Dangerous/Deceptive Ads? (Score:2)
Judging from how spam works now on YT... (Score:3)
...I have a hunch that this will backfire badly. Really badly.
The current "spin" of spam on YT is that one "person" posts something, followed in rapid succession with others chiming in and praising whatever the first one peddled. Currently it's mostly craptocurrency scams where such-and-such is the big secret to success and you should look them up on (insert messenger here).
The thing is that it's not just a single account doing this, there's often a dozen or more different accounts, often some years old, that offer this "advice".
Now let's ponder for a moment how these moderation bots will work. Well, they'll likely follow complaints by users, maybe combined with a few keywords.
I foresee that these spambots will get a new revenue stream. Don't like someone? Can't out-smart them in the comments? Let us squelch him by reporting him with a few dozen accounts.
It's the Reddit game all over again.
Anything to avoid the real issues (Score:1)
I know it's tough to identify them when they were sped up by 2% or someone added a border around the video, but hey, at least you took down that guy who was critical of a new NFT scam because the creator says using his name in a video is a DMCA violation.
Re: (Score:2)
"Troll 'em all! Let Dog sort 'em out!"
just another way (Score:2)