Hacker News Reader

Comment by knaik94 | Hacker News Reader

The update is more crazy than the initial flagging. Youtube went back and doubled down saying the video violates their "sex and nudity policy". This is a channel that uploads how a speedrunning record was set https://twitter.com/summoningsalt/status/1575972292395667456 . The creator only does voiceover explanations, all of the content is gameplay footage, and occasionally a picture in picture video of a speedrunner while they are achieving a specific record. The video is age-restricted again.

There comes a point where Youtube is not saving money by automating their content moderation but Youtube decides to keep pushing forward. At some point a competitor that is good enough will arrive and once large creators like this one, around 1.5 million followers, almost every video with over 2 million views, 8 videos with over 5 million views, leave they will pay attention. Until content creators have an alternative choice, Youtube knows they can implement any black box policy and ignore 99% of the backlash.

My personal theory is that they are doing manual review but only after an appeal. They definitely have software that flags videos for vulgar language internally, and I imagine only those parts and a couple minutes surrounding it are reviewed. They refuse to elaborate or make their methods public because they see it as a way to combat spam. It's becoming clear that the moderation team is mostly software at this point.

I have a different theory that Youtube is also classifying channels into kid friendly and unfriendly based on video content while ignoring the context the channel was made for. Game are generally much more popular among kids, and so this channel was marked as kid friendly. The channel is getting treated with rules that should only apply to kid targeted channels, unfortunately this benefits Youtube way more than the creator. Youtube Kids has ads.

View on HN

> There comes a point where Youtube is not saving money by automating their content moderation but Youtube decides to keep pushing forward.

When you deal with the scale that YT are operating at, automating content moderation is the only cost effective way.

> At some point a competitor that is good enough will arrive

Competitors have always existed. They existed before YT got big, still exist, and new competitors have been created since. The problem isn’t lack of competition, it’s the network effect. People go on YT, few specifically load YT to go on specific channels. Or to put it in an old world context, people don’t buy a TV for a specific TV show. They buy one to watch lots of TV shows knowing there are multiple shows they could enjoy.

So content creators leaving YT doesn’t equate to all their subscribers leaving YT. If anything, what that actually equates to is that content creator loosing viewers. And bare in mind that most content creators will have significantly more viewers (regular + irregular) than the number who’ve subscribed.

> When you deal with the scale that YT are operating at, automating content moderation is the only cost effective way.

Why?

If youtube had 30 million in revenue, could they afford a couple moderators?

Their revenue is 1000x higher, so can they not afford 1000x as many moderators?

Let's take some ballpark estimates. Half a million videos are uploaded per day (probably accurate), and 10% of them reach a thousand views or are on channels that typically reach a thousand views (probably accurate), and 10% of those need a human moderator (probably way too high). Investigation takes 10 minutes with each moderator doing 40 per work day (just a guess, should be sufficient), and averaging 25 per day over the entire year.

Multiply all those numbers together, and you get an estimate of 200 moderators to handle all videos of any import. For a company with 30 billion dollars in revenue.

Am I super far off on one of those numbers? The fact that it's so hard for medium to large channels to get the attention of a human seems entirely like mismanagement to me, not a real cost issue.

It was 30,000 hours of new content an hour on average a few months ago. That's 720,000 hours a day. The average video is reportedly 11.7 minutes, which makes it about 4M new videos per day. You're off by about an order of magnitude.

Edit: 10% of videos with more than a thousand views is reportedly acurate. That makes it 50,000 videos per day according to your numbers, and with 25 videos reviewed a day on average by person, that makes it 2,000 persons with your numbers, not 200, so you're off by another order or magnitude.

For 10% of 4M, that would be 16,000 people, or, what, 10% of Alphabet's workforce? Youtube itself reportedly has 2,800 employees.

Huh, I didn't think a 2016 number would be that wildly different.

Okay, 4 million videos. So if we adjust the moderation percent to 5% that would mean 800 moderators.

And focusing on videos with 10k views would need 1/4 of that. Videos with 100k views would be 1/10 of that. And this is a channel that gets millions of views.

There's only going to be a couple thousand videos per day at the million+ tier. A handful of people could provide white glove moderation to all of them.

Edit to your edit: I think you skipped the part where most videos don't need to be reviewed. Take that 16k and divide it by 10 or 20 or 50.

The problem we have is that popular content creators are the most susceptible to the problems caused by YTs moderation. Your solution only reinforces that problem since you’re excluding unpopular content creators.

You’re also making the assumption that humans are going to moderate consistently but that simply isn’t true. Different people will draw the line at different points and our mood will also vary the line too. Review guidelines would only only help with measurable criteria but so often what constitutes as acceptable or not is a lot more nuanced.

> Your solution only reinforces that problem since you’re excluding unpopular content creators.

If you're not even getting a thousand views then it's not going to be a big deal if some of your videos get messed with. Annoying but not important to your life. Just about anyone that would seriously call themselves a content creator, even as a hobby, should be above that bar.

I'm in favor of good moderation for everyone, sure, but I want to hammer home just how easily they could provide good moderation to any video with traction.

> If you're not even getting a thousand views then it's not going to be a big deal if some of your videos get messed with.

I don’t think uploaders would agree with that sentiment.

For starters not every video uploaded is by online personalities. A lot of corporate entities use YT. I’ve been on mental health courses that had material on YT too. These videos aren’t going to receive thousands of videos but the content is just as critical for their operations (arguments about whether they should use YT is another topic).

And you’re still overlooking the injustice that the larger personalities would feel if they have content demonetised or even banned while smaller channels have the same content available.

It’s a nice idea you have but it’s creating an uneven playing field so would have just as many problems as the status quo.

Edit: also let’s not forget that your suggestion is easy to weaponise. If you’re a smaller channel competing with a larger one then you can abuse the moderation system by reporting the other channel. Thus any moderation system has the be universal or else it’s going to be abused.

> I don’t think uploaders would agree with that sentiment.

Which uploaders? I spent a while looking through my subscriptions on youtube, and nothing I'd call a creator is below that line, not even this obscure meme channel that got banned once. Just an acquaintance's personal channel and the archives of risc-v presentations, and neither of those channels would be bothered very much by a rogue bot flagging.

> For starters not every video uploaded is by online personalities. A lot of corporate entities use YT. I’ve been on mental health courses that had material on YT too. These videos aren’t going to receive thousands of videos but the content is just as critical for their operations (arguments about whether they should use YT is another topic).

Well those aren't going to be monetized or affected by this kind of thing. And if youtube goes weird they can use any other video host; they don't need youtube or its userbase.

> It’s a nice idea you have but it’s creating an uneven playing field so would have just as many problems as the status quo.

Add those few hundred moderators in addition to whatever the hell exists already and it should improve a lot while having roughly zero downside.

> And you’re still overlooking the injustice that the larger personalities would feel if they have content demonetised or even banned while smaller channels have the same content available.

I don't know what you mean or why that would happen.

> also let’s not forget that your suggestion is easy to weaponise. If you’re a smaller channel competing with a larger one then you can abuse the moderation system by reporting the other channel. Thus any moderation system has the be universal or else it’s going to be abused.

How is that a problem? The moderator will look at it and reject it, and the small channel might get punished if it keeps doing that.

And small channels can already report big channels. I really don't understand this scenario.

> Which uploaders?

I gave examples.

> Well those aren't going to be monetized or affected by this kind of thing.

I’m pretty sure they’ll be affect by their channel getting suspended ;)

> And if youtube goes weird they can use any other video host; they don't need youtube or its userbase.

It’s not as straightforward as you suggest when busy orgs with no dedicated IT resource has to re-upload thousands of hours of training material to a new platform they likely haven’t heard of before.

You’re point of view here is pretty narrow compared to how the wider usage of YT.

> it should improve a lot while having roughly zero downside

In my experience the people who claim “zero downsides” are the last people who should be making business decisions. Every design has a trade off and all too often there are unforeseen consequences too. If you honestly believe there are zero downsides then you haven’t spent long enough in the industry at the level of seniority required to understand the the consequences.

Sorry if this sounds condescending but I’ve worked with so many engineers who have believed they shit gold only for them to discover they knew far less about the job then they believed they did.

> I gave examples

I think they meant specific examples (as opposed to hypothetical or vague examples)

> I’m pretty sure they’ll be affect by their channel getting suspended ;)

Unless I'm misunderstanding the other guy incorrectly, what they're proposing is additive.

If I understand right, you're essentially saying "But people without manual moderation might be unfairly impacted by automation".

But it's a bigger problem without the custom moderation the other guy is suggesting?

> In my experience the people who claim “zero downsides” are the last people who should be making business decisions.

Fortunately, it sounds like the other guy isn't making any business decisions in this specific context. So why not just stick to addressing the (de)merits of their suggestion, and leave the veiled insults out?

> Every design has a trade off and all too often there are unforeseen consequences too.

Stating a self-apparent truism doesn't bring much value either. Even if someone here wasn't aware of this idea, stating it as an abstract doesn't really help the discussion much, and seems like a bad faith comment to make.

> If you honestly believe there are zero downsides then you haven’t spent long enough in the industry at the level of seniority required to understand the the consequences.

Attacking the guys seniority doesn't add any value and is a fallacy. If you seriously believe there are downsides, why not state them instead of insulting the dude?

> Sorry if this sounds condescending but I’ve worked with so many engineers who have believed they shit gold only for them to discover they knew far less about the job then they believed they did.

My dude, this doesn't just sound condescending, it is condescending. This trailing comment borders on outright gaslighting.

The fact you're aware it is condescending but decide to say it anyway means you're not sorry either - otherwise you wouldn't have said it in the first place?

I'll spare you a self-righteous diatribe about bitter, unempathetic co-workers, since I'm sure you'll agree it's not relevant.

I’m management now and definitely an empathetic one. But there’s plenty of studies about the assumptions people make about their own capabilities

https://en.m.wikipedia.org/wiki/Dunning–Kruger_effect

So the moment someone confidently posts an absolute like “zero downsides”, I immediately see red flags that perhaps that person hasn’t fully explored that solution.

This is compounded by the fact that there are no shortage of cognitive biases (such as, but not limited to, the Ikea Effect) that can lead one to overestimate the value of their own ideas.

So when I see absolutes, that’s always a queue for me to engage in a deeper discussion with the individual to ensure their confidence is justified.

> So why not just stick to addressing the (de)merits of their suggestion,

I had up until then. They couldn’t look past their idea and replied that there was zero downsides to it. Hence the concern I raised.

> Attacking the guys seniority doesn't add any value and is a fallacy

Fair point. It wasn’t a tactful response. Thank you for pointing that out :)

> I think they meant specific examples (as opposed to hypothetical or vague examples)

I gave a specific example and the other suggestions were anything but vague.

You dont even need to look far to see my point, just look at the videos submitted to HN and you’ll see that YT is more than just internet personalities adding content for monetisation:

https://news.ycombinator.com/from?site=youtube.com

You gave an example of something that wouldn't change and called it a downside. So yes you need to do better than that for examples.

If you're taking a "no free lunch" approach, the downside is that it costs money to hire these people. Does that make the balance clearer?

Adding human review to more decisions, and changing nothing else, can possibly cause some problems but they should be very minor compared to the number of robot-caused problems they will fix.

> I gave examples.

If those are your only examples, I addressed them the line after that. Please take my comment as a whole.

> Every design has a trade off and all too often there are unforeseen consequences too. If you honestly believe there are zero downsides

It's always possible something somewhere could go wrong but "don't change anything on smaller videos" is pretty hard to screw up! If that's too risky, then your standards are too high and you interpreted "roughly zero" much too precisely.

If you’re waiting until a few hundred thousand views before moderating a video then you’re already too late.

And if you think ignoring millions of unpopular videos watched by thousands of people (in many cases, each) isn’t risky then you’ve underestimated the scale of content on YT.

I don’t even think the “roughly” disclaimer will save you at YTs scale.

> If you’re waiting until a few hundred thousand views before moderating a video then you’re already too late.

Oh. I'm not suggesting that at all. I'm saying that bigger videos should get a better system, and smaller videos can use the current system.

It's minimally risky because it doesn't change anything for those videos. They get the current treatment.

I think "No change -> roughly zero downside." is a reasonable conclusion, at least for those videos.

The moderators only really need to review the videos already flagged by the automatic process, and the classification ML thingy can add the timestamps with positions the "objectionable content" is featured at, so the mods wouldn't need to view the video in whole.

How many of the videos are flagged down, though, we probably don't have any solid statistic on that.

You don't need to review ALL new content, just content that gets traction or is manually flagged by the uploader as being wrongly marked by the AI.

YouTube doesn't need more moderators, they need to be compelled through negotiation or law to be fair.

What would that look like? Any creator could send a classification (such as age restriction) which they disagree with to an independent moderation agency, and the party which is decided for or against pays for it. There are of course issues with maintaining the independence of those agencies, but this is a problem which has been solved successfully in the past.

Also (this is table stakes, it's pathetic this isn't already done) an exact timestamp for any part of a video used to classify it must be provided by YouTube.

So it's about reviewing ~5 seconds of video in context, not the whole thing. There may be two, three of those timestamps, but if someone has to look at a full minute of a video to determine if it's age-appropriate? It probably isn't.

Maybe I have a different understanding of how people use YT but everyone in my circle has a YT account because they have a gmail account and therefore subscribe and actively watch things based on that. Because gmail has become the standard every single person also has a youtube account. In new world terms, Youtube is one of the most used apps on my TV in my house and the first question I was asked in terms of tech support by my parent was "How do I connect the TV Youtube to my account? I want my recommendations to show up."

Competitors don't really exist. The closest is most likely Baidu. But even Baidu isn't close enough to be considered competition. There is very clearly no English language alternative. Knowing alternatives isn't the same as people actively using them.

Even content creators that don't leave will be less excited about making content. People will watch mostly similar videos on Youtube by design. Once a large content creator leaves and demonstrates that it's a viable option, other content creators will follow. Once most of the people you enjoyed leave a platfrom, and they leave in a small amount of time, people will stop going to youtube.

Having more viewers than subscribers doesn't matter nearly as much as "the algorithm" and how many people get recommended a video. "The algorithm" makes or breaks a video's momentum, and that's obvious by videos like this content creator's that bluntly put how this kind of mismanagement put their livelihood at risk. People not logged in will not be recommended a flagged video, it will be unprioritized in search results too. I have first hand experience on that. Viewers vs Subscribers is meaningless to argue when even your subscribed users over the age of 18 barely get recommended your new video.

Looking from a world view, users in the EU require users to share with Youtube a valid driver's license, Proof of Age card, or passport in order to get 18+ videos unblocked. Content creators have a hard time convincing viewers to just press a button on a webpage.

Some of these things are not obvious without having analytics numbers to watch when you release a somewhat popular video.

You’re making a lot of illogical counter arguments here.

> Maybe I have a different understanding of how people use YT…

You’re not disagreeing with my point though. Just talking about it from the other end. But the point you’re making is the same as me: people don’t use YT because a specific content creator is on there. They use it because of the network effect.

> Competitors don't really exist.

Of course they do. There’s literally dozens of big videos sharing platforms. What you’re arguing is that there isn’t any platforms equivalent in size, which is true but that doesn’t mean that there are not competitors on the market.

The problem isn’t a lack of competition, it’s the network effect. It’s the fact that most people watch YT, not $CREATOR who happens to also be on YT.

> Once a large content creator leaves and demonstrates that it's a viable option, other content creators will follow.

YT has already had large (millions of subs) creators leave their platform and it’s had little difference. Hence my point that people are not using YT for specific content creators.

> Having more viewers than subscribers doesn't matter nearly as much as "the algorithm" and how many people get recommended a video.

You’re making a straw man argument here because I wasn’t talking about the algorithm. I was talking about users. I was making the point that subscribers aren’t the full picture in terms of the number of viewers a creator might have.

> People not logged in will not be recommended a flagged video, it will be unprioritized in search results too

You’re making the assumption here that non-subscribed viewers aren’t logged in. That’s usually not the case. Lots of people will be logged in by default (because of GMail et al) but still not subscribing to every creator they enjoy.

Example: I have subscribed to literally zero creators and still see everything you’re claiming I wouldn’t.

I see more clearly where you are coming from. I agree, we are looking at mostly the same conclusion from different perspectives.

But "the algorithm" isn't a straw-man argument. I am saying viewers vs subscribers is a false dichotomy to care about. People at some point will stop going to Youtube when there is competition along with their favorite creators they actively watch.

The main problem is competition. There are no viable competitors that feature match youtube in terms of feature richness and mobile experience. Having no effective competitors in size means Youtube doesn't need to do anything to keep content creators or users on its platform. People go to youtube because they know they will find something interesting. If there is another interesting site, they will start going to it as well and just go to both.

The network effect is largely due to groups of creators on a platform. If you look at an example competing platform like Twitch, there are many people that just don't come to Youtube because their favorite group or individual creator does not livestream there. The number of people that watch Vods of previous live streams when it is later reuploaded to youtube show how few people convert between platforms. It's even more interesting what happens when a creator does switch. I have heard anecdotal stories that they don't get the same viewer base back, they mostly foster a new group of fans. It is especially hard to convert when they do an immediate switch due to contracts. The numbers on Twitch and Tiktok show they have a dedicated audience.

Unfortunately those platforms aren't viable for long form content. When competition in a different space started and Youtube was behind, we saw how large Tiktok became. Youtube is still trying and partially failing with its shorts feature.

If major creators leaving the platform doesn't effect youtube enough for it to change its poor policy, the overall experience will get worse on average for content creators and viewers. The content will become optimized to remove more and more adult oriented themes. That will cause a shift in the demographic.

I didn't say content disappears, I said it becomes harder to find when it gets age restricted. It becomes deprioritized in search results and doesn't get recommended. I've even realized that if you don't search for the correctly spelled terms, Youtube will often refuse to autocorrect to age-restricted content in the results. Which I saw based on my tests with certain videos of mine and the accounts of my friends and family. It was multiple tests over 15 accounts. It does sometimes show up but it's becoming more and more rare. I am surprised you get recommended age restricted content, I don't despite subscribing to the creators and do get recommended their other videos.

There have been plenty of video platforms with a great mobile experience that have gone under. And I’d argue that YTs mobile app isn’t even that good.

You can’t make technical argument for why something gains traction. Most people don’t choose a platform because it’s technically better.

> But "the algorithm" isn't a straw-man argument. I am saying viewers vs subscribers is a false dichotomy to care about.

I called it a straw man argument because you were arguing against a point I wasn’t making (though it was a little harsh calling it that on my part because I think this is really just a misunderstanding)

The point I was saying about viewers had nothing to do with algorithms nor anything else technical. It was just to remind people that the number of people on the platform is greater than the number of subscribers. So the network effect is larger than the figures suggest.

Discussions about algorithms et al are interesting but not related.

I think "the algorithm" is the only reason Youtube is popular at all. It's also the reason Tiktok became popular. "The algorithm" more than anything else explains both Youtube's popularity for long form content and its failure for short form vertical video content as well as livestreaming. "The algorithm" recommends new videos based on what you previously watched and that depends on creators emulating others and creating content in similar areas. "The algorithm" dictates not only what the viewer can find but also what creators will spend time making and what areas new creators will try out. It's hard to predict what new thing will become popular on Youtube but it is easy to predict when the performance of similar videos is so readily accessible. View count is the first thing anyone reads after the title and creator of a video by design.

That’s a fair point. :)

> When you deal with the scale that YT are operating at, automating content moderation is the only cost effective way.

Is this something that should be a limit to scale then? If you instead ran a petrol station that grew so large that you couldn't economically hire enough people to safely and fairly service all your patrons, you've reached a natural limit of the scale of the business. I think it's fair to say that if they're failing to moderate the platform in a way that doesn't harm it's users fairly regularly then they've reached a limit to their scale. Remembering as well that a lot of their users are clients, making business transactions with YouTube, not just users watching content.

How do you define “harm”? And who are YTs customers?

The problem here is that “harm” is subjective and YTs customers (advertisers) are not the same as their users (viewers), who are also different from their content creators.

I’m not defending YT here. Just explaining how it’s gotten to the place it is.

Harm I feel is when YT doesn't pay them for their contribution to the platform anymore, based on any myriad of reasons that the YT moderation can trigger falsely. There's also the extra complication of brand and personal impact of having videos removed or flagged falsely. Of course you have to work within the platform rules, but many people are doing exactly that and still getting flagged by the algorithmic moderation. Especially in the music YouTube space.

Content creators are working with YT for distribution of content in return for a cut of ad revenue, so it's more than an account on an internet platform like a forum, it's a business relationship, which would make them a client or vendor.

But this comes back to the point about who the customer is. YT are not harming his customers. They’re harming (from their perspective) those annoying expenses.

I don’t think you can equate it to a business relationship because those same relationships in broadcasting looks totally different (disclaimer: I used to work in broadcasting).

Content creators need to wise up that they’re not part of the business relationship in YTs business model. The fact they get paid at all is a by product.

The customer for YouTube is definitely the advertisers I do agree there.

I think the relationship between YT and creators is more akin to an event venue, definitely nothing like broadcasting. In venues, the relationship is symbiotic, without artists there would be no events, without venues, artists would have no where to host a large audience. YT similarly does not have an audience without it's creators, and so it requires their work.

YT is of course a mixed bag, since there are lots of accounts that treat it more like a video hosting platform. But for those making it their business to make YouTube videos, it's definitely a commercial relationship.

> The problem here is that “harm” is subjective

Sure, but we have national laws, regulations, and cultural norms that specify harms and YouTube just shrugs their shoulders and says "we don't care".

I agree it sucks but again, I feel compelled to remind people who YTs customers are.

Content creators benefit from YouTubes business model but they are neither YTs customers, employers nor even their intended users (ie the people watching the ads).

This is where the problem lies. Content creators are effectively just an expense. And it’s compounded by the fact that if one creator leaves there are plenty of others with big aspersions. So creators are an expense that’s easy to replace.

This is a question I think about a lot.

It seems to me that a tech company can only pick two of the following:

1. Free-to-use products 2. High-touch support 3. Profitability

I really just don’t see how free-to-use, instant sign-up services with billions of users can feasibly guarantee a human in the loop for every situation that arises.

Your position seems to be that if a company can’t provide human support, then it ought not expand their services. I can understand this position, but it has pretty major trade-offs: this would probably exclude large swaths of the world from uploading videos or using email.

There are many communities that 1) are unable to pay for email or video hosting services and 2) are not a big target of advertisers’ budgets.

That might be a shame but it could also be an opportunity for those countries to build their own ecosystems around it. As much as it sounds like a positive to the citizens of the other countries on an individual scale, YouTube and Google being a free global platform stops any chance of local alternatives popping up.

Though I am a massive advocate for making software more distributed and local, for the communities they are servicing. So it's worth noting I have that bias. I imagine a much more diverse software landscape, with more providers and more opportunities on smaller scales, rather than global behemoths that hoover up all the users and opportunities.

> When you deal with the scale that YT are operating at, automating content moderation is the only cost effective way.

At some point the Silicon Valley excuse of "we can't afford to do it" is going to run into dumb but powerful regulators who simply do not care and who will impose strict law to force it to happen.

YouTube has managed to persuade some regulators that the advertisers and the channel owners are solely responsible for ad placement, and that YouTube (which controls which products can be advertised, which places the ads, which takes most of the money for the ads, which has contracts with the advertisers and the channels where ads are placed, which earns most of its revenue through ads) has nothing to do with ads. That's worked for them in the US, but it's a fact that regulators in other countries are losing patience.

For example, YouTube needs a way to stop alcohol being advertised to children.

At the moment YouTube are saying that it's the responsibility of the drinks company and the channel to stop this happening. That works in the US, but it doesn't work in Europe, and YouTube saying "we don't know how old people are because COPPA" is met with "so what? It's still a problem, and you need to fix it".

Even in classical broadcasting in Europe there’s an element outside of the broadcasters control. In the U.K. we have a “watershed” where content gets more adult after 9PM. But kids can watch TV after then and still see adverts for alcohol.

The solution to that is the same as with YT, parents need to accept some responsibility to vet what their kids watch (I say this as a parent myself).

I’m not saying this excuses YT of any responsibility though. Just that this problem has always existed.

The watershed is a good example that supports my point.

We have law that says "don't advertise to children", and that's quite tricky so if a beer company says "we only placed ads in programmes that aired after 9pm which had adult themes" they're in the clear. If they place ads in shows at 7pm that are aimed at families they're in trouble.

The YouTube version is the channel saying "I have no idea who watches my channel because YouTube won't and can't tell me", advertisers saying "We advertise with YouTube, we expect them to place the ads responsibly, and we can't rely on YT analytics because COPPA means many children watch while logged into a parents account" and YouTube saying "we don't care, it's entirely the responsibility of the channels and advertisers".

It's this last "we don't care" bit that is causing problems. As the watershed example shows you don't have to have a bullet proof solution. You just need to make some good faith attempt to create conventions around ad placement that reduce the numbers of alcohol ads being served to children.

Advertiser have no less of an idea who watches videos on YT than they do on TV. Advertisers choose their slot on TV. They also choose their demographic on YouTube (it’s called “targeted advertising”).

Hence why you’re more likely to get ads about tech products on a tech video vs a music video.

> Competitors have always existed. They existed before YT got big, still exist, and new competitors have been created since.

Point is, for now Google has and have had for over a decade, a bottomless source of funding.

So they pushed YouTube as a free quality product for over a decade until every viable competitor was dead or irrelevant.

And now they can do whatever they want:

- awful moderation

- censoring

- remove downvotes for everyone just because their video was downvoted

- show 11 ads in a row

- etc

I used to like Google a lot.

A year ago I voluntarily dropped out of interviews with them after the first interview.

I don't want to work at such a place. As an ordinary guy who used to be poor I'm not sure I'd even be safe or fit in there.

When I was young I didn't understand why anti-dumping laws existed and why EU enforced them harshly. Today I wish they did against such abusers.

This is true for how GMail became dominant but most video sharing platforms were already free before YouTube, and still are now.

Sure there are some non-free services, but YouTube (even pre-Google) didn’t invent the free video hosting paradigm.

I think to address this specific situation all YouTube literally needs to do is to allow you to check your video for copyright/content/etc violations before upload in a robust way. That REALLY shouldn't be impossible for google, literally just run all the checks you're running now at upload time. Then if the robust check failed, a human moderator should be able to apply a manual override.

This sort of thing would both be more economical, allow you to get a more through review in advance, and actually be less arbitrary than wholly human moderation.

> There comes a point where Youtube is not saving money by automating their content moderation but Youtube decides to keep pushing forward. At some point a competitor that is good enough will arrive and once large creators like this one, around 1.5 million followers, almost every video with over 2 million views, 8 videos with over 5 million views, leave they will pay attention.

We're waiting for that competitor for last 10 years. It ain't fucking happening.

Hell, MS tried that with Mixer, they outright bought out few of the big content creators on Twitch, didn't worked.

Funny, last time I watched a YouTube Kids video, it was a tutorial on how to create a blowtorch using Febreeze and a lighter. The kid nearly burns his hand in the video because he holds it at the wrong angle.

> It's becoming clear that the moderation team is mostly software at this point.

I always assumed that the appeal process was manned by "Mechanical Turks" kind of employees that have to fill insane KPI quotas and don't have the time to do proper reviews. However most appeals probably just hits some brick wall before even reaching the staff.

Surely they should be asking creators whether they classify their vids as kid friendly?

For young kids I wouldn't want any swearing or other allusions to adult themes (not that I want my young kids watching YouTube).

If you want to aim at young kids the rules should be a lot stricter than the current ones.

The current mess is too lax for young kids and too strict for anyone older. And seems to treat educational and entertainment content equally. I want kids to know about the Holocaust. I don't want them watching a possibly racist comedy sketch about gassing Jews though.

I wonder when AI will reach a point where it can reasonably detect and classify more esoteric forms of humor. Deadpan humor is very difficult to explain to a computer, the reason it is funny is because the delivery sounds like it is not a joke. You have to intuit that the speaker would never ever believe what they are saying for it to be funny.

At some point even comments will become a signal Youtube uses to decide how to classify videos.

And no, there have been many channels over the years that have complained about the age-restriction issue even though they expressed clearly that the are a mixed-audience channel.

To show a much much more severe example, here's a tweet showing youtube misclassifying Happy Tree Friends as kids friendly, a cartoon style animated show about cute looking animals that get gruesomely killed in various cruel and disturbing situations. Clearly a parody of things like Tom and Jerry in many ways. Reasonable and artistically meaningful, especially when it came out in 1999, but clearly no human was in that misclassification process. They are 1 to 5 minute episodes.

https://twitter.com/Rat_Tail_Tony/status/1215015868515655680

>And no, there have been many channels over the years that have complained about the age-restriction issue even though they expressed clearly that the are a mixed-audience channel.

I know. What I'm saying is channels should have the choice of declaring themselves kid friendly and be judged accordingly. The current rules are too lax for kid friendly videos and videos are automatically being put in that slot which is the worst solution for everyone.

They already ask that

Reply on news.ycombinator.com