YouTube’s LGBTQ Problem Is a Business-Model Problem
onezero.medium.com
onezero.medium.com
s/(algorithmic|human) moderation/a rule of law/g
The solution is to not be the sole entity with the power to make decisions about unsolved social issues. Relinquish the power to make decisions about what kind of content is "acceptable" by implementing a rule of law managed by some kind of publicly accountable, transparent governance,
Google (and anybody else that controls something the public relies upon as infrastructure) has a ch9oice: keep the power to make socially-important decisio9ns and accept the blame and responsibility for those decisions, or abandon that power to make the whole mess someone else's problem.
They started making these decisions, because Capitalism.
They're publicly traded. Shareholders demand more profits. Advertising is the vast majority of their Youtube profits. Advertisers are risk averse. No fortune 500 company wants their ad running on white supremacist/violent/sexually explicit content. There is collateral damage, but caring about that only marginally increases profit.
Google's need to comply with these laws is part of the genesis for the moderation (e.g. copyright protection, child pornography).
In fact, aggressive government laws for censorship are why Google chooses not to operate in China.
The topic at hand is about monetization of ads -- the "acceptability" of the content is determined by the ad buyers. Not some idea of social good -- it's just the market.
The problem here is about execution not policy. The plaintiffs argue Google's algorithms are not working as intended, not that their intentions are misguided.
Are you suggesting that Google should create additional authority to regulate itself beyond what is legally required and needed for their business? E.g. create some expert panel that can compel Google to censor certain content? That would create more problems than it would solve.
Sooner or later, even if Search was still being censored, the other services would probably be blocked, or have unacceptable conditions imposed on them. The amount of effort to make YouTube and Gmail bend to PRC desires I think would have been a bridge too far. Once Xi Jinping took office especially.
I mean, I've seen super mild videos on Bilibili, where say a Mainland athlete or contestant loses to a Taiwanese one get removed. YouTube would be overloaded quite quickly by 50 cent army takedown requests.
There is some pretty crazy stuff on Chinese video sharing platforms, censorship in China is arbitrary and selective. That alone would make it very difficult for a western company to operate a media company in china.
The article is very long, but definetely worth a read. Section III and VII specifically detail some of the things going on with their China business.
The Tldr is that a big reason for google to stop their censored search in China was due to internal resistance as well as animosity from Congress.
This is mostly unrelated to what I'm talking about.
> The topic at hand is about monetization of ads
That's one (important) aspect of "YouTube’s LGBTQ Problem", but it is not the only issue. As others have mentioned, some of Google's attempts to "fix" the problem have, for example, been used as a weapon by various hate groups to get LGBTQ channels deleted (not just demonetized), or by fascists to remove anti-fascist reporting.
> the "acceptability" of the content is determined by the ad buyers
A big part of the problem is that this isn't actually true in practice. The "acceptability" of content is determined by Google's algorithm. They chose to use an opaque algorithm that they hope implements a very good approximation of what ad buyers consider "acceptable". As we've seen many times over the last few years, this has resulted in problems on all sides, including ad buyers.
> it's just the market
"just" a market ignores that markets are created in an legal environment that can be modified (e.g. adding, modifying, or removing regulations or policies)
> The problem here is about execution not policy.
The problem for Google is that they are choosing to have any policy at all. They are implementing opaque policies and thus are the only party that can be responsible for the resulting mess.
> The plaintiffs argue
Not relevant to my point;' I'm addressing the fundamental problem with Google's management of youtube (or similar platforms), not any particular case.
> Are you suggesting that Google should create additional authority to regulate itself beyond what is legally required and needed for their business?
More or less, yes. If decisions about monetization, video censorship, channel deletion, etc are going to be made, they need to happen through some kind of well-defined due process (aka being "ruled by laws"). Google could define that process unilaterally - which would be a major improvement from the current opaque and arbitrary status quo - but the best option would be to start some type of open/public entity/process for creating and managing what is meant by "due process on youtube".
> create some expert panel that can compel Google to censor certain content?
That already exists! That panel is whatever group inside Google chose the algorithm(s) that has been demonetizing and/or censoring videos over the last few years. I'm suggesting that transparent, publicly accountable panel is better than a hidden panel of unknown Google employees. Yes, both types will have problems, but at least we can see and attempt to fix the panel when it is public.
I guess basically what I'm saying is that Google's content moderation problem is really a problem with our society, our politics, and our press.
[1] https://www.buzzfeednews.com/article/ryanhatesthis/will-sili...
[2] https://www.buzzfeednews.com/article/mitchprothero/isis-rese...
Were those activists you mentioned campaigning to explicitly get respected analysts removed? Not from what I remember.
Lets compare this with a hypothetical: Lets say activists successfully got gov to newly introduce a speed limit. Some time later, a huge chunk of the emergency service personnel is in prison for breaking this new law. Would you actually blame this on activist's? Or maybe a retarded court system, probably combined with incompetent legislators.
Back to Youtube. They are either incompetent at making rules or incompetent at reasonably enforcing them, probably cause they entirely offload the latter on some shitty algorithm. Don't get me wrong, algorithms are great as first heuristic to save work. But there has to be a way to get to higher court levels, even if there is some form of cost.
A better analogy would be if campaigners demanded the police be forced to arrest anyone who broke the speed limit with no exemptions, called anyone who opposed them supporters of dangerous speeding, and then were shocked whem most of the emergency services ended up in jail.
I don't think there's even any way around this problem. The activists want anyone who reports on the far right in ways they consider excessively sympathetic to be banned too, and you can't implement "how activists feel" as the policy dictating what's allowed on something the size of YouTube. It just doesn't scale.
Savvy YouTubers (especially those who produce content that could be perceived as controversial) have taken control of their own revenues through sponsorship and patronage. YouTube doesn't have a business model problem - YouTubers do. You can't expect a platform to sell such vast amounts of ad inventory while also judging each piece of content in a nuanced fashion. That's the devil's bargain of algorithmic advertising and it isn't going away. If you don't like that bargain, sell your own damned inventory, or cut advertisers out of the loop completely and ask your viewers to make a direct contribution.
I have not yet observed much progress towards these rules. The few "ministry of truth" style steps that have occurred don't exactly look promising https://en.wikipedia.org/wiki/Netzwerkdurchsetzungsgesetz
I wonder what all this paternalism will do to society long term.
In the short term we mostly see the people who are agitated by being on the wrong side of popular opinion (or are just caught by hamfisted enforcement), or others who expect levels of protection they're still not getting. But that's just the exceedingly obvious parts.
Every social intervention risks unintended consequences, no matter how well intentioned and angelic the goals are.
Censorship in TV and movies have irreparably damaged our collective psyche for generations (e.g.: the most obvious and uncontroversial area to say it's warped us are in attitudes around sex).
Where does social media censorship go from here? And what effects will it have on the population at large over time?
This is also the case with African Americans, and there is a similar problem of machine learning flagging African American content as hate speech. https://www.vox.com/recode/2019/8/15/20806384/social-media-h...
The problem may be that we cannot depend on machine learning algorithms to do an acceptable job of moderation, and that a lot more humans need to be hired as moderates. However, doing that really changes the economics of the platform
Why can't there be a balance between full automation and human review, such as an appeal system or 'whitelist' program? Are the economics really the obstacle, or is it more of a corporate stance towards not adjudicating subjective controversies to avoid liability under the Communications Decency Act Section 230?
I fail to see how a "but both sides" proclamation leads to any progress (other than maintaining the status quo which is awful good for a select group of people).
Because this will devolve into different groups flagging their opponents' videos. This will especially hurt the voices of minorities and other vulnerable populations.
- if there are sufficient flags, trigger a moderator review
- flaggers who correctly flagged content get +1 score or -1 score if not
that would make it harder for people to consistently flag content maliciously. Like age-old moderation systems such as calling 911 or police. If you abuse calling 911, you can get fined.
I imagine this has been tried before, and am genuinely curious why this is worse than flawed ML.
https://www.adweek.com/digital/youtube-heroes-will-let-users...
But it didn't work. YouTube is too diverse and too popular, it has no clear "naturally" anointable moderator types.
But at some point, sites decided that experimenting with stuff like that was unimportant compared to just bringing in more users. Letting everyone vote all the time obviously has huge issues, but it's emotionally satisfying to downvote, so it keeps users engaged.
The people most prone to dominate the site with voting on everything and coordinating efforts, are also the ones which will complain most loudly if they don't dominate the discussion. So just give them what they want.
It worked. Experimenting with self-moderation mechanisms went extremely out of fashion.
I'd love a system were you could get 2 vote choices per post and then have 8 or more vote options.
Like/dislike (or agree/disagree), off topic/on topic, factually correct/factually incorrect, funny/hateful, user is a prick/user is a swell guy, or whatever seems like a good option for the discussion forum.
I mean, something can be funny, but off topic at the same time. Or something can be factually correct but the person making the post can be a real jerk about it.
I also have a hard time believing the GP that there is no experimentation anymore. Probably a lot more subtle, but you can do a lot of stuff behind the scene without giving feedback to the user. I'd be shocked if there is not at least some form of shadowbanning bad actors.
Not to mention that moderating this kind of stuff can seriously mess you up: https://www.theverge.com/2019/6/19/18681845/facebook-moderat...
You end up looking at not just mean words but CP, seriously gory stuff, fringe sexual deviancy, and more. Do we really want to do this to moderators?
I don't know specifically, but there must be something.
This is just not a difficult problem, YouTube just doesn’t want to solve it
If they started with that YouTube wouldn't have gotten off the ground as people wouldn't go with the "unilaterally acts like they are being controlled by the Chinese Government" service.
I don't like the fact that YouTube with the recommender and Autoplay is stochastic mind control. It's programmed to make people watch loads of YouTube, and the mechanism that it has discovered to do that is to nudge people to obsess over fringe things, causing them to watch videos about these things for hours per day.
What if the filters intended to prevent hate speech end up harming the targets of that hate speech instead?
What if YouTube's algorithms push conspiracy theories and manifestos of domestic terrorists into people's feeds?
These are not "deal with it or leave" issues; these are existential crises that we must hold platforms accountable to solving.
Also, you'd be surprised how smart kids are if you give them the chance, have the right UI, etc.
I'm actually surprised the HN community seems to look down on crowdsourcing these efforts, kinda humorous.
This is why civilized countries have developed rule of law, including arguments and rebuttals, overseen by an impartial arbiter.
Yes, that is costly. But it is the accepted solution we have as a society, and it's time "Big Tech" starts accepting rule of law.
Which laws? I will acknowledge they are gaming the system somewhat by claiming to be an impartial platform (thereby gaining copyright protections) while removing what they find distasteful (but not illegal, so moderation based on opinion). However, it sounds like you believe removing videos that are not illegal in and of themselves is itself illegal. We don't have a law prohibiting that, so I'm not sure to what you are referring. Protecting speech also means protecting the right not to speak.
it sounds like you believe removing videos that are not illegal in and of themselves is itself illegal.
It's not even on my radar. I'm concerned with the transparency and consistency of the process, not the outcome of individual decisions.
Yes, tech companies should accept the rule of law. But people need to accept that the rule of law does not preclude running a site full of angry, bigoted idiots with no productive commentary.
Classic example being that in some areas of the UK, close friends will greet each other with a joking insult. If you don't know they're friends or you don't have the verbal clues to work from (the 'I'm joking' intonation), it looks a lot like the start of a bar fight.
If humans can't get 100% accuracy on this - how can we expect it from business rules, machine learning and so on?
To pinch a line from ST:TNG, "Life is an exercise in exceptions".
Getting profit-loss data on Youtube is very difficult. If they are profitable, it can't be by a large amount.
So, if the plan is to have a LOT of real people just trying to monitor YT all day long it will really gut YT's economics, not just change it.
Destin of Smarter Every Day on YT went and talked to FB and their content mod team a few months back [0]. He talked with Rob Leathern of FB's Cyber Security Department. Rob mentiond that they now use ~30k real people to monitor FB. Soon thereafter, The Verge released a scoop on the horrific conditions these 30k people must endure [1] in that job.
YT would likely have to have a proportinate number of people in similar conditions as well, or likely scale up to make things less worse and then have to pay more people.
YT simply cannot possibly afford to have enough people moderate YT.
Likely, this version of YT will become something of a blip in history. All this investment will likely not pan out as the economics of video platforms like YT cannot work as they are today. The real question is what is the subscription amount that does make YT work the way we want it to work?
[0] https://www.youtube.com/watch?v=FY_NtO7SIrY Also, watch Destin's body language with Nate Gleischer (~14:30 mark). It's so strange. Something wierd happened in that interview with Nate.
[1] https://www.theverge.com/2019/6/19/18681845/facebook-moderat...
For example, YouTube doesn't want violent content. Just having the word "kill" in your title can get the content demonetized, even if the content is about a video game and properly categorized as such. It's been this way for years.
The rationale “we can’t make as much money as we’d like to if we followed the law” just won’t wash.
Citation very much needed.
They also need to not censor so much that it's a publisher and maintain themselves as a platform legally.
I don't think that that's preferable for society.
Imagine a YouTube that cost $10 to upload a video and $1 per comment. The quality of the content would be much higher and the hate speech would drop to zero. Wouldn’t that be preferable for society?
You're effectively restricting speech to the rich, and the poor just have to sit there and listen. That would be sad and unnecessary.
I create content for YouTube and if I had to pay $10 per video then I would never make another YouTube video again. It doesn't matter that there are thousands of people (and tends of thousands+ sometimes), because I simply couldn't afford to spend time and effort to make a video and then spend extra money to share it.
You could make the argument that perhaps society is better off if I don't create content, but there are at least some people that claim they enjoy it.
Isn't that basically Vimeo's business model?
Video classification by YT is a bit more tricky, since they themselves are one of the involved parties and thus any incentives put in place to improve the system would cost YT.
Google and company has no reason to not do the minimal acceptable amount. They aren't bound by morals, nor do they owe "society" anything. Shareholders also demand more and more profits, and so the result you see is the inevitable outcome, and it can't be any other way. Unless "society" wills it by voting in people who then make new laws to change the outcome.
To use a historical example, it's easier to moderate an IRC channel of 100 people than an IRC network of 10,000.
Usenet only fell apart because of the spam, and legal attacks by the RIAA and MPAA and BSA.
I found the moderated sci.* and comp.* groups more pleasant for asking questions and giving answers. The alt.* hierarchy had its moments, but mostly it was an exercise in tolerance. Some groups were largely self-moderating, others had such a poor SNR that there was no point putting the time in.
The RIAA/MPAA/BSA raids were only really relevant if you were into pirating copyrighted material.
I think that this actully is violent content - so this is actually an example where the algorithm worked fine (which does not contradict your argument that there exist false positives - they are just a lot more subtle).
That's not exactly subtle.
I understand there could be other reasons for being demonetised, but it's hard to see what it could be in a very very inoffensive video about how to do a vintage dance.
If you are referring to the game example given some posts above: "Killing" means that the respective object was alive before - a property that a space ship with no living beings onboard does not have. So the respective author named and labelled the video improperly.
To earn "ad bucks", you have to watch an ad. The ad watching takes place in a dedicated ads section of the site, completely independent of all videos.
The vast majority of monetized videos would cost 1 ad buck, so watch one ad. The ad watching is just a giant "watch ad" button. The algorithm goes through all all the info about me and shows me a relevant ad, just like it does already.
If you want to opt out of ad bucks, you can just subscribe to that channel and pay some money. YouTube takes some, the creator gets the rest.
Why not just use this model instead of this politically driven sham that currently exists?
I'm the viewer, I'm the product. My attention has value - so why not let me to worry about what videos I want to support and leave the corporations out of it? YouTube is nothing without the viewer.
If YouTube is a platform and not a publisher, then they should act like the phone company. If I want to call some psychic hotline, just connect the call and take your cut.
Advertisers don't care about what ads people want to watch. They care about what ads convert.
In my view this should eliminate the ad blocking group entirely.
I guess the solution is to use some technology to ensure that the viewer is watching the ads, like in Black Mirror S01E02.
Maximum ad balance = 3 ad bucks
Watching an ad requires clicking the watch button, so you can't queue them up.
People can walk away, minimize or block the ad even now.
But the youtube isn't a public institution - it's a private platform - does that apply as well even if they would put e.g. banned topics/words in their terms and conditions document? (I'm just curious)
In fact, on top of human raters exhibiting classification problems in the face of ambiguity, there is also the issue of bias. Radiolab on NPR had a great series on Facebook's human censors that shows it just opens up a different pandora's box.
There is no clear, non-messy way to do censorship that doesn't end up with a sizable number of people mad, and with some subset of people hit by accidental or deliberate false positives.
I'm afraid there's no easy answer and it will inherently be constant battle and fight between sides.
How do you propose making advertisers happy?
Easy to cry censorship; hard to address real business issues.
1. Don't have "autoplay". Make the user choose. Autoplay is not good for society. "Tip your head back, turn your brain off, and let the algorithm feed you what will keep you here."
2. Give the advertisers more fine grained controls on what types of videos their ads show on. If a stink goes up because Dove doesn't want to be on a PewDiePie video, then that isn't Google's fault. It's Dove's. Let the advertiser add PDP to the blacklist, or make a more exclusive whitelist.
It's like Twitter. "Oh, oh no... how do we stop people seeing all this awful content?"
A) Stop deciding for the user (and therefore the entire website) what is awful
B) Give the user better blocking controls.
However, this won't happen, because Google and Twitter like being able to shape what the "algorithm" shows you, and keeping that control in their hands. This is why the chronological timeline isn't coming back.
https://news.slashdot.org/story/19/08/15/2058257/advertisers...
Google recently won a lawsuit which complained YouTube was putting conservative content in "restricted" mode.[1] Others complain Google is not doing enough to avoid shoving too many extreme videos at people.[2]
[1] https://www.reuters.com/article/us-alphabet-google-youtube-c...
[2] https://www.nytimes.com/2019/08/11/world/americas/youtube-br...
https://twitter.com/gaywonk/status/1134264395717103617?lang=...
There are homophobic ads on YT? What does that even look like? Can you link to an example?
(Moderators will often, but not always, say if they've rescued a page. It's possible this got aggressively flagged off the front page and was rescued.)
No?
> In Poland, for example, Youtube, in wholesale, blocks channels which are criticize on LGBTQUERTY.
YouTube is US based and hate speech is looked down upon in the US. Don't like it? Find a service not based in the US.
> The question is why the are able to censor content.
It's a private website, they can do whatever they want. Again, find another site if you want hate content.
> I don't expect that my supplier turn off gas for me, if he thinks, that I use it to kill people; it's not his business, ot's the matter of law enforcement, the police, etc.
???
A) if they know you're using it to kill people they would be an accomplice in the eyes of the law
B) they would have every right to refuse service
Vendors always discriminate on what they ship to customers. It is never an open platform. I know it's very confusing, but Youtube literally says that you can only distribute certain kinds of things using their service and that they will block other things. I suppose they don't put it in 40 point font on their front page, but it's also not hard to find: https://www.youtube.com/t/terms (incidentally this came up in Japanese for me, so I suppose it might come up in Polish for you).
When people talk about freedom of speech and censorship, what they are talking about is society's role in gagging you. Generally speaking, usually countries don't force you to carry specific content (ha ha! Unless you are Canadian.... Canadian inside joke...), but likewise nobody is forced to carry all content, or accept all trade that might come their way. You are allowed as a business to choose what you want to do and who you want to do business with -- unless it's with a protected segment of society. That's a pretty basic freedom and someone who clearly values freedom as you do can surely see why it is important.
You might argue (to greater or lesser success) that Youtube is a pubic carrier and should be forced to carry everything for free... I suppose... But I'm betting you'll be in the "lesser success" category there. You might argue that your particular subgroup of people who criticise minority gender issues are worthy of special protection and therefore should have a protected ability to spread your message... I suppose... But seriously, I don't think you'll get very far with any of it.
From my perspective, I just think Youtube is within their rights to do what they want on this kind of issue. We may like it or dislike it, but it's the only sane way to handle it.
https://gizmodo.com/youtube-no-we-wont-remove-these-videos-o...
(And no, they're not blocked in Poland.)