5 ms·

Imperfect moderation is inevitable at the massive scale of social media companies. YouTube receives an enormous quantity of video every day and it’s clearly imp

Imperfect moderation is inevitable at the massive scale of social media companies. YouTube receives an enormous quantity of video every day and it’s clearly impossible to have humans review all of it. If we humans did review every single minute of video, the subjective element of moderation would still lead to headlines like this where someone could identify instances of inconsistencies. Given that she re-uploaded the video and I’m watching it on YouTube right now, I assume the initial moderation strike was more of a mistake. Or more likely, maybe it was reported by a number of people who were upset by it for some reason and it entered an auto-moderation queue. I have a hard time getting too worked up about news stories of singular moderation mistakes. Does anyone realistically expect YouTube to have perfectly consistent moderation? Even HN, a significantly smaller community, has very uneven and inconsistent moderation due to the user-influenced flagging features.

JCharante 4y ago
How hard is it to use manual review for videos that have at least a million views?
- PragmaticPulp 4y ago
  What makes you think they don't use manual review for highly-viewed videos? The video in question wouldn't have triggered that manual review threshold because it's not even close to a million views. I would actually assume that highly-viewed videos are moved up much higher in the manual review priority queue, as it only makes sense. The controversies like the one in the article are generally among moderately popular channels with 10K - 100K views, not in the millions.
  - Dylan16807 4y ago
    > The video in question wouldn't have triggered that manual review threshold because it's not even close to a million views. What makes you say that? I don't have a source on that specific video, but socialblade says her channel was getting hundreds of thousands of views per day in the week after that video went up, three weeks since her previous video. Edit: Bing cache says 720k as of some point. Archive.org says 1.5 million as of three days ago. I would have set the manual review threshold no higher than 100k anyway.
poulpy123 4y ago
nothing is perfect but google has a market capitalization of 1.5 trillion dollars, i'm sure they can afford to remove of put warning on dangerous videos
Zigurd 4y ago
That's true. However, bad moderation is most harmful when clickbait-for-profit gets profitable enough to have their boiler room crew mass-report responsible people debunking the clickbait. This is the bad stuff. It's also the stuff that can be caught with enough anti-spam effort. The example in this article fits that model: A toxic creator farm getting away with a coordinated mass-reporting campaign against an individual. It's just bad anti-spam and YouTube deserves the criticism.
wmeredith 4y ago
> Imperfect moderation is inevitable at the massive scale of social media companies. YouTube receives an enormous quantity of video every day and it’s clearly impossible to have humans review all of it Then maybe they shouldn't spread it across the globe?
MetaWhirledPeas 4y ago
> YouTube receives an enormous quantity of video every day and it’s clearly impossible to have humans review all of it. Micro moderation might not be possible, but meta moderation certainly is. If a story such as the fractal burning has made the news that means it has passed through plenty of human brains along the way. That sort of thing should be noticed by meta moderation. This reminds me of the YouTuber who uploaded 2 Million Videos over the course of years and was not shut down. https://www.youtube.com/watch?v=Olkb7fYSyiI https://www.youtube.com/watch?v=Olkb7fYSyiI The only way for that not to be dealt with is by gross negligence.
swatcoder 4y ago
> Imperfect moderation is inevitable at the massive scale of social media companies. YouTube receives an enormous quantity of video every day and it’s clearly impossible to have humans review all of it. Yes, this is the root of the content tech boom of the last 20 years: we’ll cut corners to make publishing and distribution as cheap as possible and displace the remaining cost to third-parties (advertisers, data buyers) so that nobody actually involved in the content has any accountability. And the negative consequences of this approach are rapidly becoming clear to society. We’ll be spending the next 20 years trying to bring those consequences back in line with our values and societal needs. It’s like technical debt, but for societal projects instead of engineering projects.
jameshart 4y ago
> YouTube receives an enormous quantity of video every day and it’s clearly impossible to have humans review all of it But not all of that content gets picked by the algorithm to be shown to millions of people. They don’t have to manually check everything. But if they are considering putting something at the top of their homepage for a large number of people, would it kill them to have someone give the content a once over to see if it might be dangerous?
pessimizer 4y ago
> Imperfect moderation is inevitable at the massive scale of social media companies. If social media companies can't function well at that scale, what's the reason for having social media companies at that scale? We should set and enforce expectations, and when companies get to the scale where they can't meet those expectations, fine them until they "rightsize."
- hedora 4y ago
  I'm a fan of removing legal carve outs for social media companies. If it's hosted on your domain, then you are liable, just as a book publisher would be. If people want to publish stuff, they can buy a raspberry pi or an s3 bucket. If people want to moderate content, they can hire an editorial staff (like magazines and podcasts do). I'm not sure how to handle Wikipedia. It's a net positive for society, and this proposal would effectively ban it. That's the only impacted site I can think of that I think is worth saving.
  - Nextgrid 4y ago
    > I'm not sure how to handle Wikipedia. It's a net positive for society, and this proposal would effectively ban it. The rule could only apply when you profit from impressions from the content. Platforms that rely on ad revenue tied to views would be covered (as the forbidden content is directly driving their revenue), platforms that are either free or use an alternative monetization model not dependent on impressions (flat monthly fee regardless of which content was consumed or how many times) would be exempt.

JCharante 4y ago

How hard is it to use manual review for videos that have at least a million views?

PragmaticPulp 4y ago
What makes you think they don't use manual review for highly-viewed videos? The video in question wouldn't have triggered that manual review threshold because it's not even close to a million views. I would actually assume that highly-viewed videos are moved up much higher in the manual review priority queue, as it only makes sense. The controversies like the one in the article are generally among moderately popular channels with 10K - 100K views, not in the millions.
- Dylan16807 4y ago
  > The video in question wouldn't have triggered that manual review threshold because it's not even close to a million views. What makes you say that? I don't have a source on that specific video, but socialblade says her channel was getting hundreds of thousands of views per day in the week after that video went up, three weeks since her previous video. Edit: Bing cache says 720k as of some point. Archive.org says 1.5 million as of three days ago. I would have set the manual review threshold no higher than 100k anyway.

poulpy123 4y ago

nothing is perfect but google has a market capitalization of 1.5 trillion dollars, i'm sure they can afford to remove of put warning on dangerous videos

Zigurd 4y ago

That's true. However, bad moderation is most harmful when clickbait-for-profit gets profitable enough to have their boiler room crew mass-report responsible people debunking the clickbait. This is the bad stuff. It's also the stuff that can be caught with enough anti-spam effort. The example in this article fits that model: A toxic creator farm getting away with a coordinated mass-reporting campaign against an individual. It's just bad anti-spam and YouTube deserves the criticism.

wmeredith 4y ago

MetaWhirledPeas 4y ago

> YouTube receives an enormous quantity of video every day and it’s clearly impossible to have humans review all of it. Micro moderation might not be possible, but meta moderation certainly is. If a story such as the fractal burning has made the news that means it has passed through plenty of human brains along the way. That sort of thing should be noticed by meta moderation. This reminds me of the YouTuber who uploaded 2 Million Videos over the course of years and was not shut down. https://www.youtube.com/watch?v=Olkb7fYSyiI https://www.youtube.com/watch?v=Olkb7fYSyiI The only way for that not to be dealt with is by gross negligence.

swatcoder 4y ago

> Imperfect moderation is inevitable at the massive scale of social media companies. YouTube receives an enormous quantity of video every day and it’s clearly impossible to have humans review all of it. Yes, this is the root of the content tech boom of the last 20 years: we’ll cut corners to make publishing and distribution as cheap as possible and displace the remaining cost to third-parties (advertisers, data buyers) so that nobody actually involved in the content has any accountability. And the negative consequences of this approach are rapidly becoming clear to society. We’ll be spending the next 20 years trying to bring those consequences back in line with our values and societal needs. It’s like technical debt, but for societal projects instead of engineering projects.

jameshart 4y ago

> YouTube receives an enormous quantity of video every day and it’s clearly impossible to have humans review all of it But not all of that content gets picked by the algorithm to be shown to millions of people. They don’t have to manually check everything. But if they are considering putting something at the top of their homepage for a large number of people, would it kill them to have someone give the content a once over to see if it might be dangerous?

pessimizer 4y ago

> Imperfect moderation is inevitable at the massive scale of social media companies. If social media companies can't function well at that scale, what's the reason for having social media companies at that scale? We should set and enforce expectations, and when companies get to the scale where they can't meet those expectations, fine them until they "rightsize."

hedora 4y ago
I'm a fan of removing legal carve outs for social media companies. If it's hosted on your domain, then you are liable, just as a book publisher would be. If people want to publish stuff, they can buy a raspberry pi or an s3 bucket. If people want to moderate content, they can hire an editorial staff (like magazines and podcasts do). I'm not sure how to handle Wikipedia. It's a net positive for society, and this proposal would effectively ban it. That's the only impacted site I can think of that I think is worth saving.
- Nextgrid 4y ago
  > I'm not sure how to handle Wikipedia. It's a net positive for society, and this proposal would effectively ban it. The rule could only apply when you profit from impressions from the content. Platforms that rely on ad revenue tied to views would be covered (as the forbidden content is directly driving their revenue), platforms that are either free or use an alternative monetization model not dependent on impressions (flat monthly fee regardless of which content was consumed or how many times) would be exempt.