Hacker News Reader

Comment by PragmaticPulp | Hacker News Reader

> We need things to change here. We need Google to moderate their services properly,

Human moderation isn’t perfectly consistent, either. It would be extraordinarily expensive to expect YouTube to have humans moderate everything manually.

What are people willing to give up in exchange for a human-moderated version of YouTube? Would we be okay if they charged a $3 “moderation fee” on every upload to cover the cost of paying someone to watch the video and evaluate how it fits YouTube’s rules? The big creators wouldn’t care, but the small time videos would immediately stop being uploaded and competitor websites would become popular overnight.

I don’t see demands for perfect human moderation of YouTube uploads as realistic in any way.

View on HN

No one expects things to be perfect or that there will be no mistakes, but there's a whole lot of space between "perfect" and "braindead stupid".

The worst part isn't so much that mistakes get made, but the way those mistakes are dealt with, which is often a frustrating and silly experience.

Exactly. A rogue moderator might make the wrong call on a video, but unlike an algorithm that call won't affect hundreds of videos at once. And also unlike an algorithm the moderator might be a different person the next day. So mistakes are not magnified the way they are with an algorithm.

> No one expects things to be perfect or that there will be no mistakes, but there's a whole lot of space between "perfect" and "braindead stupid".

There is but I am not terribly convinced the difference is employing a badly paid human.

Could this be a browser plugin? No idea how "invasive" a plugin can be, but ad-blockers and privacy plugins seem to be able to change a lot on webpages (including blocking ad videos on Youtube).

So theoretically it seems possible to build a plugin that moderates comments on web pages ... "No one expects things to be perfect" ... so it does not hurt to try?

> It would be extraordinarily expensive to expect YouTube to have humans moderate everything manually.

You could flip this around, however? Why is YouTube promoting videos that no responsible human has reviewed?

These videos don't just trip and fall into people's feeds. They're selected by YouTube's recommendation algorithm, presumably targeting maximum engagement. That decision is automated, but it's still an editorial decision -- it's not based on an objective, chosen criterion like "most recent" or even "most viewed."

This isn't how YouTube is currently structured (nor the other social media platforms), but I think it's fair to suggest that if a platform is in the business of picking and choosing winners, it should bear at least a moral responsibility for the content it promotes.

YouTube could set a views threshold to make manual reviews scale. My guess is 99% of what people watch is the top 0.1% most popular videos which makes things manageable.

Highly dangerous video seen by 5,000 people not a big deal. Highly dangerous video seen by 5,000,000 people is.

And this small percentage of popular videos are made by a smaller amount of reputable or disreputable channels. Give me the tools and a few months and I'll clean up the English-speaking toxic content by pruning it at the source. But that won't happen because removing these dangerous, stupid, click-bait farms hurts the business. Fine, whatever. But stop claiming moderation is "impossible" or "too expensive." That is a heinous lie.

> Would we be okay if they charged a $3 “moderation fee” on every upload to cover the cost of paying someone to watch the video and evaluate how it fits YouTube’s rules? The big creators wouldn’t care, but the small time videos would immediately stop being uploaded and competitor websites would become popular overnight.

That would actually be pretty ideal - most consumers watch YouTube for the bigger channels and popular content anyway, so they would stay on the platform, which would validate that revenue model. The smaller producers and consumers would both leave, which would help build competitors to YouTube. The big producers who remain will be slightly disincentivized to spam low-quality videos (including those text-to-speech automated ones that some people here have found). And, Google will have no excuse for not using humans to moderate videos anymore (can't cite "cost concerns" when you're literally pricing it in).

Seems like a win-win-win to me.

> Human moderation isn’t perfectly consistent, either. It would be extraordinarily expensive to expect YouTube to have humans moderate everything manually.

Those aren't just random channels though, they are top youtube channels, surely they can manage to review the most popular ones at least...

And we're talking about tips which can make people die if they follow them, that's the most clear cut case, there's nothing opinion based.

There's just no excuses there.

> I don’t see demands for perfect human moderation of YouTube uploads as realistic in any way.

Then don't add the word "perfect."

I'd be happy with a law that said Youtube were responsible for people who hurt themselves copying videos which they were recommended which were wildly unsafe.

If the result of that is Google has to charge people a moderation fee for videos they then recommend, then fine, that's the cost we have to pay.

What about a $3 fee that automatically comes out of the ad revenue when your video exceeds $Arbitrary limit (they could start at 200,000 views to start with).

Moderation is very complicated especially since it can be subjective based on the background and skill level of who is watching the video. Would it be possible to crowd source this? For example, I could add a filter to my search saying exclude 'profanity' and videos with profanity would be removed. Or exclude videos that are 'dangerous' or 'for entertainment only' and those would be removed. My thought is that we should use the shear numbers of persons watching the videos to our advantage. Honestly, I sort of do this by looking at the comments since many times you can gather the flavor of the video based on the comments. This could also be used to political bias. Videos with a right-wing leaning would get tagged as leaning right and vice-versa. Also, content creators would be incentivized to accurately tag their videos and perhaps the algorithm could penalize content creators with discrepant self-tag versus what viewers tag as.

Anything "crowdsourced" can and will be gamed by spammers. Such tags would become useless very quickly without traditional moderation.

Reply on news.ycombinator.com