Hacker News Reader

I would argue it's much, much, worse than what you're saying. I mean this part:

> Google will never do it, because it would 1) require paying humans to do work, which is expensive, and 2) it wouldn't drive engagement and generate clicks.

Every regular business out there needs support. In the form of pre-sales, sales, post-sales, actual support folks, etc.

Google's entire business model is predicated on there being no meaningful human support.

If they're forced to implement the level of support their worldwide, what-they-consider-top-notch operation would actually require, their business model goes bust.

Ok, I'm probably exaggerating, but their profit margins would go down from ~25% to probably something like 5%.

They will <<never>>, ever do it, unless someone puts a legal gun to their head.

Thats the biggest problem with all of this stuff. They designed their whole business to not pay people. Not only their margins would shrink, but the perceived value of the company would shrink _a lot_. They will never do it.

>Every regular business out there needs support. In the form of pre-sales, sales, post-sales, actual support folks, etc.

While I'm not privy to YouTube's organizational structure, I'd be very surprised if their customers don't receive plenty of human support.

Note that you are not their customer. Advertisers are.

The fact that they regularly ban accounts and publishers kinda of goes against that.

And I am a customer, I pay for their services.

> So they just take the easy way out, so they can keep on doing what they do.

I don't think this is a good takeaway, they can always replace bans with deprioritizing, so banning is a conscious choice.

> Social networks seem to be unable to optimize for anything other than "engagement,"

This is mostly due to the profit motive. Optimizing for time spent on the platform benefits shareholders.

I agree that adding human reviewers could help; but it will not resolve the root cause of YT being designed to boost disinformation in the name of engagement.

It’s encoded in their business model: the paying customer and the source of revenue is the advertiser. Thus, their interests are radically different from those of the users.

I believe there is something fundamentally wrong about such models, especially applied at scale, that I would not even oppose some light, focused regulation that brings it in check.

Otherwise, adding human moderation could maybe help in short term but the business model with find its way to ruin things again.

It may be an outcome, but it certainly wasn’t ”designed to boost disinformation.” If you watch a pro-vaccine videos, it will suggest more videos portraying the vaccine positively.

My take on things as time has gone on? It’s depressing but the problem is people. It doesn’t matter if you’re watching a pro-vaccine or anti-vaccine video, the comments are filled with hundreds or thousands of anti-vaccine comments. People are the common denominator of what’s wrong with the Internet today.

I agree it’s imprecise to say it is designed to boost disinformation (which I quoted from the parent); it is more precise to say it was designed to maximise revenue.

Which, due to the business model, is more or less incompatible with presenting the truth. Truth is often uncomfortable, full of caveats and subject to change; rarely does it make for a good story. If it turns out that presenting outright false information (including lying by omission, etc.) as established fact keeps eyeballs for longer and engages more people, then the algorithm would be doing just that.

Sure, if you watch technical content and train the algorithm by watching videos from solid, trusted sources, you can (for now) get it to give you sound recommendations, but I doubt this is reliable or representative. Thus, I believe acting as if YouTube was designed to boost misinformation is not necessarily wrong, and may be more or less an approximation of the average case.

And the problem is not quite people. If people are expected to constantly counteract the efforts of a sophisticated machine that doesn’t have their interests in mind and can spend a million dollars just optimizing its GUI and algorithms, which are then deployed against each individual user (who comes alone, tired after work, low on willpower and with lowered defences from the safety of their home) to exploit the darkest in them in order to show them more ads, then the problem is not quite people—the problem is that we are dealing with an adversary.

You assign this to social media, but it's is seemingly true of politics as well. (Reference vaccines in 2016 GOP Primary debate about vaccines)