3 ms·

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.

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.

freyir 5y ago
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.
- strogonoff 5y ago
  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.

freyir 5y ago

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.

strogonoff 5y ago
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.