Re:YouTube algorithm recommends videos that violate its own ...
by acituan on Wednesday July 07, 2021 @07:22PM
(#27766897)
I think these are great questions, because there is a way it touches on fundamental values such as the freedom of speech.
Human attention/willpower/judgement etc are exhaustible resources, and they don't scale like a machine does. Even if the machine is conveying mere human utterances and likeness, the recommendation itself is ultimately the "speech" of a machine, which was conjured by thousands of highly paid engineers and data scientists to do some "magic" of keeping you talking to it. The attractiveness of the recommendations of a machine can out-tire the willpower of the reciprocating human, especially considering the ubiquitousness of the machine; be it on your tv, phone, car, videos, music, news, social media.
So I think it is possible to formulate the reaction you see as a defense against building machines that abuse organic human attention. Going back to free speech, there wouldn't be defending the right to free speech of the machine if it could denial-of-service the recipients' judgement with all the engagement tricks in the book and at scale.
And the spirit of the objection is not about the radicalness of the content; but rather the frequency of which such content as brought to the user by the machine. In a free marketplace of ideas of humans, such content would a) have lower incididence than youtube b) would have much different propagation/decay function than youtube c) you could give direct, high fidelity feedback to those humans. People's ability to filter out bullshit collectively in a real social setting is much stronger than our 1:1 encounters with youtube. There is no talking back to youtube, there is no friends you can together vet the ideas you encounter on youtube.
Human attention/willpower/judgement etc are exhaustible resources, and they don't scale like a machine does. Even if the machine is conveying mere human utterances and likeness, the recommendation itself is ultimately the "speech" of a machine, which was conjured by thousands of highly paid engineers and data scientists to do some "magic" of keeping you talking to it. The attractiveness of the recommendations of a machine can out-tire the willpower of the reciprocating human, especially considering the ubiquitousness of the machine; be it on your tv, phone, car, videos, music, news, social media.
So I think it is possible to formulate the reaction you see as a defense against building machines that abuse organic human attention. Going back to free speech, there wouldn't be defending the right to free speech of the machine if it could denial-of-service the recipients' judgement with all the engagement tricks in the book and at scale.
And the spirit of the objection is not about the radicalness of the content; but rather the frequency of which such content as brought to the user by the machine. In a free marketplace of ideas of humans, such content would a) have lower incididence than youtube b) would have much different propagation/decay function than youtube c) you could give direct, high fidelity feedback to those humans. People's ability to filter out bullshit collectively in a real social setting is much stronger than our 1:1 encounters with youtube. There is no talking back to youtube, there is no friends you can together vet the ideas you encounter on youtube.