4 ms·

Supreme Court admits they're not the best choice to decide internet future

8 points by zekrioca 4y ago

brucethemoose2 4y ago
Just going by the article, it really doesn't seem like the nature of what was going on is being articulated here. - The ISIS video was not recommended by random chance, it was recommended because the "algorithm" thought it was engaging. Its a deliberate profit driven choice that Google failed to clamp, not some innocent software bug. - Calling the Google's recommendation engine an "algorithm" is kinda like calling the Apollo program a research project. Its far from some nebulous side feature like they are describing it. - They are also trivializing YouTube's UI design which, again, is a finely tuned engagement machine. Maybe the justices are right about tech not being their area of expertise. But business ethics in relation to that "good faith" clause is precisely their thing... if the lawyers can boil things down to that.
- rektide 4y ago
  Accusing an algorithm of being bad because it doesnt use random chance seems like a bad joke to me. This is obviously unideal, and if Google was letting this happen regularly & doing nothing about it, I'd have plenty of sympathy for outrage & snark. But from what I can tell they take this quite seriously & spend enormous effort on these hard, difficult problems of trying to work out at scale what content is ok, and what is bad. It's in their own interest to remain a place advertisers can associate with, and that limus test seems fine; there's such few overall problems it's basically a miracle. But we have an overwhelming pent up anger & hair-trigger world where the very very very infrequent imperfections & real fuck-ups can be amplified inordinately, are lightning rods of attension. No one in risk management will ever promise you 100% no problems ever. No criticism is worth dealing with if we cant also talk about the millions of cases of bad material that have been mitigated. Having a one sided rage-fest against small specific star counter-examples is a low & unrealistic sensationalization not worth entertaining.
  - brucethemoose2 4y ago
    I'm fine with a refined personalization algorithm, but I am not fine but not an engagement optimizing one specifically. Engagement optimization is, by definition, addictive, and it tends to pick up outrage-generating extremist content because it is really engaging. Whats more, Google has every incentive to compensate as little as possible, as less engagement means less money. Its my belief that chasing engagement so fervently is at the heart not just the tech industry's woes, but many of the world's issues. Its warping user's minds, addicting them, and polarizing them, its destroying privacy as a side effect, and Google (and others) cant stop themselves because of competitive pressure.
    - rektide 4y ago
      In my view, there's not really a ton of other directly measurable things. We can ask users all kinds of questions to try and gauge what impact we're having, but that takes extreme work, doesn't scale, & has all manners of awful self-reporting bias. By contrast, using whether or not someone keeps watching & how much they upvote/interact does not seems like some horrible end of times disaster. Which is basically what engagement is. It's a directly observable metric, and it correlates pretty closely with whether or not what you are doing sucks for users or is something they seem to enjoy enough to keep doing. Are there bad aspects? Yeah likely. But to me, 99% this looks like trying to make a product people enjoy & like to use. Having such a strong opinion as you have, describing & attributing so many horrific negative effects - for making a product that has a measurably positive impact on basic user enjoyment - seems a bit wild to me. It's the tendency to view technology as magic, as some phenomenally super powerful ultra-capable advanced wizard-machine with incredibly fine grained & exact & perfect control. It's just not so. There's just not that much knowledge or sense inside the system to know what it's recommending or what it's doing, it's just blindly following macro-trends to try to make some course ass suggestion that maybe might help. There probably are some fairly gross/bad decisions that are bad for users being made, and there is definitely some shaping of intent, choice of what to push - a man behind the curtains of the algorithm, picking some biases, and some of those choices are probably closer to hell than heaven, but overall I just don't think there's that much real control or competency in how we pick stuff, in how these choices are made. I wish I could find it, but there was a wonderful critique of Zuboff's Age of Surveillance Capitalism that said, yes, the situation isn't great, but the book is also overselling the powers & potency these systems have. They're just not that smart, not so well guided as to be the high-functioning brainwashing machines the modern mania has made them out to be.. And I think it's selling enormous fear to worry so deeply that "It's warping user's minds, addicting them, and polarizaing them". Overall I think that describes such a vanishingly small problem, such a non-phenomenon in the global scale of things. Very very very few people fall down these dark paths. Countless more have perfectly fine experiences, experiencing no or only the most mild side-effects of the bad side. It races into paranoia to exacerbate the problem & magnify only the bad so heavily, leaves rationality behind & portrays a clutching desperation. It's the "the cities have become unlivable hells" narrative, but virtual: the online world has become an unlivable hell. Please, I feel like we've lost moderation, are spiralling out into wildness. Not everything is amazing, but it's just not so bad. Such a willfull blindness & such extremeism to ignore so very much of the picture & radicalize yourself around Fear Uncertainty and Doubt. and that position leaves us so weak, so susceptible to influence; there are real problems & real despair a plenty in the world, but youtube is, like, not really a huge factor in that, at all. It barely has a mark.

8 points by zekrioca 4y ago

Just going by the article, it really doesn't seem like the nature of what was going on is being articulated here. - The ISIS video was not recommended by random chance, it was recommended because the "algorithm" thought it was engaging. Its a deliberate profit driven choice that Google failed to clamp, not some innocent software bug. - Calling the Google's recommendation engine an "algorithm" is kinda like calling the Apollo program a research project. Its far from some nebulous side feature like they are describing it. - They are also trivializing YouTube's UI design which, again, is a finely tuned engagement machine. Maybe the justices are right about tech not being their area of expertise. But business ethics in relation to that "good faith" clause is precisely their thing... if the lawyers can boil things down to that.

rektide 4y ago
Accusing an algorithm of being bad because it doesnt use random chance seems like a bad joke to me. This is obviously unideal, and if Google was letting this happen regularly & doing nothing about it, I'd have plenty of sympathy for outrage & snark. But from what I can tell they take this quite seriously & spend enormous effort on these hard, difficult problems of trying to work out at scale what content is ok, and what is bad. It's in their own interest to remain a place advertisers can associate with, and that limus test seems fine; there's such few overall problems it's basically a miracle. But we have an overwhelming pent up anger & hair-trigger world where the very very very infrequent imperfections & real fuck-ups can be amplified inordinately, are lightning rods of attension. No one in risk management will ever promise you 100% no problems ever. No criticism is worth dealing with if we cant also talk about the millions of cases of bad material that have been mitigated. Having a one sided rage-fest against small specific star counter-examples is a low & unrealistic sensationalization not worth entertaining.
- brucethemoose2 4y ago
  I'm fine with a refined personalization algorithm, but I am not fine but not an engagement optimizing one specifically. Engagement optimization is, by definition, addictive, and it tends to pick up outrage-generating extremist content because it is really engaging. Whats more, Google has every incentive to compensate as little as possible, as less engagement means less money. Its my belief that chasing engagement so fervently is at the heart not just the tech industry's woes, but many of the world's issues. Its warping user's minds, addicting them, and polarizing them, its destroying privacy as a side effect, and Google (and others) cant stop themselves because of competitive pressure.
  - rektide 4y ago
    In my view, there's not really a ton of other directly measurable things. We can ask users all kinds of questions to try and gauge what impact we're having, but that takes extreme work, doesn't scale, & has all manners of awful self-reporting bias. By contrast, using whether or not someone keeps watching & how much they upvote/interact does not seems like some horrible end of times disaster. Which is basically what engagement is. It's a directly observable metric, and it correlates pretty closely with whether or not what you are doing sucks for users or is something they seem to enjoy enough to keep doing. Are there bad aspects? Yeah likely. But to me, 99% this looks like trying to make a product people enjoy & like to use. Having such a strong opinion as you have, describing & attributing so many horrific negative effects - for making a product that has a measurably positive impact on basic user enjoyment - seems a bit wild to me. It's the tendency to view technology as magic, as some phenomenally super powerful ultra-capable advanced wizard-machine with incredibly fine grained & exact & perfect control. It's just not so. There's just not that much knowledge or sense inside the system to know what it's recommending or what it's doing, it's just blindly following macro-trends to try to make some course ass suggestion that maybe might help. There probably are some fairly gross/bad decisions that are bad for users being made, and there is definitely some shaping of intent, choice of what to push - a man behind the curtains of the algorithm, picking some biases, and some of those choices are probably closer to hell than heaven, but overall I just don't think there's that much real control or competency in how we pick stuff, in how these choices are made. I wish I could find it, but there was a wonderful critique of Zuboff's Age of Surveillance Capitalism that said, yes, the situation isn't great, but the book is also overselling the powers & potency these systems have. They're just not that smart, not so well guided as to be the high-functioning brainwashing machines the modern mania has made them out to be.. And I think it's selling enormous fear to worry so deeply that "It's warping user's minds, addicting them, and polarizaing them". Overall I think that describes such a vanishingly small problem, such a non-phenomenon in the global scale of things. Very very very few people fall down these dark paths. Countless more have perfectly fine experiences, experiencing no or only the most mild side-effects of the bad side. It races into paranoia to exacerbate the problem & magnify only the bad so heavily, leaves rationality behind & portrays a clutching desperation. It's the "the cities have become unlivable hells" narrative, but virtual: the online world has become an unlivable hell. Please, I feel like we've lost moderation, are spiralling out into wildness. Not everything is amazing, but it's just not so bad. Such a willfull blindness & such extremeism to ignore so very much of the picture & radicalize yourself around Fear Uncertainty and Doubt. and that position leaves us so weak, so susceptible to influence; there are real problems & real despair a plenty in the world, but youtube is, like, not really a huge factor in that, at all. It barely has a mark.