6 ms·

> YouTube's algorithm recommended extremist/conspiracy content Would you agree that the algorithm recommended this content because people believe in, or at lea

by jrcii 9y ago

> YouTube's algorithm recommended extremist/conspiracy content Would you agree that the algorithm recommended this content because people believe in, or at least are interested in, conspiracies? Even if we feel that's a mistake, who are we (or Google) to be the arbiters of truth?

gareim 9y ago
A friend asked me to watch a clip of Ben Shapiro so we could sincerely talk about what the other side believes. I forgot that I was on a fresh OS install, not yet logged in. Well, right after that one video, every single recommended video to me became a "Ben Shapiro SLAMS liberal...". I get that YouTube had no other metric to gauge me by, by there are 10 or 12 recommendations and there's no law that says YouTube can't ease into recommendations. Whether you're for or against Shapiro, this kind of behavior induces people into bubbles and radicalizing in the feedback loop.
- m00g00 9y ago
  I mean, that style of title exists accoross the political spectrum on YouTube and it's beyond stupid. Whenever I see "X DESTROYS Y" titles I assume the video will be a waste of 3-10 minutes with no particular moment of linguistic destruction on either side and is in fact just two people bickering. Just clickbait titles is all it is. Regardless, I don't see how any of that can really count as "extremist" or "conspiratorial" content.
- mediumdeviation 9y ago
  Actually, I noticed YouTube's recommendations seems _heavily_ biased towards what you watched recently. This is likely due to how people watch videos - much like people binge watching TV serials, they are very much likely to continue watching videos from the same or similar channels. In most cases this is harmless - play a song, and YouTube will automatically create a playlist of songs in the same genre. Watch a cooking video, and it'll give you a dozen other recipes by the same chef. Unfortunately this doesn't work for political videos, as you've noticed. But there is absolutely no incentive for YouTube to insert videos from opposing viewpoints into the recommendations. YouTube is just giving people what they want.
  - saalweachter 9y ago
    YouTube is sort of caught between a rock and a hard place here. So you watch a video about how the history of vaccines and how many lives have been saved by vaccination. Should YouTube recommend a counterpoint video about how vaccines are going to kill you and everyone you love? If you are recommending "other sides of issues", either you start showing a lot of really crazy things to people who wouldn't otherwise see it, or you need to start taking an editorial position that some issues are "settled" and one side is just "wrong", and do a one-way gating where you'll show a breath of sanity in the follow-up recommendations to "bad" videos but not show "bad" recommendations on videos that get the subject "right". Which one would you prefer?
    - dragonwriter 9y ago
      Well, I'd prefer annotated recommendations with reasons, and analysis of quality as a big feeder into the recommendations (so, both opposing and supporting--and even supporting in part and opposing in part--viewpoints, especially with stronger support, would be natural recommendations, with their relationships to the thing you just viewed annotated.) OTOH, that's probably a lot harder to do (especially automatically; intuitively, it seems likely to be essentially an AI-complete problem, though I wouldn't be too surprised if someone found a way to do a "good enough" version for common classes of content without a full AI) than what typical recommendation engines do; the annotated related items thing is found some places, but they are IME always manually-curated, narrow-focus databases (e.g., commercial legal authority databases.)
    - MrLeap 9y ago
      It's funny how the problem is so _us_ this time. Turn a mirror on our souls and we lose nuance, mediums, neutrals, and common ground. "Ben Shapiro makes a good point" becomes "BEN SHAPIRO DESTROYS SUBHUMAN SCUM." Who is that message for? It's not a discussion or debate, it's linguistic flashing lights to simulate a community around a simulated conflict. Youtube red, solving alienation in a more profitable way than the last red revolution. The crime of social media is that they've created public solitary confinement. The internet is now a place where you go to be alone with your thoughts as simulated by companies. So boring.
thisacctforreal 9y ago
The crux seems to be that Google's algorithms are designed to maximize time spent watching. It's frankly absurd how successful they have been at this, especially in younger generations. It's one thing when their team of PhD's extract attention into cat videos, and another thing to extract it into potentially unhealthy amounts of extremist material and conspiracies. I don't know what the answer is, but I do miss the days before infinite scroll and autoplay.
- crmd 9y ago
  One of my friends is the music director of a US freeform radio station. He’s a legend in the industry known for his encyclopedic knowledge of cool but highly obscure artists over the past 40 years. Over the past year, he has often been left mouth open when YouTube’s algorithm starts picking what’s next after we pick a starting point. “How the hell does it know about these guys?” That’s how good the robots have become.
- jrcii 9y ago
  It occurs to me that the widespread interest in conspiracies may stem from a general distrust of information from authorities, which in turn is the result of perceiving the dishonesty upon which all of politics is built, for example.
dkarl 9y ago
Google recommends that stuff to me, and I don't believe in it or watch it. Watch math videos, get flat earth recommendations. Watch a few videos about the migration of Germanic tribes in Europe during the decline of the Roman Empire, get white supremacist recommendations. My best guess? They want you to sit and watch YouTube for hours, so they recommend stuff watched by people who sit and watch YouTube for hours. This stuff reminds of the word "excitotoxins," which is based on a silly idea yet seems to capture the addictive effect of stimulation. People are stimulated by things that seem novel, controversial, and dangerous. People craving stimulation will prefer provocative junk over unsurprising truth.
- smoyer 9y ago
  I binge watch youtube in a small window while doing certain kinds of work and specifically don't want anything I'm too interested in. This leads to future "weird" recommendations. I also have to say that there's a certain comic relief to videos by flat-earthers (no moon landing, ancient astronaut) and other conspiracy theorists. Some mix valid points with hyperbole but some are so far out there that they're comical.
- emilfihlman 9y ago
  >Google recommends that stuff to me, and I don't believe in it or watch it. So? Obviously people are watching them. Or are you saying that YouTube has a strong hidden conservative bias?
- SerLava 9y ago
  >Watch a few videos about the migration of Germanic tribes in Europe during the decline of the Roman Empire, get white supremacist recommendations. Ugh, that happened to me. I was halfway through a video making weird, interesting, yet really benign claims when suddenly they connected all the pieces in support of some bizarre white supremacist shit. Just because I wanted to know about proto-indo-europeans. I looked over to the recommended videos, and it was already like 20% armchair white supremacist videos.
JoshCole 9y ago
No. Content recommendation algorithms reward engagement metrics. One of the metrics they reward is getting a user's attention, briefly. In the real world, someone can get my attention by screaming that there is a fire. Belief that there is a fire and interest in fire are not necessary for my attention to be grabbed by a warning of fire. All that is needed is a desire for self-preservation and a degree of trust in the source of the knowledge. Compounding the problem, since engagement is improved and people make money off of videos, there is an incentive in place encouraging the proliferation of attention grabbing false information. In a better world, this behavior would not be incentivized. In a better world, reputation metrics would allow a person to realize that the boy who cried wolf was the one who had posted the attention grabbing video. Humanity has known for a long time that there are consequences for repeated lying. We have fables about that, warning liars away from lying. I don't think making that explicit, like it is in many real world cases of lying publicly in the most attention attracting way possible, would be unreasonable.
- dcow 9y ago
  No one hears the lies once they're removed. That can be good, but it also might work to prevent more people from criticizing or exposing them and thus, even if only slightly, validate them in the eyes of liars. This might yield nastier lies in the future.

by jrcii 9y ago

gareim 9y ago

A friend asked me to watch a clip of Ben Shapiro so we could sincerely talk about what the other side believes. I forgot that I was on a fresh OS install, not yet logged in. Well, right after that one video, every single recommended video to me became a "Ben Shapiro SLAMS liberal...". I get that YouTube had no other metric to gauge me by, by there are 10 or 12 recommendations and there's no law that says YouTube can't ease into recommendations. Whether you're for or against Shapiro, this kind of behavior induces people into bubbles and radicalizing in the feedback loop.

m00g00 9y ago
I mean, that style of title exists accoross the political spectrum on YouTube and it's beyond stupid. Whenever I see "X DESTROYS Y" titles I assume the video will be a waste of 3-10 minutes with no particular moment of linguistic destruction on either side and is in fact just two people bickering. Just clickbait titles is all it is. Regardless, I don't see how any of that can really count as "extremist" or "conspiratorial" content.
mediumdeviation 9y ago
Actually, I noticed YouTube's recommendations seems _heavily_ biased towards what you watched recently. This is likely due to how people watch videos - much like people binge watching TV serials, they are very much likely to continue watching videos from the same or similar channels. In most cases this is harmless - play a song, and YouTube will automatically create a playlist of songs in the same genre. Watch a cooking video, and it'll give you a dozen other recipes by the same chef. Unfortunately this doesn't work for political videos, as you've noticed. But there is absolutely no incentive for YouTube to insert videos from opposing viewpoints into the recommendations. YouTube is just giving people what they want.
- saalweachter 9y ago
  YouTube is sort of caught between a rock and a hard place here. So you watch a video about how the history of vaccines and how many lives have been saved by vaccination. Should YouTube recommend a counterpoint video about how vaccines are going to kill you and everyone you love? If you are recommending "other sides of issues", either you start showing a lot of really crazy things to people who wouldn't otherwise see it, or you need to start taking an editorial position that some issues are "settled" and one side is just "wrong", and do a one-way gating where you'll show a breath of sanity in the follow-up recommendations to "bad" videos but not show "bad" recommendations on videos that get the subject "right". Which one would you prefer?
  - dragonwriter 9y ago
    Well, I'd prefer annotated recommendations with reasons, and analysis of quality as a big feeder into the recommendations (so, both opposing and supporting--and even supporting in part and opposing in part--viewpoints, especially with stronger support, would be natural recommendations, with their relationships to the thing you just viewed annotated.) OTOH, that's probably a lot harder to do (especially automatically; intuitively, it seems likely to be essentially an AI-complete problem, though I wouldn't be too surprised if someone found a way to do a "good enough" version for common classes of content without a full AI) than what typical recommendation engines do; the annotated related items thing is found some places, but they are IME always manually-curated, narrow-focus databases (e.g., commercial legal authority databases.)
  - MrLeap 9y ago
    It's funny how the problem is so _us_ this time. Turn a mirror on our souls and we lose nuance, mediums, neutrals, and common ground. "Ben Shapiro makes a good point" becomes "BEN SHAPIRO DESTROYS SUBHUMAN SCUM." Who is that message for? It's not a discussion or debate, it's linguistic flashing lights to simulate a community around a simulated conflict. Youtube red, solving alienation in a more profitable way than the last red revolution. The crime of social media is that they've created public solitary confinement. The internet is now a place where you go to be alone with your thoughts as simulated by companies. So boring.

thisacctforreal 9y ago

The crux seems to be that Google's algorithms are designed to maximize time spent watching. It's frankly absurd how successful they have been at this, especially in younger generations. It's one thing when their team of PhD's extract attention into cat videos, and another thing to extract it into potentially unhealthy amounts of extremist material and conspiracies. I don't know what the answer is, but I do miss the days before infinite scroll and autoplay.

crmd 9y ago
One of my friends is the music director of a US freeform radio station. He’s a legend in the industry known for his encyclopedic knowledge of cool but highly obscure artists over the past 40 years. Over the past year, he has often been left mouth open when YouTube’s algorithm starts picking what’s next after we pick a starting point. “How the hell does it know about these guys?” That’s how good the robots have become.
jrcii 9y ago
It occurs to me that the widespread interest in conspiracies may stem from a general distrust of information from authorities, which in turn is the result of perceiving the dishonesty upon which all of politics is built, for example.

dkarl 9y ago

Google recommends that stuff to me, and I don't believe in it or watch it. Watch math videos, get flat earth recommendations. Watch a few videos about the migration of Germanic tribes in Europe during the decline of the Roman Empire, get white supremacist recommendations. My best guess? They want you to sit and watch YouTube for hours, so they recommend stuff watched by people who sit and watch YouTube for hours. This stuff reminds of the word "excitotoxins," which is based on a silly idea yet seems to capture the addictive effect of stimulation. People are stimulated by things that seem novel, controversial, and dangerous. People craving stimulation will prefer provocative junk over unsurprising truth.

smoyer 9y ago
I binge watch youtube in a small window while doing certain kinds of work and specifically don't want anything I'm too interested in. This leads to future "weird" recommendations. I also have to say that there's a certain comic relief to videos by flat-earthers (no moon landing, ancient astronaut) and other conspiracy theorists. Some mix valid points with hyperbole but some are so far out there that they're comical.
emilfihlman 9y ago
>Google recommends that stuff to me, and I don't believe in it or watch it. So? Obviously people are watching them. Or are you saying that YouTube has a strong hidden conservative bias?
SerLava 9y ago
>Watch a few videos about the migration of Germanic tribes in Europe during the decline of the Roman Empire, get white supremacist recommendations. Ugh, that happened to me. I was halfway through a video making weird, interesting, yet really benign claims when suddenly they connected all the pieces in support of some bizarre white supremacist shit. Just because I wanted to know about proto-indo-europeans. I looked over to the recommended videos, and it was already like 20% armchair white supremacist videos.

JoshCole 9y ago

No. Content recommendation algorithms reward engagement metrics. One of the metrics they reward is getting a user's attention, briefly. In the real world, someone can get my attention by screaming that there is a fire. Belief that there is a fire and interest in fire are not necessary for my attention to be grabbed by a warning of fire. All that is needed is a desire for self-preservation and a degree of trust in the source of the knowledge. Compounding the problem, since engagement is improved and people make money off of videos, there is an incentive in place encouraging the proliferation of attention grabbing false information. In a better world, this behavior would not be incentivized. In a better world, reputation metrics would allow a person to realize that the boy who cried wolf was the one who had posted the attention grabbing video. Humanity has known for a long time that there are consequences for repeated lying. We have fables about that, warning liars away from lying. I don't think making that explicit, like it is in many real world cases of lying publicly in the most attention attracting way possible, would be unreasonable.

dcow 9y ago
No one hears the lies once they're removed. That can be good, but it also might work to prevent more people from criticizing or exposing them and thus, even if only slightly, validate them in the eyes of liars. This might yield nastier lies in the future.