5 ms·

> I would rather take the stance of feed algorithms and moderation logs be PUBLICLY available. Transparency instead of censorship. I think that removing CDA Se

> I would rather take the stance of feed algorithms and moderation logs be PUBLICLY available. Transparency instead of censorship. I think that removing CDA Sec 230 protections for algorithmically-curated feeds is the answer. It's one thing to have a basic FIFO feed, it's another to hide or reveal content based on your own internal engagement/revenue targets. At the point where you're crafting bespoke engagement-maximizing feeds, you're creating a gestalt creative work that has a life of its own and that is no longer merely a passthrough for the works of users of the platform. In other words, consider Facebook's curated feeds as Facebook's speech, not only the speech of their users, so Facebook faces liability for that speech.

quotemstr 5y ago
Your proposal would make HN-style comment ranking illegal. Do you really want to live in the world you propose?
- mullingitover 5y ago
  I considered the widely-used 'hot' algorithm and I don't think that really qualifies as a 'creative work' on the part of the site. The 'top from last hour/day/month/year' and 'hot' algorithms are really elementary sorting methods that are in the same league as FIFO.
  - quotemstr 5y ago
    Who gets to decide what algorithms are simple enough to be legal? You? Why? All a ranking algorithm is doing is making suggestions that certain content be read before other content. That's protected speech.
    - mullingitover 5y ago
      > Who gets to decide what algorithms are simple enough to be legal? You? Why? I'm not talking about the legality, I'm talking about the liability. If one user posts defamatory/libelous speech and Facebook decides to blast it to 1 billion users because their secret sauce algorithm found that it really cranks the revenue dial, Facebook should share some culpability in that defamation lawsuit. A basic 'hot' algorithm that's not designed by squadrons of PhDs running machine learning on user behavior analysis isn't really the same thing, culpability-wise, imho.
      - quotemstr 5y ago
        > basic 'hot' algorithm that's not designed by squadrons of PhDs running machine learning on user behavior analysis isn't really the same thing, culpability-wise, imho A "hot" algorithm can blast high engagement libel in front of people as well as a fancy algorithm. If the "hot" algorithm and the fancier algorithm are both content neutral, on what basis can you distinguish the two as a matter of law? Does the hot algorithm become illegal if a PhD implements it? I'm at a loss about what distinction you're actually trying to draw. Your post, like many others on this thread, isn't articulating exactly what about FB's conduct you find objectionable?
        mullingitover 5y ago
        You keep putting words in my mouth, I'm not saying anything about outlawing anything, just assigning liability. > Your post, like many others on this thread, isn't articulating exactly what about FB's conduct you find objectionable? The objectionable conduct is pretty thoroughly covered in the article, but to review: > "Facebook over and over again has shown it chooses profit over safety"
        quotemstr 5y ago
        Illegal, liable, doesn't matter --- you want to use the state to drive certain kinds of ranking off the internet. Fine. What, precisely, is the line between algorithms acceptable to you and algorithms not? What is the precise conduct that would make liability attach to a ranking algorithm? You can emote, but you can't describe what exactly it is you would turn into a law.
        mullingitover 5y ago
        > You can emote, but you can't describe what exactly it is you would turn into a law. I thought I made it pretty clear from the outset I was talking about removing CDA Sec 230 protections for sites using bespoke (i.e. proprietary) curation algorithms for their feeds.
- specialist 5y ago
  HN's rankings are not personalized, so no filter bubbles. HN is fully moderated, FB is barely moderated. HN nerfs their algos. FB weaponizes theirs. HN applies their rules fairly, evenly. FB does not. FB has ads. HN does not. FB weaponizes PII. HN does not. Etc. Comparing HN to FB is like bicycle vs freight train, firecracker vs nukes, nice bath vs a hurricane.
  - quotemstr 5y ago
    No, actually, it's not any of these things. It's on HN that dang applies special rules to your comments if he doesn't like you. FB's algorithm is impersonal. Your list of criteria is legally inactionable. Neither you nor anyone else can write down what precisely it is that FB should be prohibited from doing. All I see is a bunch of unproductive rhetoric about how this is bad or that is bad. Blah blah, filter bubbles, moderation, whatever. What is the exact definition of the criteria you would use to prohibit ranking algorithms?
    - specialist 5y ago
      Regulating Facebook, making it safe for human consumption, is trivial. No targeted advertising. No algorithmic news feed. Voila. Fixed. RSS feed readers can do it. So can Facebook. > dang applies special rules to your comments if he doesn't like you. What's FB's equivalent to HN's guidelines? When has @dang violated his own guidelines? Are you suggesting that automated curation, moderating is without bias? > FB's algorithm is impersonal. FB's algorithms are hyper personal. Machine mediated curation is sociopathic, utterly lacking empathy, not impersonal. Repeating myself: every FB user gets their own news feed, every HN user sees the same thing.
      - quotemstr 5y ago
        > No targeted advertising. Strong disagree. Why the hell shouldn't I be able show ads to people who want to see them? > No algorithmic news feed. HN is an algorithmic feed.
        5y ago
raxxorrax 5y ago
I agree. Especially since political ambition won't be to curb misinformation, it will be to get control about which misinformation will be spread. Harm to users will be an excuse and like misinformation it will be quite difficult to quantify.

quotemstr 5y ago

Your proposal would make HN-style comment ranking illegal. Do you really want to live in the world you propose?

mullingitover 5y ago
I considered the widely-used 'hot' algorithm and I don't think that really qualifies as a 'creative work' on the part of the site. The 'top from last hour/day/month/year' and 'hot' algorithms are really elementary sorting methods that are in the same league as FIFO.
- quotemstr 5y ago
  Who gets to decide what algorithms are simple enough to be legal? You? Why? All a ranking algorithm is doing is making suggestions that certain content be read before other content. That's protected speech.
  - mullingitover 5y ago
    > Who gets to decide what algorithms are simple enough to be legal? You? Why? I'm not talking about the legality, I'm talking about the liability. If one user posts defamatory/libelous speech and Facebook decides to blast it to 1 billion users because their secret sauce algorithm found that it really cranks the revenue dial, Facebook should share some culpability in that defamation lawsuit. A basic 'hot' algorithm that's not designed by squadrons of PhDs running machine learning on user behavior analysis isn't really the same thing, culpability-wise, imho.
    - quotemstr 5y ago
      > basic 'hot' algorithm that's not designed by squadrons of PhDs running machine learning on user behavior analysis isn't really the same thing, culpability-wise, imho A "hot" algorithm can blast high engagement libel in front of people as well as a fancy algorithm. If the "hot" algorithm and the fancier algorithm are both content neutral, on what basis can you distinguish the two as a matter of law? Does the hot algorithm become illegal if a PhD implements it? I'm at a loss about what distinction you're actually trying to draw. Your post, like many others on this thread, isn't articulating exactly what about FB's conduct you find objectionable?
      - mullingitover 5y ago
        You keep putting words in my mouth, I'm not saying anything about outlawing anything, just assigning liability. > Your post, like many others on this thread, isn't articulating exactly what about FB's conduct you find objectionable? The objectionable conduct is pretty thoroughly covered in the article, but to review: > "Facebook over and over again has shown it chooses profit over safety"
        quotemstr 5y ago
        Illegal, liable, doesn't matter --- you want to use the state to drive certain kinds of ranking off the internet. Fine. What, precisely, is the line between algorithms acceptable to you and algorithms not? What is the precise conduct that would make liability attach to a ranking algorithm? You can emote, but you can't describe what exactly it is you would turn into a law.
        mullingitover 5y ago
        > You can emote, but you can't describe what exactly it is you would turn into a law. I thought I made it pretty clear from the outset I was talking about removing CDA Sec 230 protections for sites using bespoke (i.e. proprietary) curation algorithms for their feeds.
specialist 5y ago
HN's rankings are not personalized, so no filter bubbles. HN is fully moderated, FB is barely moderated. HN nerfs their algos. FB weaponizes theirs. HN applies their rules fairly, evenly. FB does not. FB has ads. HN does not. FB weaponizes PII. HN does not. Etc. Comparing HN to FB is like bicycle vs freight train, firecracker vs nukes, nice bath vs a hurricane.
- quotemstr 5y ago
  No, actually, it's not any of these things. It's on HN that dang applies special rules to your comments if he doesn't like you. FB's algorithm is impersonal. Your list of criteria is legally inactionable. Neither you nor anyone else can write down what precisely it is that FB should be prohibited from doing. All I see is a bunch of unproductive rhetoric about how this is bad or that is bad. Blah blah, filter bubbles, moderation, whatever. What is the exact definition of the criteria you would use to prohibit ranking algorithms?
  - specialist 5y ago
    Regulating Facebook, making it safe for human consumption, is trivial. No targeted advertising. No algorithmic news feed. Voila. Fixed. RSS feed readers can do it. So can Facebook. > dang applies special rules to your comments if he doesn't like you. What's FB's equivalent to HN's guidelines? When has @dang violated his own guidelines? Are you suggesting that automated curation, moderating is without bias? > FB's algorithm is impersonal. FB's algorithms are hyper personal. Machine mediated curation is sociopathic, utterly lacking empathy, not impersonal. Repeating myself: every FB user gets their own news feed, every HN user sees the same thing.
    - quotemstr 5y ago
      > No targeted advertising. Strong disagree. Why the hell shouldn't I be able show ads to people who want to see them? > No algorithmic news feed. HN is an algorithmic feed.
      - 5y ago

raxxorrax 5y ago

I agree. Especially since political ambition won't be to curb misinformation, it will be to get control about which misinformation will be spread. Harm to users will be an excuse and like misinformation it will be quite difficult to quantify.