3 ms·

There is a legitimately important debate in one specific area of section 230: the treatment of algorithms. Section 230 is predicated on the idea that sites lik

There is a legitimately important debate in one specific area of section 230: the treatment of algorithms. Section 230 is predicated on the idea that sites like HN have no editorial function and therefore have no liability over posted content. This safe harbor is extremely important and needs to be protected. Unfortunately algorithmic recommendation makes that line very grey. As a compromise I'd push for algorithmic transparency: show me how you're ranking content. This would both clarify whether companies are engaging in deliberate editorial activities via algorithm (and therefore should be held liable for the content they promote) while also improving both user privacy (we'd be able to see how our data is being used) and online trust.

supriyo-biswas 4y ago
HN has multiple algorithms to prevent sink controversial topics and the like, and the removal of "How", "X ways" etc. could also be seen as editorialization. I honestly don't see a way to balance the two interests, unless a voluntary opt-out to the algorithm (such as HN's /new) also suffices to comply with the new law.
- BaseballPhysics 4y ago
  You missed my point: show me the algorithm. You just described how you think HN works but unless you work there you can't have first hand knowledge. I know I don't. So show me the algorithm. Let's have an open conversation based on facts instead of speculation. As an aside, I also think the lines are easier drawn than you think. For example: is the algorithm based purely on overt signals from the userbase (votes, flags, etc)? Or is it based on an analysis of the content itself, which is then used to fuel a recommendation engine? If the latter, are the recommendations based purely on analyzing past expressed preference? Or is there some other baked in weighting? Algorithms aren't dark mysterious things. They're entirely explainable, and if opened up you could have a real discussion about what is and isn't editorial promotion of content.
  - judahmeek 4y ago
    A lot of content moderation / censorship isn't by algorithm. Recommendations, sure, but not the removal of content.
    - BaseballPhysics 4y ago
      Then it's a good thing I wasn't addressing moderation in general but only algorithmic curation at scale. What about my comments suggested otherwise?
basch 4y ago
Section 230 says you CAN have editorial control over content and STILL not have liability. That selectively choosing what content appears is NOT your speech. It’s maybe a somewhat questionable reasoning, but it led to an internet where spam, hostility, nonsense etc can be filtered without worry or risk of consequence. There might be room for improvement into transparency of what is and isn’t allowed. But as of right now, platforms are allowed to be arbitrary. Publishing algorithms would make it easy to defeat spam filters and anti evil filters. They are black boxes for a reason.
- trieste92 4y ago
  > but it led to an internet where spam, hostility, nonsense etc can be filtered without worry or risk of consequence. No, it led to an internet where the listed things proliferate
  - basch 4y ago
    Not everywhere
- BaseballPhysics 4y ago
  > But as of right now, platforms are allowed to be arbitrary. And IMO they shouldn't be. > Publishing algorithms would make it easy to defeat spam filters and anti evil filters. They are black boxes for a reason. Security through obscurity. Besides, if I have to choose between slightly poorer spam detection and algorithmic transparency on the sites that are setting the public discourse, I'm gonna pick transparency. Wouldn't you?
  - basch 4y ago
    Should a Christian website be allowed to foster a Christian community, without disruptive trolls? Should a Republican website be allowed to create a Republican eco chamber? Should Reddit be liable for the moderator of a food community removing a post about android tech support? Should sites where dang censors whatever he wants for the sake of the community values suddenly become liable for the speech that isn’t removed? No. It’s my speech not his. Let’s be real, the threat of benevolent censorship is what keeps this place civil and upstanding. You might say, but but transparency. Now you are creating a new problem. For example, the Christian website above. If I go there extolling the virtues of Satan, you are forcing them to host copies of speech they find deplorable, in their public censorship reports. And if you obscure or roll up censorship reports, now you need independent speech removal auditors. I would choose a better spam filter and a free market for you to make a competing service with transparency. The free market breaks down when institutions become proprietary and monolithic. A better angle is the government to better force protocols, decentralization, federation etc. Even email is kinda controlled by two players, but at least if I got kicked off one I can still email all my old friends from my new account on the other. Data portability, service interoperability, and privacy better address a lack of transparency than forcing homogenous censorship rules on all sites.

supriyo-biswas 4y ago

HN has multiple algorithms to prevent sink controversial topics and the like, and the removal of "How", "X ways" etc. could also be seen as editorialization. I honestly don't see a way to balance the two interests, unless a voluntary opt-out to the algorithm (such as HN's /new) also suffices to comply with the new law.

BaseballPhysics 4y ago
You missed my point: show me the algorithm. You just described how you think HN works but unless you work there you can't have first hand knowledge. I know I don't. So show me the algorithm. Let's have an open conversation based on facts instead of speculation. As an aside, I also think the lines are easier drawn than you think. For example: is the algorithm based purely on overt signals from the userbase (votes, flags, etc)? Or is it based on an analysis of the content itself, which is then used to fuel a recommendation engine? If the latter, are the recommendations based purely on analyzing past expressed preference? Or is there some other baked in weighting? Algorithms aren't dark mysterious things. They're entirely explainable, and if opened up you could have a real discussion about what is and isn't editorial promotion of content.
- judahmeek 4y ago
  A lot of content moderation / censorship isn't by algorithm. Recommendations, sure, but not the removal of content.
  - BaseballPhysics 4y ago
    Then it's a good thing I wasn't addressing moderation in general but only algorithmic curation at scale. What about my comments suggested otherwise?

basch 4y ago

Section 230 says you CAN have editorial control over content and STILL not have liability. That selectively choosing what content appears is NOT your speech. It’s maybe a somewhat questionable reasoning, but it led to an internet where spam, hostility, nonsense etc can be filtered without worry or risk of consequence. There might be room for improvement into transparency of what is and isn’t allowed. But as of right now, platforms are allowed to be arbitrary. Publishing algorithms would make it easy to defeat spam filters and anti evil filters. They are black boxes for a reason.

trieste92 4y ago
> but it led to an internet where spam, hostility, nonsense etc can be filtered without worry or risk of consequence. No, it led to an internet where the listed things proliferate
- basch 4y ago
  Not everywhere
BaseballPhysics 4y ago
> But as of right now, platforms are allowed to be arbitrary. And IMO they shouldn't be. > Publishing algorithms would make it easy to defeat spam filters and anti evil filters. They are black boxes for a reason. Security through obscurity. Besides, if I have to choose between slightly poorer spam detection and algorithmic transparency on the sites that are setting the public discourse, I'm gonna pick transparency. Wouldn't you?
- basch 4y ago
  Should a Christian website be allowed to foster a Christian community, without disruptive trolls? Should a Republican website be allowed to create a Republican eco chamber? Should Reddit be liable for the moderator of a food community removing a post about android tech support? Should sites where dang censors whatever he wants for the sake of the community values suddenly become liable for the speech that isn’t removed? No. It’s my speech not his. Let’s be real, the threat of benevolent censorship is what keeps this place civil and upstanding. You might say, but but transparency. Now you are creating a new problem. For example, the Christian website above. If I go there extolling the virtues of Satan, you are forcing them to host copies of speech they find deplorable, in their public censorship reports. And if you obscure or roll up censorship reports, now you need independent speech removal auditors. I would choose a better spam filter and a free market for you to make a competing service with transparency. The free market breaks down when institutions become proprietary and monolithic. A better angle is the government to better force protocols, decentralization, federation etc. Even email is kinda controlled by two players, but at least if I got kicked off one I can still email all my old friends from my new account on the other. Data portability, service interoperability, and privacy better address a lack of transparency than forcing homogenous censorship rules on all sites.