5 ms·

It only "is not a positive change", suddenly, now that websites friendly to the accepted dogma du jour are at risk of being exposed to it!

manfredo 6y ago
So let me get this straight: the total removal of moderation, and the subsequent inclusion of content which espouses ideas of racial superiority and extermination of peoples is a positive change? Calling for violence against groups of people is not illegal. Only specific, actionable incitement of violence [1]. I don't think you really understand what total absence of moderation really means. 1. https://en.m.wikipedia.org/wiki/Brandenburg_v._Ohio https://en.m.wikipedia.org/wiki/Brandenburg_v._Ohio
- pjkundert 6y ago
  I'm wondering how you jumped from "K-means Clustering", which implements much stronger and more fairly applied moderation than any current scheme, to "total removal of moderation".
  - manfredo 6y ago
    K-means clustering is insufficient to guarantee that no content that results in lawsuits or charges gets uploaded. These are sites with a billions of users. A failure rate of even 0.001% still means a lot of content that can result in charges or lawsuits gets through. Removal of section 230 means companies can even be criminally liable for things their users post. It's not a question of fairness at that point.
    - pjkundert 6y ago
      Of course not - since that is an impossible task. We can’t even build LEGO sites that make it impossible for kids to build and upload genitalia. Nor should we - if you don’t want your kids seeing LEGO dicks, help them downvote a bunch of users that upload those, and voila, they don’t see any posts by any users in that cluster. Conversely, your “beevis and butthead” kid sees nothing but dicks. Fill your boots... The only requirement is long-lives identities. Create a new identity anytime you want; it’s just that no veteran users will see anything you post until you establish a long track record of reasoned debate, and work yourself through clusters with logarithmic aging baskets, until you finally “graduate” to being seen by people of substance in your target audience. And, you can easily move yourself out again, by a few hot-headed off-the-cuff remarks. Welcome to the real world! Actions have consequences. The thing about being an unbiased conduit vs. a editorializing publisher is significant; if your clustering algorithms are standardized and published, and aren’t subject to editorial interference — you’re safe from lawsuits, just like the phone companies. It is unlikely that you will be able to escape from the requirement to give up metadata on users to the FBI — just like the phone company. Removal of section 230 for companies that have proven themselves unworthy by clearly political editing of users posting feeds deserve what they’re going to get, when they lose these protections, in my opinion. If they want to run an opinionated newspaper, then they’ve chosen the heavy lifting required to ensure no rando publishes an illegal article. Tough beans.
      - manfredo 6y ago
        > The only requirement is long-lives identities. Create a new identity anytime you want; it’s just that no veteran users will see anything you post until you establish a long track record of reasoned debate, and work yourself through clusters with logarithmic aging baskets, until you finally “graduate” to being seen by people of substance in your target audience. And, you can easily move yourself out again, by a few hot-headed off-the-cuff remarks. Welcome to the real world! Actions have consequences. And? None of this changes the fact that without Section 230 protections, companies are dead. There's simply no way to reliably prevent any and all illegal content from being posted. Sites like HN could not exist. There's no way Paul Graham, or whoever actually owns HN, is going to risk criminal charges to keep the forum running. > It is unlikely that you will be able to escape from the requirement to give up metadata on users to the FBI — just like the phone company. Giving metadata to law enforcement never had anything to do with Section 230. All section 230 says is, people hosting content are not liable for content created by other people. > Removal of section 230 for companies that have proven themselves unworthy by clearly political editing of users posting feeds deserve what they’re going to get, when they lose these protections, in my opinion. > If they want to run an opinionated newspaper, then they’ve chosen the heavy lifting required to ensure no rando publishes an illegal article. Tough beans. So it's not a blanket removal. It's a targeted removal of section 230 protections from sites that are perceived to have biased moderation. This is even worse. Now politicians can go to Facebook or YouTube and say, "hey, you better promote positive content about me and negative content about my opponents or I'm going to strip your Section 230 protections." Such a change to Section 230 likely won't lead to less biased sites, it'll lead to even more biased sites as they lick the boots of politicians to keep their protections. There's nothing stopping a government from claiming that k-clusters are biased. This mechanism of some sites getting liability protections and others not getting them is rife for abuse.

manfredo 6y ago

So let me get this straight: the total removal of moderation, and the subsequent inclusion of content which espouses ideas of racial superiority and extermination of peoples is a positive change? Calling for violence against groups of people is not illegal. Only specific, actionable incitement of violence [1]. I don't think you really understand what total absence of moderation really means. 1. https://en.m.wikipedia.org/wiki/Brandenburg_v._Ohio https://en.m.wikipedia.org/wiki/Brandenburg_v._Ohio

pjkundert 6y ago
I'm wondering how you jumped from "K-means Clustering", which implements much stronger and more fairly applied moderation than any current scheme, to "total removal of moderation".
- manfredo 6y ago
  K-means clustering is insufficient to guarantee that no content that results in lawsuits or charges gets uploaded. These are sites with a billions of users. A failure rate of even 0.001% still means a lot of content that can result in charges or lawsuits gets through. Removal of section 230 means companies can even be criminally liable for things their users post. It's not a question of fairness at that point.
  - pjkundert 6y ago
    Of course not - since that is an impossible task. We can’t even build LEGO sites that make it impossible for kids to build and upload genitalia. Nor should we - if you don’t want your kids seeing LEGO dicks, help them downvote a bunch of users that upload those, and voila, they don’t see any posts by any users in that cluster. Conversely, your “beevis and butthead” kid sees nothing but dicks. Fill your boots... The only requirement is long-lives identities. Create a new identity anytime you want; it’s just that no veteran users will see anything you post until you establish a long track record of reasoned debate, and work yourself through clusters with logarithmic aging baskets, until you finally “graduate” to being seen by people of substance in your target audience. And, you can easily move yourself out again, by a few hot-headed off-the-cuff remarks. Welcome to the real world! Actions have consequences. The thing about being an unbiased conduit vs. a editorializing publisher is significant; if your clustering algorithms are standardized and published, and aren’t subject to editorial interference — you’re safe from lawsuits, just like the phone companies. It is unlikely that you will be able to escape from the requirement to give up metadata on users to the FBI — just like the phone company. Removal of section 230 for companies that have proven themselves unworthy by clearly political editing of users posting feeds deserve what they’re going to get, when they lose these protections, in my opinion. If they want to run an opinionated newspaper, then they’ve chosen the heavy lifting required to ensure no rando publishes an illegal article. Tough beans.
    - manfredo 6y ago
      > The only requirement is long-lives identities. Create a new identity anytime you want; it’s just that no veteran users will see anything you post until you establish a long track record of reasoned debate, and work yourself through clusters with logarithmic aging baskets, until you finally “graduate” to being seen by people of substance in your target audience. And, you can easily move yourself out again, by a few hot-headed off-the-cuff remarks. Welcome to the real world! Actions have consequences. And? None of this changes the fact that without Section 230 protections, companies are dead. There's simply no way to reliably prevent any and all illegal content from being posted. Sites like HN could not exist. There's no way Paul Graham, or whoever actually owns HN, is going to risk criminal charges to keep the forum running. > It is unlikely that you will be able to escape from the requirement to give up metadata on users to the FBI — just like the phone company. Giving metadata to law enforcement never had anything to do with Section 230. All section 230 says is, people hosting content are not liable for content created by other people. > Removal of section 230 for companies that have proven themselves unworthy by clearly political editing of users posting feeds deserve what they’re going to get, when they lose these protections, in my opinion. > If they want to run an opinionated newspaper, then they’ve chosen the heavy lifting required to ensure no rando publishes an illegal article. Tough beans. So it's not a blanket removal. It's a targeted removal of section 230 protections from sites that are perceived to have biased moderation. This is even worse. Now politicians can go to Facebook or YouTube and say, "hey, you better promote positive content about me and negative content about my opponents or I'm going to strip your Section 230 protections." Such a change to Section 230 likely won't lead to less biased sites, it'll lead to even more biased sites as they lick the boots of politicians to keep their protections. There's nothing stopping a government from claiming that k-clusters are biased. This mechanism of some sites getting liability protections and others not getting them is rife for abuse.