4 ms·

We are censored every day. A newspaper editor censors their writers, your boss censors your reports, your friends censor your bar room bullshit by calling you

We are censored every day. A newspaper editor censors their writers, your boss censors your reports, your friends censor your bar room bullshit by calling you out on it, your spouse tells you you will not argue with your brother in law again. But this is all implicit censoring - adhering to social norms through social signals (a raised eye brow, being told to buy your round and shut up) The thing is computers do not do any implicit censoring - and so we are expecting FAANG to step up. But we need firstly a set of agreed rules, then the ability to automatically understand the rules before we even get to due process. The issue is firstly one of scale - HN kinda sorta works because it had clear rules from the start, abs had a lot of moderation, from early users and mods. Multiply by 1000x the volume and suddenly dang's keyboard explodes. Facebook is trying to produce moderation at scale with thousands of human moderators, but they probably will never "succeed" - because success is not defined well enough. Let me restate that. There are thousands of groups on FAANG - each one has its own evolving rules on what is and is not acceptable. But facebook can only really deal with one set of rules. It will try and manage the lowest common denominator - no child porn, no murder, but it slowly heads towards "define free speech" The main indicator here is that both sides of every political debate think FAANG is censoring them.

lifeisstillgood 6y ago
Just to expand a little - If you are the moderator and both sides think you are biased, you are probably in the right zone. The problem is do we as society, really want a handful of private companies deciding what the rules are - deciding for example if a Philipino politician can use FAANG to talk to his electorate (because the electorate is all on facebook, say de facto facebook are garejeeping this) Society does not particularly want this, Zuckerberg definitely does not want this. So we need to find a way to have regulations that deal with a set of poorly defined globally circumstances (Philipino election law as well as German hate speech laws), some areas of which are the companies problem, some are individual, some are societal and some a mixture. And we want those solutions to be technically achievable, auditable and not constrain us in the future. That's quite a lot for a few thousand human moderators in poorly air conditioned offices to deal with.
- confidantlake 6y ago
  So if those that think the world is round and those who think it is flat both think the moderator is biased he must be doing something right? Both sides are wrong because technically the earth isn't exactly flat? What about climate change deniers vs climate scientists?
  - lifeisstillgood 6y ago
    I was thinking "Democrats and Republicans" not Flat earthers vs ... umm everyone else. The problem is not the obvious cases. Child porn is a crime in every country, it can be hunted down online. Society generally agrees on that. Similarly for flat earthers. 99% of people in a pub or street will roll their eyes at this shit. It's only because it's all over youtube and so are we that we think it's big. Problems where most societies mostly agree are not the issue. We could build some kind of rebuttal system - a distributed URL based downvoting system. This would help with the labelling / tagging of dubious posts - at the prominence level of the original post. (ie some W3C tagging for fact checking) But this is also not about agreeing the same set of facts - people rarely disagree over facts unless facts lead to different political positions. And that's the real issue. Climate change is probably too huge to cover but let me try - Climate change does not have a solution - the bomb has dropped we are just waiting for the shockwave to hit. Now we can stop pouring fuel on the fire (metaphor breaking sound) by moving off fossil fuel, but also we need to handle the massive changes involving water rights (hello India China) immigration as people get displaced by flooding (hello everywhere). Denial of this is a human response - because otherwise you have to welcome massive change and people who are not confident their needs will be looked after by the establishment. This is the real reason for the disparity in numbers of deniers in US and canada / europe - if US built a welfare state tomorrow it would see a massive drop off in climate change denial. So I think I lost my thread - but fact checking is only part of the politics problem - rapid rebuttal for every stupid online statement is a fine thing to aim for - something good recommendation algorithms could really help with ! But it stops there. Yes it is fine to mandate Ford and GM build cars with airbags and seatbelts - but it is crazy to ask them to build more parking spaces or solve rush hour. Same goes for online political discussions - YouTube must "fix" (god knows what how) recommendations and facebook must fix the newsfeed and twitter must something. But it has to stop somewhere around the point that someone reading a post is informed - that it is clear and obvious there is debate around what is being said. After that it is about human biases and cognitive failures - ie politics. We should work to ensure everyone has the same set of facts. Yes. But the same set of facts when run through different mental models come up with different actions. And I don't think it is right to have FAANG ensure everyone has the same mental model. That's what politics is about - and the fact that two people can have the same facts but use different mental models is basically the difference between every political party ever - and is why everyone thinks they are being censored - because if they weren't censored then obviously everyone would vote their way. On a side note this is something software gives us hope for - I can conceive of building a politics observatory - a sort of version of that UK economy coloured water experiment - where you can view the economy globally - the shows made in China shipped to Fermany and sold - and see how immigration or welfare changes the model. I think it's possible to have people build up a real software model - to try and reflect in software what their own biases are. I think if this was an app everyone had it would have interesting repercussions- would everyone tend towards a common model? Could you ask for explanations why your model on (prejudice dunjoir) was wrong? Hang on - I just described a university education.
SmokeyHamster 6y ago
The problem with Facebook's censorship isn't one of scale. I think they're pretty good at censoring large numbers of people, as we've seen over the last several years. The problem is with determining the standard by which people are censored. Currently, like most large social media companies, Facebook's standard is largely arbitrary, opaque and unequally enforced. Posting audio of Melania Trump's personal phone calls is fine, even though it violates Facebook's policy on posting someone else's personal data without their permission, but posting the contents of Hunter Biden's laptop receives an instant ban. Posting info about the Steele Dossier was fine, even though none of it was verified, but posting any articles about the Biden Ukraine scandal received bans...because none of it was verified. I'm less concerned with censorship and more concerned with the censorship standard being uniformly applied, and right now, on most sites, it's not. Most of the staff they use to censor people are young people with left-leaning politics, so even well meaning censors will inadvertently censor conservatives far more often and without explanation. I think minds.com probably does censorship the best. There's a public log of what gets removed, and it can be appealed to a public group of randomly selected users to judge whether or not the removal was warranted. That's both scalable (since they don't have to hire thousands of moderators) and is responsive and accountable to the community it's trying to foster.

lifeisstillgood 6y ago

Just to expand a little - If you are the moderator and both sides think you are biased, you are probably in the right zone. The problem is do we as society, really want a handful of private companies deciding what the rules are - deciding for example if a Philipino politician can use FAANG to talk to his electorate (because the electorate is all on facebook, say de facto facebook are garejeeping this) Society does not particularly want this, Zuckerberg definitely does not want this. So we need to find a way to have regulations that deal with a set of poorly defined globally circumstances (Philipino election law as well as German hate speech laws), some areas of which are the companies problem, some are individual, some are societal and some a mixture. And we want those solutions to be technically achievable, auditable and not constrain us in the future. That's quite a lot for a few thousand human moderators in poorly air conditioned offices to deal with.

confidantlake 6y ago
So if those that think the world is round and those who think it is flat both think the moderator is biased he must be doing something right? Both sides are wrong because technically the earth isn't exactly flat? What about climate change deniers vs climate scientists?
- lifeisstillgood 6y ago
  I was thinking "Democrats and Republicans" not Flat earthers vs ... umm everyone else. The problem is not the obvious cases. Child porn is a crime in every country, it can be hunted down online. Society generally agrees on that. Similarly for flat earthers. 99% of people in a pub or street will roll their eyes at this shit. It's only because it's all over youtube and so are we that we think it's big. Problems where most societies mostly agree are not the issue. We could build some kind of rebuttal system - a distributed URL based downvoting system. This would help with the labelling / tagging of dubious posts - at the prominence level of the original post. (ie some W3C tagging for fact checking) But this is also not about agreeing the same set of facts - people rarely disagree over facts unless facts lead to different political positions. And that's the real issue. Climate change is probably too huge to cover but let me try - Climate change does not have a solution - the bomb has dropped we are just waiting for the shockwave to hit. Now we can stop pouring fuel on the fire (metaphor breaking sound) by moving off fossil fuel, but also we need to handle the massive changes involving water rights (hello India China) immigration as people get displaced by flooding (hello everywhere). Denial of this is a human response - because otherwise you have to welcome massive change and people who are not confident their needs will be looked after by the establishment. This is the real reason for the disparity in numbers of deniers in US and canada / europe - if US built a welfare state tomorrow it would see a massive drop off in climate change denial. So I think I lost my thread - but fact checking is only part of the politics problem - rapid rebuttal for every stupid online statement is a fine thing to aim for - something good recommendation algorithms could really help with ! But it stops there. Yes it is fine to mandate Ford and GM build cars with airbags and seatbelts - but it is crazy to ask them to build more parking spaces or solve rush hour. Same goes for online political discussions - YouTube must "fix" (god knows what how) recommendations and facebook must fix the newsfeed and twitter must something. But it has to stop somewhere around the point that someone reading a post is informed - that it is clear and obvious there is debate around what is being said. After that it is about human biases and cognitive failures - ie politics. We should work to ensure everyone has the same set of facts. Yes. But the same set of facts when run through different mental models come up with different actions. And I don't think it is right to have FAANG ensure everyone has the same mental model. That's what politics is about - and the fact that two people can have the same facts but use different mental models is basically the difference between every political party ever - and is why everyone thinks they are being censored - because if they weren't censored then obviously everyone would vote their way. On a side note this is something software gives us hope for - I can conceive of building a politics observatory - a sort of version of that UK economy coloured water experiment - where you can view the economy globally - the shows made in China shipped to Fermany and sold - and see how immigration or welfare changes the model. I think it's possible to have people build up a real software model - to try and reflect in software what their own biases are. I think if this was an app everyone had it would have interesting repercussions- would everyone tend towards a common model? Could you ask for explanations why your model on (prejudice dunjoir) was wrong? Hang on - I just described a university education.

SmokeyHamster 6y ago

The problem with Facebook's censorship isn't one of scale. I think they're pretty good at censoring large numbers of people, as we've seen over the last several years. The problem is with determining the standard by which people are censored. Currently, like most large social media companies, Facebook's standard is largely arbitrary, opaque and unequally enforced. Posting audio of Melania Trump's personal phone calls is fine, even though it violates Facebook's policy on posting someone else's personal data without their permission, but posting the contents of Hunter Biden's laptop receives an instant ban. Posting info about the Steele Dossier was fine, even though none of it was verified, but posting any articles about the Biden Ukraine scandal received bans...because none of it was verified. I'm less concerned with censorship and more concerned with the censorship standard being uniformly applied, and right now, on most sites, it's not. Most of the staff they use to censor people are young people with left-leaning politics, so even well meaning censors will inadvertently censor conservatives far more often and without explanation. I think minds.com probably does censorship the best. There's a public log of what gets removed, and it can be appealed to a public group of randomly selected users to judge whether or not the removal was warranted. That's both scalable (since they don't have to hire thousands of moderators) and is responsive and accountable to the community it's trying to foster.