4 ms·

> ...deleted thousands of automated accounts... This article is not about deleting accounts of normal human users. It’s about halting coordinated propaganda ca

> ...deleted thousands of automated accounts... This article is not about deleting accounts of normal human users. It’s about halting coordinated propaganda campaigns.

harshreality 8y ago
I don't think that's how it works, according to everything I've heard reported by the media. When Twitter or Facebook talk about automated accounts, I believe the people saying that have good (but naive) intentions (in short, I don't think they have much grasp of classical liberal theory or philosophy regarding free speech), but what they actually mean in practice seems to be this: "Only bots display X communication pattern, this account shows a similar communication pattern, therefore this account is a bot." In short, they work backwards. They don't determine something's a bot and then nuke it (they do that too, but that's not what these mass bannings and reporting of it are all about). They observe offensive communications, and if that fits into a certain pattern, the communications alone prove the account is a bot. I'm sure they do ban bots, true bots, to the best of their ability, with automated detection that blocks signups or blocks accounts from communicating when it can determine that there's unauthorized bot-like behavior on the protocol/timing level. That's not what this is about. This is about (semi-)manually blocking of "bots" based on communication content after their automated detection couldn't be sure the account was a bot. How do you think they know they're bots? If someone held a serious penalty over their heads, and told them if they blocked a single non-bot claiming it's a bot, and got it wrong, they'd be in trouble, do you think any of these "bots" would be blocked? I'm not even saying they're not bots (I can't know), but that those companies aren't as certain that they're banning bots as you assume they are. "Bots" seems to have become a code word that also includes agent provocateurs, whether they're Russian operatives or people in some Asian sweat shop being paid to retweet certain things. Newspeak, much? And if you label them truthfully, not as bots, but as propagandists, you have a huge problem because banning based on propaganda is done selectively — provocateurs with some agendas are banned, but others are not.
- MilnerRoute 8y ago
  I think that "communication pattern" is inaccurate information. But maybe that points to the real issue. Do we agree that harmful inaccurate information should be screened for and eliminated? That leads to a follow-up question: how do you screen for inaccurate information at scale?
  - prolikewh0a 8y ago
    >Do we agree that harmful inaccurate information should be screened for and eliminated? What prevents them from removing say, any information about workers rights because THEY deem it harmful? This is a very dangerous road to go down, that leads right to fascism.
masonic 8y ago
If they have a legitimate means of distinguishing automated accounts, why in the hell aren't they automatically culling all of them or dishonoring their signup attempts in the first place?

> ...deleted thousands of automated accounts... This article is not about deleting accounts of normal human users. It’s about halting coordinated propaganda campaigns.

harshreality 8y ago

I don't think that's how it works, according to everything I've heard reported by the media. When Twitter or Facebook talk about automated accounts, I believe the people saying that have good (but naive) intentions (in short, I don't think they have much grasp of classical liberal theory or philosophy regarding free speech), but what they actually mean in practice seems to be this: "Only bots display X communication pattern, this account shows a similar communication pattern, therefore this account is a bot." In short, they work backwards. They don't determine something's a bot and then nuke it (they do that too, but that's not what these mass bannings and reporting of it are all about). They observe offensive communications, and if that fits into a certain pattern, the communications alone prove the account is a bot. I'm sure they do ban bots, true bots, to the best of their ability, with automated detection that blocks signups or blocks accounts from communicating when it can determine that there's unauthorized bot-like behavior on the protocol/timing level. That's not what this is about. This is about (semi-)manually blocking of "bots" based on communication content after their automated detection couldn't be sure the account was a bot. How do you think they know they're bots? If someone held a serious penalty over their heads, and told them if they blocked a single non-bot claiming it's a bot, and got it wrong, they'd be in trouble, do you think any of these "bots" would be blocked? I'm not even saying they're not bots (I can't know), but that those companies aren't as certain that they're banning bots as you assume they are. "Bots" seems to have become a code word that also includes agent provocateurs, whether they're Russian operatives or people in some Asian sweat shop being paid to retweet certain things. Newspeak, much? And if you label them truthfully, not as bots, but as propagandists, you have a huge problem because banning based on propaganda is done selectively — provocateurs with some agendas are banned, but others are not.

MilnerRoute 8y ago
I think that "communication pattern" is inaccurate information. But maybe that points to the real issue. Do we agree that harmful inaccurate information should be screened for and eliminated? That leads to a follow-up question: how do you screen for inaccurate information at scale?
- prolikewh0a 8y ago
  >Do we agree that harmful inaccurate information should be screened for and eliminated? What prevents them from removing say, any information about workers rights because THEY deem it harmful? This is a very dangerous road to go down, that leads right to fascism.

masonic 8y ago

If they have a legitimate means of distinguishing automated accounts, why in the hell aren't they automatically culling all of them or dishonoring their signup attempts in the first place?