5 ms·
It is unfortunately well known that moderators (human and machine) can't reliably distinguish between X and discussion/criticism about X. So it's quite common f
by wmf 6y ago
It is unfortunately well known that moderators (human and machine) can't reliably distinguish between X and discussion/criticism about X. So it's quite common for academic discussion of topics like fascism, conspiracy theories, terrorism, trolling, etc. to be mistakenly censored. Either YouTube/Facebook/Twitter need to invest 1000x more on better moderation or academics should host their content on university sites.
- px43 6y ago"The Cleaners" is a great documentary about this. https://vimeo.com/ondemand/thecleaners2018 https://vimeo.com/ondemand/thecleaners2018 One really eye opening thing that was super obvious in retrospect is the societal context that's required for a lot of the content moderation. No matter how many people you cram into a room in Malaysia, they aren't going to be able to fully appreciate the nuances of things like satire targeted towards a US audience. For the same reason, you can't have a team of well paid moderators in San Francisco moderating political discussions happening in Burma. This is what makes the problem of content moderation so intractable. There are extreme financial and political incentives trying to evade moderation, and no matter how many people you throw at the problem, from any number of backgrounds, your coverage is still going to be dismal. Couple that with the fact that this is probabally the most traumatizing job in the world, with incredibly high suicide rates. It's really really hard to build up a moderation team that can make a sizable dent in the flood of horrific content that floods these platforms. Your best bet is a swarm of AI doing its best, and as many humans as you can hire double checking their work, which is the state of moderation on most major platforms.