We research this stuff - a lot of nation-state and financed activity is farms of low-paid people assisted by software running multiple and shared accounts. Misi

We research this stuff - a lot of nation-state and financed activity is farms of low-paid people assisted by software running multiple and shared accounts. Misinfo analysts (including ML tools) can suspect this bc the accounts clock in & clock out and follow a talking points script. Except bc a low paid person is adding a human touch, account abuse teams see it and do nothing due to uncertainty, despite being against TOS. Sometimes it is more obvious. People had MAGA campaign apps that lets Trump's team remotely activate otherwise-legit social media accounts. So the account is real, but popular messages were bot behavior. Twitter is even less likely to want to take down such an account, despite the prohibited behavior.