[pkrein]

I saw a very interesting talk by Mike Tamir, Berkeley and Uber Autonomous Group, about an extension his research group built to detect sensationalism and emotional appeals. The idea was that detecting writing designed to anger, incite and sensationalize (versus state facts) basically separated everything we’d call fake news from journalism: https://www.fakerfact.org/about

[knzhou]

This would only detect the crudest type of bias —- it does absolutely nothing against selective reporting or editorial slant. People read sensational news precisely because they think “respectable” outlets are selectively and disingenuously deciding to ignore it.