Interesting article about FB's harm management process

This article is the complete text of FB’s after action assessment on its response to “Stop the Steal” and the 1/6 insurrection. Well worth reading in full as a reference point for harm and risk management at global scale.

A few points that jumped out at me. FB found that their response was hampered by:

  1. the fact that they didn’t have “Delegitimization” in their taxonomy (for future reference, it’s “D14N”)
  2. the attack was carried out by a network and aimed at a system property (faith in the election results) rather than directed at harming individuals
  3. they had little policy against coordinated authentic harms, i.e. people openly calling for globally harmful outcomes

A boatload of technologies deployed:

“our research during the US2020 IPOC came from rapid work on topic classifiers, CIRD pipelines, regex and classifier tracking in HELLCAT, and manual analysis via CORGI modeling.” Not sure what all of those mean!

1 Like