I feel conflicted about upvoting. The "Facebook knew and failed to act" framing is ridiculous. So is complaining about casual access being closed off for a document that is clearly highly likely to leak. (And honestly it looks like something that should have been need to know from the start.)
But the leaked document itself is fascinating, both as a post mortem of sorts and as an example of high quality anti abuse work. It is an area where little gets published for obvious reasons, and every group needs to learn the same lessons from scratch. Great read.
Very interesting read, however when trying to do additional research on the toolage mentioned in the article (such as HELLCAT classifiers, CORGI modelling and CIRD pipelines), I'm not able to find anything (outside of the article linked and Facebook groups containing pictures of corgis). Are these just internal codenames for popular mechanisms for modelling data, or are these internal tools that have never seen the light of day?
I believe they are. The article describes some of the graphics that were in the report. The description of title page graphic says it includes a cartoon corgi in a firefighter suit, putting out a fire. CORGI sounds like an internal system that helps with automatic moderation by detecting certain types of content or behavior.
Based on the table, HELLCAT seems like a keyword/sentiment analysis system that surfaces outliers (this group uses terms like "Mainstream Media (MSM)" at a higher average rate than a control group, etc).
> “The authors never intended to publish this as a final document to the whole company," a Facebook spokesperson said in a statement. "They inadvertently published it to a broad audience and they simply restricted it to the internal working group it was intended for."
> The spokesperson added that it was the authors who restricted access to the report.
I am old enough to remember the TV images and sounds of violence in Beirut and the fighting on the streets during the time of the IRA. The storming of the capital seems like incorrect framing of what looked a lot like a desperate attempt to get attention.
These are 2nd amendment types, uncolleged, who all forgot to bring their assault rifles, riot gear and bomb-making equipment to a violent takeover of the US capitol.
There weren’t even videos of burning cars or distant gunshots. To call something that looked more like a college sit-in an “armed insurrection” sounds like framing the pitiful as dangerous and unstable, leading to false imprisonments.
But the leaked document itself is fascinating, both as a post mortem of sorts and as an example of high quality anti abuse work. It is an area where little gets published for obvious reasons, and every group needs to learn the same lessons from scratch. Great read.