[glangdale]

I get the theory behind that quiz 'trick question', but does it make any actual legal sense? If the discrimination against a subset happens because of the component of the subset that's protected, then surely it's an attack on a protected category. So if you say "black children are all stupid" the likely implication is that you're saying "unlike white children" (otherwise why would you include black).

Is this even true legally? Could a school discriminate against black children because, hey, they are children? Or a movie theater: "Let's not offer youth discounts to Latinos".

I am not a lawyer, so I don't know the legalistic view, but it seems to be an absurdity.

[throwanem]

It's not generally speaking a matter of law, since the context is that of a US-based private company moderating the behavior of its voluntary users, and we generally treat this in a very laissez-faire fashion by default. The extent to which Facebook heeds the laws of those countries other than the US, in which it does business and would like to continue doing so, is determined almost entirely by the value it places on that business.

Of course, this situation changes in any such case as new legislation is made to address perceived or actual need, but that hasn't really happened yet here. The same is not true of, say, refusing trade to members of a specific ethnic group because of that ethnicity - a case in which laws of the sort I describe have indeed been made.

(And, to be clear, I have serious qualms with pretty much every facet of the way in which Facebook does business, from its content moderation choices, to the fashion in which it monetizes its userbase, to the extent to which the scale of userbase it's deliberately developed - in support of that questionable style of monetization - might make "voluntary" an inaccurate way to describe the choice people have of whether or not to participate there. My description above is not normative, but positive, and should be regarded as such.)

[danso]

From my interpretation, it's not so much a legal worry as it is a statement of company beliefs -- for starters, according to the reproduced slides, religion is not a protected category for Facebook, even though it enjoys federal protections, while gender identity is, even though there's no federal law. Additionally those legal protections are for certain categories. Age has federal protections when it comes to employment, but not for housing.

None of those statuses enjoy special protection when it comes to free speech, so FB is making its own decisions here. In the slide titled "Subsets", you see a rough algorithm that censors are supposed to use:

PC + PC = PC (protected class, I assume)

PC + NPC = NPC (non-protected class)

Irish women = PC

Irish teens = NPC

-------------

These guidelines presumably help censors have a quick test to tell between hate speech and controversial speech, so that posting a Chris Rock stand-up segment won't get you banned. It's obviously imperfect and in flux. The slides note that migrants became a "quasi protected category" after the Syrian refugee crisis "triggered a global conversation around the role Facebook plays in protecting migrants"