4/10/22, 10:31 Oversight Board | Independent Judgment. Transparency. Legitimacy. With regards to how the slur list is enforced, Meta stated in the "South Africa Slurs" case ( 2021-011-FB-UA) that its “prohibition against slurs is global, but the designation of slurs is market-specific.” It explained that “[i]f a term appears on a market slur list, the hate speech policy prohibits its use in that market.” Meta’s explanation is confusing as to whether its enforcement practices, which may be global in scope, mean that market-designated slurs are also prohibited globally. Meta explained that it defined a market as “a combination of country(ies) and language(s)/dialect(s)” and that “the division between…market[s] is primarily based on a combination of language /dialect and country of the content.” Meta’s content reviewers are “designated to their market based on their linguistic aptitude and cultural and market knowledge.” According to Meta, this content involved the Arabic and Maghreb markets on the slur list. It was routed to these markets “based on a combination of multiple signals such as location, language, and dialect detected in the content, the type of the content and the report type.” It is not sufficiently clear to the Board how the multiple signals work together to determine which markets a piece of content would engage, and whether content containing a word which is a slur in a given market would only be removed if the content relates to that market, or whether it would be removed globally. The Community Standard itself does not explain this process. Meta should issue a comprehensive explanation of how slurs are enforced on the platform. There are multiple areas of opacity in the current policy, including whether slurs designated for particular geographies are removed from the platform only when posted in those geographies or when viewed in those geographies, or regardless of where they are posted or viewed. Meta should also explain how it handles words that are considered a slur in some settings but have an entirely different meaning, one that does not violate any of Meta's policies, elsewhere. The structure of the Community Standard on Hate Speech may also cause confusion. Although the prohibition on slurs appears below the heading for tier three hate speech, the Board finds it unclear whether the prohibition does belong to tier three as slurs do not necessarily target people with segregation or exclusion, which are the focus in the rest of that tier. II.Legitimate aim Any restriction on expression should pursue one of the legitimate aims listed in the ICCPR, which include the “rights of others.” The policy at issue in this case pursued the legitimate aim of protecting the rights of others ( General Comment No. 34, para. 28) to equality, protection against violence, and discrimination based on sexual orientation and gender identity (Article 2, para. 1, Article 26 ICCPR; UN Human Rights Committee, Toonen v. Australia (1992); UN Human Rights Council Resolution 32/2 on the protection against violence and discrimination based on sexual orientation and gender identity). III. Necessity and proportionality The principle of necessity and proportionality provides that any restrictions on freedom of expression “must be appropriate to achieve their protective function; they must be the least intrusive instrument amongst those which might achieve their protective function; [and] they must be proportionate to the interest to be protected” ( General Comment 34, para. 34). It was not necessary to remove the content in this case as the removal was a clear error which was not line with the exception in Meta’s Hate Speech policies. Removal was also not the least intrusive instrument to achieve the legitimate aim because, in each review resulting in removal, the entire carousel containing 10 photos was taken down for alleged policy violations in only one of the photos. Even if the carousel had included one image with impermissible slurs not covered by an exception, removal of the entire carousel would not be a proportionate response. Meta explained to the Board that “the post is considered violating when any photo contains a violation of the Community Standards” and “[u]nlike Facebook, it is not possible for Meta to remove a single image from an Instagram multi-photo post.” Meta stated that an update to the content review tool had been proposed with the aim that reviewers could remove just the violating photo in a carousel, but the update had not been prioritized. The Board does not find this explanation clear, and believes that deprioritizing the update could lead to systemic overenforcement where entire carousels are taken down even though only parts of them are deemed violating. The Board also notes that, where a user posts the same series of photos on Facebook and Instagram, the different treatments of this kind of content on the two platforms would lead to inconsistent results which are not justified by any meaningful policy difference: if one of the photos is violating, this will cause removal of the whole carousel on Instagram, but not on Facebook. 2.Non-discrimination https://www.oversightboard.com/decision/IG-2PJ00L4T 9/11