Meta drew a line between allowed and recommended

The test

On 31 August 2020, Meta's newsroom published "Recommendation Guidelines," describing rules developed with input from 50 experts specializing in recommender systems, expression, safety and digital rights, which govern what Facebook and Instagram will actively recommend to people who do not follow the account posting it. The announcement draws an explicit line: this content is still allowed on the platforms, Meta says, it just will not be shown in places where the company recommends content. The guidelines apply to non-recommendable content and accounts, not to removal decisions.

What the evidence says

The underlying document, "About recommendations on Facebook," lists specific categories excluded from recommendation, including content that discusses self-harm, suicide or eating disorders, sexually suggestive content, content promoting regulated products or cosmetic procedures, and what it calls unoriginal content that is largely repurposed from another source without adding material value. It also excludes accounts with recent policy violations or that build followings through practices like purchasing likes. The 2020 newsroom post is Meta's own announcement of a policy decision; the help-center article is the living version of the same policy as it reads today, and the two agree on the core distinction between what stays up and what gets recommended.

The sample and the variance

Both documents describe policy applied to Meta's own recommendation surfaces platform-wide, not a measured enforcement rate, an audited sample of affected accounts, or a percentage of content actually withheld from recommendation. Neither source states how often the rule is applied in practice or how consistently automated systems versus human reviewers make the underlying calls. A policy statement describes intended treatment; it is not a disclosure of enforcement volume, and this record should not be read as one.

What to try next

As an editorial suggestion beyond what the sources state, a publisher whose reach seems to have narrowed without an account warning could check content against the specific listed categories, particularly the unoriginal-content language, since a repost with no added material value is named directly. This is a checklist drawn from the policy text, not a guarantee that meeting it restores any specific level of recommendation.

• Does any of your recent content fall into a category the help-center article names, such as unoriginal repurposed content?
• Has your account had a recent policy violation that the guidelines describe as a factor separate from the content itself?
• Are you treating "not recommended" and "removed" as the same outcome, when Meta's own document distinguishes them?

Meta's 2020 announcement and its current help-center article agree on the same allowed-versus-recommended distinction; together they describe a standing policy, not a measured account of how often it changes any one creator's reach.