The News, Explained
Meta announced on October 13, 2026 that it is rolling out Your algorithm, a Facebook feature for directly shaping recommendations in Feed and Reels. People can add or manage topics they want to see and describe a preferred recommendation direction in their own words. Source
The entry point is the three-dot menu on a post. Meta says it updated this menu across posts to make the controls easier to reach. Inside Your algorithm, a person can add and manage topics for Feed and Reels, such as asking the service to surface more of a particular sport or hobby. Source
Facebook may also ask a follow-up question after someone chooses Interested or Not interested on an individual post. The person can write what they liked, what kind of related material they want more of, or which material they want reduced. Meta says recommendations update in real time and can change within the same session. Source
Existing behavior continues to inform ranking. Meta describes a system that adds freeform feedback to thousands of signals, including likes, comments and viewing activity. Large language models interpret the interests, context and intent in a sentence and combine that meaning with the signals Facebook already uses. A new sentence therefore becomes another ranking input while earlier activity remains part of the system. Source
The announcement gives no country-by-country rollout sequence, supported app versions or completion date. Accounts in the same region may receive the menu at different times during the staged rollout. Meta also leaves several data conditions unspecified: how long freeform feedback is retained, whether people can review or delete their input history, and how these sentences affect advertising personalization. Source
Meta says the change should make recommendations more relevant and nuanced. The post provides no before-and-after reduction in unwanted recommendations and no independent evaluation. Practical value will depend on how quickly and consistently a topic or sentence changes the result and whether that preference persists over time. Source
OYOPICK’s Take
Moving recommendation controls from a settings list into a conversational interface is meaningful. People have usually had to express preferences indirectly by scrolling past, hiding or blocking posts. A direct sentence could reduce the time needed to steer the feed if the system follows it reliably.
Control also requires visibility and reversibility. People should be able to see which requests are active, how long they will remain effective and what happens when a written request conflicts with earlier behavior. A change history and a one-step undo for topics reduced too aggressively would make the result easier to understand.
Natural-language feedback can reveal more personal context than a topic label. Clear explanations of retention, use across recommendations and advertising, and the paths for review and deletion would help the feature expand convenience while preserving meaningful data control.