Meta's new Threads trick lets you literally boss the algorithm around

Threads just gave users a magic phrase to steer the algorithm: "Dear Algo."

Meta rolled out a new AI-powered feature on Threads that lets you dictate what shows up in your feed by literally talking to the recommendation engine in public.

Here’s how it works: you type a public post starting with "Dear Algo," followed by what you want to see more or less of. For example, "Dear algo, show me more posts about podcasts" or asking for fewer posts about a TV show you haven’t finished yet.

Once posted, your feed adjusts to that prompt for three days.

Even weirder? It’s completely social. If you spot someone else’s "Dear Algo" post that aligns with your taste, you can repost it to immediately apply their content vibe to your own feed.

The feature is now live in the U.S., the U.K., Australia, and New Zealand, with more regions on the way.

Controlling the feed out loud

Historically, training your algorithm has been a passive, frustrating chore. Platforms like X and Bluesky rely on "Not Interested" buttons or static mute lists. Threads is trying something different by parsing casual, natural language requests.

Meta explicitly designed this to handle fleeting moments, like wanting live NBA commentary tonight without wrecking your feed for the next month.

It also comes at a spicy moment in the social media turf war. Market intelligence firm Similarweb recently reported that Threads hit 141.5 million daily mobile active users as of January 7, pulling ahead of X's 125 million mobile daily users, even though X still controls desktop traffic.

The catch is privacy. Because prompts must be published as public posts, everyone can see what you’re trying to hide or discover. Meta calls this a community experience, but plenty of lurkers probably won't love broadcasting their niche guilty pleasures.

Why it matters

We think this is one of the cleverest product hacks Meta has shipped in a while, even if it comes with an obvious trade-off.

On the user side, the three-day temporary window is genuinely practical. Algorithms are notorious for over-indexing on whatever rabbit hole you fell down yesterday. Letting users temporarily override the feed for an event or hobby without permanently breaking recommendations solves a real headache.

Industry-wide, this pushes social feeds into the conversational AI era. Instead of guessing your intent based on dwell time and frantic taps on "show less," platforms will just let you speak plain English to the model.

Meta's real trick, though, was forcing it to be public. Instead of burying a prompt box inside app settings, they turned algorithm tuning into user-generated content and viral distribution. If you hate posting publicly, it’s annoying. But as a growth hook to drive engagement against X, it’s practically textbook Meta.