Threads Algorithm Update Brings Manual Control

The latest Threads algorithm update tackles the inscrutable black box often considered the villain of social media experiences. Whether it is serving up rage-bait, irrelevant memes, or content from people one has never heard of, the black box that decides what users see can be a source of constant frustration. However, Meta’s text-based platform, Threads, is currently experimenting with a surprisingly analogue solution to this digital problem: simply asking the robot politely.

Threads has begun testing a new feature that allows users to manually tune their feed by posting a specific prompt. By typing “Dear algo” followed by a request, users can explicitly instruct the system on what they wish to see more – or less – of.

A Direct Line to the Code

Highlighted by Connor Hayes, the new Threads algorithm update bridges the gap between passive consumption and active curation. Users might desire more Premier League football content or perhaps less political discourse. To achieve this, they simply draft a post starting with the phrase “Dear algo.”

However, this is not a permanent resetting of the user’s digital profile. As Hayes clarified, this is a temporary override.

“When people add ‘Dear Algo’ to a post, it will tell your feed what you want to see more or less for up to three days.”

Threads suggests this three-day window manages short-term interests or moods. Therefore, it does not act as a lifetime preference setting. For instance, a user might seek live updates from a music festival over a weekend. However, they usually prefer returning to their standard content mix by Monday morning.

The Ripple Effect of Engagement

While the explicit instruction only lasts for 72 hours, the potential for long-term change remains. Users engage with the delivered content through likes, replies, and reposts. Consequently, the system naturally learns to prioritise those topics permanently. The “Dear algo” note serves as a jump-start to break the current feed cycle. Subsequently, the machine learning system latches onto these new variables.

Public Declarations and the Threads Algorithm Update

One curious aspect of this trial is that these requests are not private commands sent to a backend server. They are public posts. Because “Dear algo” messages are treated as standard updates, they will appear on the user’s timeline and be visible to their followers (provided their profile is public).

This adds a social layer to the curation process. As the platform notes, because these requests are visible, other users can see them, reply to them, or connect over shared interests. For example, requesting bookshop recommendations signals the algorithm to fetch relevant content. Moreover, it shows human followers that the user feels passionate about this topic.

The Industry Shift Toward User Agency

Threads are not operating in a vacuum with this experiment. A wider industry trend acknowledges that AI curation has become too dominant. Consequently, this situation leaves users feeling powerless.

• YouTube: The video giant recently launched its own test to allow users to use AI prompts to spark inspiration or refine recommendations.
• X (formerly Twitter): The platform is integrating its Grok chatbot to help users better define their interests and curate their consumption.
• Instagram: Meanwhile, Instagram is trialling a dedicated list of topics for adjusting feed preferences. This structured method offers a rigid alternative to the conversational approach Threads currently takes.

It seems the major platforms have collectively realised that the “black box” approach is alienating users who want to feel like they are in the driver’s seat.

Psychology Behind the Threads Algorithm Update

While the ability to command the algorithm sounds appealing on paper, the practical reality may be quite different. There is a distinct difference between what users say they want (total control) and what they actually do (passive scrolling).

Social media platforms historically provide granular privacy controls and feed settings. Yet, only a small fraction of the user base actually utilises them. The vast majority of users, particularly in the post-TikTok era, have been trained to expect the app to do the heavy lifting. The desire is to open the app and be immediately entertained without having to perform administrative tasks or “train” the software.

The Theoretical Value

Therefore, the true brilliance of the “Dear algo” feature may not be in its utility, but in its psychology. Users frequently complain that “the algorithm is ruined” or that they are being force-fed content they dislike. By providing a visible, easy-to-use tool to fix this, Threads removes the validity of that complaint.

If a user is unhappy with their feed, the responsibility now shifts partially to them. If they have not written “Dear algo, show me less of this,” they have not utilised the tools at their disposal. Even if very few people actually type out these requests, the mere existence of the feature offers a sense of reassurance. It suggests that the platform is listening and that the user has agency, even if they choose never to exercise it.

For now, the feature remains a limited test. A global rollout requires users to embrace this opportunity to converse with the code. Conversely, they might simply prefer complaining about the ghost in the machine.