Try something.

Open the social platform you use most.

Do not scroll yet.

Before the first video moves, before the first photograph appears, before you see a headline, predict what the next ten pieces of content will contain.

Fitness?

Politics?

Relationships?

Travel?

Celebrity?

Money?

Animals?

Beauty?

Anger?

Now scroll.

What is interesting is not simply whether you were correct.

It is that you were able to make a prediction at all.

We usually describe recommendation systems as machines learning us.

We click.

We watch.

We pause.

We skip.

The system collects signals and decides what might hold our attention next.

But there is another direction to the relationship.

The system returns a version of our behaviour to us.

Again.

And again.

And again.

Three videos about running can become seven.

An interest in moving abroad can become a feed filled with people leaving their countries.

One relationship video can become an evening of strangers explaining attachment, rejection, dating and heartbreak.

Eventually repetition begins to resemble importance.

This is where the identity question becomes more interesting than the technology question.

Human beings have always learned about themselves through mirrors.

Family tells us what kind of child we are.

School tells us what we are good at.

Friends reveal which version of us receives laughter.

Culture provides categories into which we can place ourselves.

Now another mirror lives in our pockets.

But this mirror does something unusual.

It rearranges what appears next according to what held our attention before.

That does not mean an algorithm decides who you are.

You can ignore recommendations.

You can confuse the system.

You can search deliberately.

You can leave.

Agency remains.

But Identity Sovereignty in an algorithmic world may require a small new question:

Do I keep seeing this because it matters to me, or does it feel as though it matters to me because I keep seeing it?

Those two experiences can look remarkably similar from inside a feed.