Top tips: How to become invisible to your own algorithms

Top tips is a weekly column where we highlight what’s trending in the tech world and share practical ways to stay ahead. This week, let’s look at a few simple ways to take back control of your recommendations, and stop your algorithm from deciding who you are.

Let's say you search for a video about running—not because you're planning to run a marathon, but because you just saw someone mention it and wondered what the hype was about. You watch one video. Then another. Maybe you watch a few shorts while you're at it.

The next morning, your feed has apparently decided that you are a runner.

Running shoes, marathon training, protein recipes, morning routines, people waking up at 5am to run 10 miles before breakfast; you haven't changed your life overnight—but your feed has.

And that's the strange thing about recommendation algorithms. They don't know why you watched something. They simply have signals to work with.

YouTube, for example, says its recommendation system looks at signals such as clicks, watch time, likes, dislikes, sharing, and survey responses to figure out what content a viewer is likely to find satisfying.

So that one video you watched out of curiosity? It can become part of the story the algorithm is building about you. And this happens all the time.

You watch a few celebrity interviews because you're bored. Suddenly, your feed knows everything about a celebrity you don't even follow. You look up one recipe before dinner. Your feed becomes a cooking channel. You spend an evening watching videos about buying a new phone. For the next few days, you're apparently in the market for every phone, case, charger, and accessory on the internet.

The algorithm isn't necessarily getting you wrong. You're just giving it incomplete information. The good news is that you can change that.

1. Stop teaching the algorithm with things you don't actually care about  

Here's something we rarely think about when scrolling: Our curiosity looks a lot like interest.

If you watch a ridiculous video because you can't believe someone actually posted it, the algorithm doesn't hear your internal monologue saying, "What on earth am I watching?"

It sees that you watched it. If you watch the whole thing, that's even more information.

And if you send it to a friend, like it ironically, or watch three more videos about the same thing just to understand what is going on, you've given the system even more signals.

This doesn't mean every action is treated the same way, or that one accidental click will permanently change your feed. Recommendation systems are much more complicated than that. But platforms do use behavioral signals to personalize recommendations. YouTube, for instance, says watch time, clicks, likes, dislikes, and shares all contribute to its recommendations.

So the next time something appears on your feed that you genuinely don't want more of, don't just keep scrolling.

Use the controls the platform gives you.

On YouTube, you can select Not interested or tell it not to recommend a particular channel. Instagram also provides Not interested and Interested options for shaping recommendations.

Don't let a five-second curiosity become a long-term interest.

2. Stop waiting for your feed to figure you out  

If you want your feed to show you more useful things, you have to give it useful signals too.

Imagine you want to learn photography. You could keep scrolling through your usual feed and hope the algorithm eventually figures it out. Or you could make the process much easier.

Search for photography tutorials. Follow photographers whose work you actually like. Watch the videos you're genuinely interested in. Save the ones you want to come back to. Explore the subject deliberately for a while.

You're essentially giving the algorithm a clearer answer to a very simple question:

What do you actually want to see?

This matters because recommendation systems are designed to learn from your behavior. YouTube describes its recommendations as constantly evolving based on signals from how people interact with content.

And there's a useful lesson here: Don't only tell the algorithm what you don't like. Tell it what you do.

If your feed has become 90% entertainment and 10% everything else, don't just complain that it's useless. Give it something better to work with.

Pick three things you genuinely want more of, maybe technology, books, fitness, photography, history, cooking, or even something completely random. For a week, deliberately interact with those subjects. You might be surprised at how quickly your feed starts changing. After all, the algorithm isn't a mind reader.

It only knows the version of you that you show it.

3. When your algorithm gets stuck, give it a fresh start  

Sometimes, though, you've already gone too far down the rabbit hole. You watched cooking videos for a week. Now your feed thinks you're opening a restaurant. You spent one weekend researching travel. Now every other post is a hotel, an airline deal, or someone standing dramatically in front of a mountain.

You can spend the next few weeks correcting individual recommendations. Or, sometimes, you can just start again.

Instagram introduced a recommendations reset that lets users clear recommended content across Explore, Reels, and Feed and start fresh. After the reset, recommendations begin personalizing again based on the accounts and content you interact with. Instagram also lets people review the accounts they follow as part of the process.

That's useful for something bigger than cleaning up your feed. It's a reminder to occasionally ask:

Is this still what I want to see?

Our interests change. The things we were obsessed with six months ago may not matter anymore. A topic we searched for once might have nothing to do with our lives today.

But algorithms don't necessarily know that unless we give them new information. So every once in a while, take a look at your recommendations.

If your feed feels strangely repetitive, overly specific, or simply doesn't feel like you anymore, use the tools available to reshape it.

And when you're researching something that you don't want influencing your recommendations, use the platform's history controls where appropriate. YouTube, for example, allows users to manage or pause watch and search history, including when they're researching something temporarily that doesn't reflect their usual interests.

So, should we try to escape algorithms?  

Probably not. Algorithms are useful. They help us find a song we end up loving, discover a creator we wouldn't have found otherwise, find a tutorial when we need one, or stumble across something genuinely interesting.

And most of us wouldn't want to go back to a completely random internet where every recommendation was just a roll of the dice. The problem isn't that algorithms know something about us. The problem is when we forget that we're constantly teaching them.

Ask yourself a different question:

When did I teach it that I liked this?

Because your feed isn't necessarily a perfect picture of who you are. It's a collection of clues. And if you don't like the picture it's putting together, you can always give it better clues.