The Algorithm, the Bubble, and Me

Oliver Hart

There is something about the internet that has been annoying me for quite a while. Not enough to actually do anything about it, obviously. Just enough to complain about it.

Oliver Hart

Mostly when I'm searching for something. It could be information, a product, a review, pretty much anything. And after a while I get this feeling that I'm seeing the same things again. The same websites. The same shops. The same kind of answers.

Oliver Hart

And I always end up thinking... Is this really it? Is this the best information available? Or is this just the information somebody has decided I'm most likely to want?

Oliver Hart

And yes, I know there are algorithms involved. Everybody knows there are algorithms involved. We blame the algorithm for everything now.

Oliver Hart

Video gets no views. Algorithm. Nobody likes your post. Algorithm. You spend two hours watching videos of people restoring rusty lawnmowers. Definitely the algorithm. Nothing to do with you.

Oliver Hart

But if I'm being fair, I knew the word. That's not the same as understanding it.

Oliver Hart

So I started reading. Not because I wanted to become an expert on algorithms. Absolutely not. I have enough things going on. I just wanted to understand why the internet sometimes feels smaller than I know it actually is.

Oliver Hart

And the first thing I realised was that I'd probably been thinking about Google in the wrong way. When I search for something, Google has to choose what to show me. Of course it does. There is far too much stuff online to show me everything.

Oliver Hart

Apparently Google first builds an enormous index of the web. Basically a gigantic catalogue of pages it has found. Then when I search, it tries to work out which results from that catalogue are useful for what I asked.

Oliver Hart

Which immediately made me think... Oh. So when I Google something, I'm not actually searching the entire internet. I'm searching what Google has found, stored and then decided is relevant to my search.

Oliver Hart

That's probably obvious to a lot of people.

Oliver Hart

And I think that's part of what has been irritating me. I've been treating search results almost like an answer. But they're not really an answer. They're a selection.

Oliver Hart

Imagine walking into the biggest library in the world. You tell the librarian you want something about motorcycles. She disappears and comes back with five books.

Oliver Hart

Great. That's useful. I don't want her to come back with seventeen thousand books and dump them on the floor.

Oliver Hart

But those five books aren't the library. They're five books somebody decided were probably useful for me.

Oliver Hart

And maybe I've forgotten that difference. Because Google is so easy. You search. Something appears. You click. Done.

Oliver Hart

The interesting question for me is what happened to everything I didn't see. Not because I think Google is secretly hiding the good stuff from me. I'm not putting aluminium foil over the router yet.

Oliver Hart

I just don't know what's missing. And I don't normally think about it.

Oliver Hart

Products are probably where I notice this the most. I'll be looking for something and suddenly it feels like there are only four companies on Earth selling it.

Oliver Hart

I change the search. Same shops. Change it again. Same shops.

Oliver Hart

And then I start wondering. Are these actually the best shops? Are they just really good at getting into Google? Are some of them advertising? Have I clicked them before? Does Google think I like them?

Oliver Hart

Or do I keep clicking them because I recognise them? Which would make this partly my own fault. That would be annoying.

Oliver Hart

I checked what Google itself says about this. And there are apparently loads of things that can affect what appears. Where you are. What language you use. What you searched for. How relevant something seems. Sometimes personalisation. And with Shopping, you're not necessarily seeing every product available anyway.

Oliver Hart

So changing browser probably isn't the dramatic escape from the system I imagined. I was hoping Firefox might be some sort of witness protection programme for internet users. Apparently not.

Oliver Hart

But this got me thinking about something much bigger than shopping. Because if a system has to choose which products I see... It also has to choose which information I see.

Oliver Hart

And social media does this constantly. YouTube does it. TikTok does it. Instagram does it. Facebook does it.

Oliver Hart

They have far more content than anyone could possibly watch. So something has to choose.

Oliver Hart

And obviously they're going to try to show me things they think I'll like. That's why recommendations are useful.

Oliver Hart

If I spend tonight watching motorcycle videos and tomorrow YouTube recommends a motorcycle video... That's not exactly Sherlock Holmes. I gave them quite a lot to work with.

Oliver Hart

But this is where I start wondering. What happens when it's not motorcycles? What happens when it's an opinion?

Oliver Hart

Let's say I believe something. Doesn't matter what. Politics. Money. Food. Climate. Parenting. Whatever subject currently has strangers shouting at each other online.

Oliver Hart

I watch one video that agrees with me. Then another. Then I like one. Then I follow somebody.

Oliver Hart

And the system notices that I seem to enjoy this stuff. So maybe I get more of it. Which makes sense.

Oliver Hart

But what happens after a year? Do I still know there are good arguments on the other side? Do I see them? Do I even want to see them?

Oliver Hart

That's where this whole bubble thing comes in. I thought it was called an echo chamber. Then I found the term filter bubble. Apparently they're not exactly the same thing. Fantastic. I'd already managed to misunderstand the terminology.

Oliver Hart

In very simple terms, a filter bubble is about information being filtered around you. An echo chamber is more like being in a room where the same ideas keep bouncing back at you.

Oliver Hart

At least that's how I'm choosing to remember it. Because the academic definitions can get considerably less fun.

Oliver Hart

And this reminded me of something completely offline. Imagine you're learning to do something. And you're doing it wrong.

Oliver Hart

But nobody tells you. Every time you do it, people say... Perfect. Exactly right. Keep going.

Oliver Hart

How long before you stop wondering whether you're doing it correctly? Probably not very long.

Oliver Hart

That's what bothers me about always being surrounded by agreement. Not just online. Anywhere.

Oliver Hart

If nobody ever tells me I'm wrong... How do I know I'm right?

Oliver Hart

And being corrected isn't fun. Nobody enjoys hearing... Actually, you're completely wrong and here's a fourteen-page PDF explaining why.

Oliver Hart

Especially if you've already spent three hours arguing about it. At that point you're emotionally invested. You've built a small house there.

Oliver Hart

But sometimes we need somebody to say no. Or... Have you thought about this?

Oliver Hart

A good friend does that. Imagine having a friend who agrees with absolutely everything you say.

Oliver Hart

I'm buying this incredibly expensive thing I can't afford. Great idea.

Oliver Hart

I'm quitting my job tomorrow with no plan. Brilliant.

Oliver Hart

I've watched six YouTube videos and I now understand something scientists have studied for forty years. Finally. Someone brave enough to see the truth.

Oliver Hart

That person isn't helping you. They're just very pleasant to have around.

Oliver Hart

So is that what algorithms are doing? Are they slowly creating a world where everything agrees with me?

Oliver Hart

I thought the answer would be yes. That would have made this episode much easier.

Oliver Hart

Unfortunately, I looked it up. And apparently it's complicated. Of course it is.

Oliver Hart

I found a review from 2024 that had looked at more than a hundred studies about filter bubbles and echo chambers. I thought... Perfect. Someone else has already done all the reading.

Oliver Hart

But the research doesn't give one clean answer. Some studies find stronger bubble effects. Others find much less. It depends on what you're measuring and how you're measuring it.

Oliver Hart

There have even been large experiments where researchers changed the political content people saw on Facebook. People really did see different information. But that didn't automatically change their political opinions.

Oliver Hart

Which ruined another nice simple theory.

Oliver Hart

Apparently we're not empty buckets waiting for Facebook to fill us with opinions. We bring quite a lot of our own nonsense with us.

Oliver Hart

And that was when I found the part of this subject I really didn't want to find. Confirmation bias.

Oliver Hart

That sounds complicated. It isn't. It basically means humans are quite good at noticing things that support what we already believe.

Oliver Hart

And we're not always quite as excited about information saying we're wrong. Which I definitely needed scientists to confirm.

Oliver Hart

Because I've done this. I think most people have.

Oliver Hart

You're thinking about buying something. You spend ages comparing options. Eventually you start wanting one particular thing.

Oliver Hart

And suddenly your searches change. You're no longer searching... Is this product any good?

Oliver Hart

Now you're searching... Why this product is brilliant. Problems with the other product. Ten reasons people love the thing I have already emotionally purchased.

Oliver Hart

Technically you're still researching. You're just researching your way towards the answer you wanted.

Oliver Hart

And that's where this became uncomfortable. Because I've spent years wondering why search engines keep showing me similar things.

Oliver Hart

Turns out I might be helping. Great. I was really hoping this would all be Google's fault.

Oliver Hart

Think about how we search when we already have an opinion. If I search... Why is X bad?

Oliver Hart

I've already told the search engine what kind of information I'm looking for. Bad things about X.

Oliver Hart

Then it finds bad things about X. And I sit there thinking... See? I knew it.

Oliver Hart

Someone else can search... Why is X good? And come back equally satisfied that the internet proved them right.

Oliver Hart

Two people. Same internet. Same search engine. Completely different research.

Oliver Hart

And this isn't just me making things up. Researchers have tested how people's existing beliefs affect what they click on and how they search. They've even found similar problems when people use academic search engines.

Oliver Hart

You can be looking at actual scientific papers and still begin with a question designed to give you the answer you want. Which is almost impressive.

Oliver Hart

We managed to bring human bias into the system specifically designed to help us find evidence. Well done, everyone.

Oliver Hart

So maybe the algorithm isn't building the whole bubble. Maybe I build part of it myself. And then the algorithm comes along and makes it comfortable.

Oliver Hart

That's a thought I keep coming back to.

Oliver Hart

Imagine I keep choosing the same chair every time I walk into a room. Eventually someone notices and starts putting that chair out for me.

Oliver Hart

That's helpful. Until one day I complain that nobody ever offers me another chair.

Oliver Hart

Whose fault is that? I genuinely don't know. Probably both of us.

Oliver Hart

And then there's another problem. Convenience.

Oliver Hart

Because I know what I could do. I could use different search engines. I could go further down the results. I could compare more sources. I could deliberately search for arguments against what I believe.

Oliver Hart

Will I do that every time I search for something? No. Absolutely not.

Oliver Hart

If I want to know when IKEA closes, I'm not conducting an independent investigation. I want to know when IKEA closes.

Oliver Hart

I'm not opening six browsers and asking... But what does Bing think?

Oliver Hart

Convenience wins because convenience is convenient. That may be the least impressive sentence I've ever said. But it's true.

Oliver Hart

We use the same browser. The same apps. The same search engine. The same shops.

Oliver Hart

Not necessarily because we've carefully decided they're the best. They're there. They work. We know how they work.

Oliver Hart

Apparently there's even a term for part of this. Status quo bias. Basically, humans have a tendency to stick with what we've already got.

Oliver Hart

Again, science has spent years discovering that people don't like unnecessary effort. I could have saved them some time.

Oliver Hart

And I think this is where my original irritation actually lives. I know there's more information out there. I know there are more shops. More opinions. More articles. More people who disagree.

Oliver Hart

But most of the time I interact with this tiny convenient window into all of it. Search results. Recommendations. Feeds.

Oliver Hart

And maybe I've slowly started confusing the window with the whole world outside it.

Oliver Hart

The library example keeps coming back to me. The librarian gives me five books. There's nothing wrong with the books. They might be excellent.

Oliver Hart

The mistake is mine if I start believing those five books are the entire library.

Oliver Hart

So what do I do about it? I don't know. And I don't think I'm suddenly going to become some perfect critical thinker after making one podcast episode.

Oliver Hart

Tomorrow I'll probably Google something, click the first result and completely forget everything I've just said. That feels realistic.

Oliver Hart

But maybe when something actually matters, I can make it slightly harder for myself. One extra search.

Oliver Hart

If I think something is true... Search for why it might be wrong.

Oliver Hart

If I'm about to spend a lot of money... Look for people who regret buying it. Then look for people who still love it years later.

Oliver Hart

Not because the negative side is automatically right. That's just the same mistake backwards.

Oliver Hart

I just want to occasionally make sure somebody else gets into the room.

Oliver Hart

Because that's probably the thing that worries me most. Not algorithms. Not Google. Not TikTok.

Oliver Hart

It's the possibility of becoming completely certain about something simply because nothing in my normal information flow gives me a reason not to be.

Oliver Hart

If I'm wrong and everyone around me keeps saying I'm right... At some point being wrong is going to feel exactly like being right.

Oliver Hart

And I don't know how often that happens. I don't know how much algorithms contribute. I don't know how much I contribute.

Oliver Hart

The research doesn't give me a nice clean answer either. Which is irritating. But probably healthier than pretending there is one.

Oliver Hart

Maybe that's what I'm taking away from all this. Not an answer. Just another thing to be aware of.

Oliver Hart

The internet isn't necessarily showing me everything. It can't. There's too much.

Oliver Hart

And the things I see aren't necessarily wrong because they've been selected for me. Selection is necessary. Recommendations are useful. I like recommendations.

Oliver Hart

I just don't want to forget that a selection was made. And that sometimes... I helped make it.

Oliver Hart

Which is a considerably less satisfying conclusion than blaming a mysterious algorithm. I was looking forward to that.

Oliver Hart

Instead I've ended up blaming a complicated mixture of technology, business, psychology... and myself.

Oliver Hart

Very efficient. Nobody leaves happy.

Oliver Hart

Before I go, there's one thing I want to mention because it matters especially for an episode like this. This podcast is made with the help of AI.

Oliver Hart

The voice you're listening to is AI generated. I also use AI to help organise my thoughts, work with the English and research things I don't know enough about.

Oliver Hart

And there are plenty of things I don't know enough about. This episode has demonstrated that quite efficiently.

Oliver Hart

But the questions start with me. The things that annoy me are mine. The thoughts and ideas I want to explore are mine.

Oliver Hart

AI helps me dig into them and turn what I find into something I can understand. And hopefully something that's interesting for you to listen to.

Oliver Hart

That doesn't mean everything here is automatically correct. AI can get things wrong. I can get things wrong. And research doesn't always give one clear answer.

Oliver Hart

Actually, considering what I've just spent this entire episode talking about... Please don't just trust me. That would be embarrassing.

Oliver Hart

Read something I read. Read something I didn't. Disagree with me. Maybe you'll end up somewhere completely different.

Oliver Hart

And if you find something that makes me rethink all of this... Good. That's kind of the point.

Oliver Hart

I'm not making this podcast because I have the answers. I'm making it because apparently my head keeps producing questions.

Oliver Hart

And now I have one more. How much of what I believe is something I actually chose... and how much just happened to be the easiest thing to find?

Oliver Hart

I don't know. But I'll probably be thinking about it tomorrow. Unfortunately.