Why Netflix Feels Limited Despite Thousands of Films: The Collapse of the Long Tail

Netflix has thousands of films. So does Amazon Prime. Yet somehow browsing feels limited, like you're seeing the same recommendations over and over.

This isn't your imagination.

There's this business theory from 2004 that everyone used to cite. Chris Anderson wrote an article in Wired about something he called The Long Tail. The basic idea was simple and made a lot of sense at the time.

Internet kills physical shelf space constraints. Distribution costs drop to basically nothing. So the economics shift completely. You'd still have your massive hits, obviously. But the real opportunity would be in all the niche stuff. Millions of obscure products, each finding their small audience. Collectively, those niche products would generate as much revenue as the blockbusters.

He used Rhapsody as proof. Early streaming music service. Of their 400,000 tracks, 98% got streamed at least once per month. The long tail wasn't just theory. It was actually happening.

Twenty years later, we should be living in that world. Obscure films thriving alongside Marvel. Experimental directors finding audiences. Genuine diversity.

Instead, we got the opposite.

What Actually Happened

Anderson was right about the technology. Infinite shelf space exists. Streaming platforms don't face physical constraints. Delivering files costs almost nothing compared to making and shipping DVDs. Algorithms exist, and they're incredibly sophisticated.

What he got wrong was assuming those algorithms would distribute attention across the tail.

They didn't. They concentrated it at the head.

There's research on Netflix originals showing that in 2018, the top 10% of titles got about 95% of all viewing. By 2022 this improved a bit to around 75%. Still massively skewed, though. Most of Netflix's catalogue is barely watched.

Music tells the same story. iTunes data showed 24% of tracks sold only one copy. 91% sold fewer than 100 copies. Remember Anderson's Rhapsody data said 98% of tracks got streamed at least once monthly. The gap between "we have this available" and "anyone actually consumes this" is enormous.

This isn't just streaming. It's how attention works online generally. Researchers call it a power law. A small number of things get most of the engagement. Everything else gets basically nothing.

YouTube studies found 10% of videos get 80% of views. The other 90% of videos share the remaining 20%.

Spotify shows the three major labels got 77% of total streams in 2021. The tail exists technically. Almost nobody is listening to it.

So Anderson was right that shelf space changes economics. Just didn't realise it wouldn't change where people's attention goes. And on platforms where revenue ties to engagement rather than purchases, attention is literally everything.

Why Streaming Algorithms Concentrate Attention Instead of Distributing It

The promise was algorithmic recommendations would solve discovery. You can't browse thousands of titles manually. The algorithm does it for you based on sophisticated analysis of what you've watched.

Sounds reasonable.

But here's what algorithms actually optimise for. Not diversity. Engagement. Which is a proxy for subscriber retention.

An algorithm that recommends some obscure Czech film from the seventies might delight a tiny subset of users. More likely, it just confuses people who then click away looking for something else. An algorithm that recommends the latest big series everyone's talking about? That keeps people watching.

Netflix's system is incredibly sophisticated. It looks at directors, actors, genres, what similar users liked, when you watch, what device you use. All sorts of variables.

But sophistication doesn't mean diversity. It means precision in delivering whatever will keep you on the platform.

And that creates a feedback loop. Popular stuff gets recommended more. Which makes it more popular. Which makes it get recommended more. Research has actually shown algorithms can increase concentration of viewership rather than spreading it out.

If the algorithm figures out that showing one massively popular show to millions of people generates more total watch time than showing each person something different, it'll just keep showing that one thing. The algorithm isn't neutral. It actively concentrates attention.

The interface design makes it worse. "Trending Now." "Top 10 in the UK." "Popular on Netflix." These rows aren't random. They're algorithmically determined based on what the platform thinks will generate engagement.

The result is millions of users see basically the same recommendations. The supposed personalisation is just mass curation with a slightly different order for each person.

Why Infinite Choice Makes People More Conservative

There's a psychological thing Anderson didn't account for. When you give people too much choice, they don't explore. They freeze up.

You're looking at thousands of films, and most people don't start digging through obscure stuff. They retreat to what's familiar. They watch what friends are watching. They pick from the top 10. They choose whatever has the most stars, most views, most cultural momentum.

Too much choice creates paralysis. Paralysis creates conformity.

Nielsen data from 2025 shows not one new series appeared in the ten most-watched streaming originals. Every single entry was a returning show, most of them approaching their final seasons.

Audiences given infinite options chose comfort.

This is completely backwards from what the long tail promised. Infinite choice was supposed to liberate niche tastes. Instead, it made people more risk-averse. The mental effort of evaluating thousands of options is so high that most people just default to whatever's already been validated by mass attention.

Streaming platforms know this. That's why their interfaces actually reduce choice instead of expanding it. Netflix homepage shows maybe 50 titles at a time. Not 5,000.

The catalogue is technically infinite. The gateway you actually see is deliberately narrow. And that gateway is controlled by algorithms that favour what's already popular.

Why Niche Streaming Services Work Differently (MUBI, Criterion Channel)

Okay, so niche services do exist and they work differently.

MUBI curates thirty films at a time. They rotate weekly like a digital arthouse cinema. Criterion Channel has 2,500+ films ranging from silent classics to contemporary art house. Shudder does horror. Arrow does cult cinema.

These prove that streaming technology isn't inherently hostile to the tail. It's the business model that matters.

They operate more like the old video shop model just at a smaller scale. Subscription revenue supports having a deep catalogue. MUBI can be genuinely diverse because it's serving cinephiles who specifically want discovery. Not casual viewers who want comfort.

Criterion can host some three-hour Hungarian film from 1967 because their subscribers actually expect that sort of thing. Even if only 2% watch it.

But these are exceptions. Not solutions.

They're serving a niche within a niche. Cinephiles willing to pay £8-12 per month for a second or third subscription on top of Netflix and Disney+.

The combined subscriber base is in the hundreds of thousands. Not tens of millions. MUBI has maybe 12 million subscribers globally. Netflix has over 280 million. The scale difference is massive.

For most films and most audiences, the platforms that matter are Netflix, Amazon, Disney+, Apple. Those are what shape what gets made and what gets seen.

A £40 million adult drama can't survive on Criterion Channel subscribers. It needs either massive theatrical success or the kind of slow-building mainstream home video revenue that doesn't exist anymore at scale. As detailed in Why Did Mid-Budget Films Disappear?, the economics of home video created a safety net for exactly these kinds of films. Streaming's subscription model can't replicate it.

Criterion proves the old model works at a small scale. Doesn't solve the structural problem of mid-budget films needing mass-market viability.

The tail survives in boutique corners. It collapsed where it actually matters. The economic mainstream that determines what gets greenlit.

What Video Shops Did (That Streaming Doesn't)

The video shop era wasn't some golden age. Selection was limited. Distribution was unequal. Blockbuster would stock fifty copies of whatever was huge and maybe one of something niche if you were lucky.

But the economics created different incentives.

A shop made money whether you rented Terminator 2 or some weird foreign film nobody'd heard of. Same £3-5 transaction either way. No algorithmic pressure to steer you towards statistically optimal choices.

Shops didn't benefit from you staying longer browsing. They benefited from you finding something you wanted and paying for it.

So there was space for slow-moving stuff. A film that got rented once every few weeks was fine. Not exciting but fine. Over years those occasional rentals built small audiences.

Physical media separated being available from being popular. A film could sit on a shelf for months without anyone touching it, and it would still be there. Didn't have to constantly justify its existence through engagement metrics.

It existed because someone had stocked it. Would keep existing until the shop actively decided to remove it.

Streaming flips this completely. Films have to constantly justify their presence through viewership data. If engagement drops below some threshold, the platform has financial reason to remove it and use that licensing budget for something that might perform better.

Being available becomes dependent on performing.

And because ownership was distributed back then, niche films had redundancy built in. Even after studios stopped making DVDs, copies kept circulating in used markets, private collections, library sales.

The tail of physical media was genuinely long because it wasn't all dependent on one platform's hosting decisions.

The New Scarcity

Anderson was right that the internet eliminated shelf space scarcity. What he didn't see coming was that it would create a different scarcity. Scarcity of attention.

Physical era, scarcity came from distribution. Only so many films could be made, shipped, stocked. Real constraint. But it meant what actually got stocked tended to get watched. Ratio of available to consumed was reasonably high.

Streaming era, scarcity is imposed by algorithms. Anything can be hosted. But only a tiny fraction gets surfaced to most users. Ratio of available to consumed is incredibly low.

Platforms basically have infinite content. Most of it is invisible to most users.

The end result is streaming has more in common with broadcast TV than with the open ecosystem Anderson imagined. A handful of titles get massive promotion and viewership. Everything else exists in theory but is functionally invisible. Rarely surfaced, rarely discovered, rarely watched.

What Got Lost

The long tail wasn't just about money. It was about culture. About weird difficult films being able to survive and find their people over time. About a director being able to make something strange and trust that somewhere, someone would care.

Physical media let this happen not because it was efficient but because it was inefficient. Because ownership was distributed. Because discovery was slow. Because a film on a shelf didn't have to justify itself quarterly based on engagement metrics.

Streaming replaced that with something faster, cheaper, more convenient for users. Also, more fragile, more concentrated, more hostile to anything that doesn't immediately generate scale.

My local video shop had maybe 5,000 films and some handwritten recommendation cards stuck to shelves. Delivered more actual diversity than Netflix's algorithm ever will. Not because it had better technology. Because it had completely different economics.

Sometimes inefficiency serves a purpose.

Anderson's vision wasn't wrong exactly. Just incomplete. He saw the internet would change supply. Didn't see it would also change demand in ways that make the long tail economically pointless for most platforms.

The future he predicted needed infinite shelf space plus a fundamental shift in how platforms make money. We got the first part. The second part never happened.

Instead, we got platforms that make money from attention. And attention doesn't flow to the tail. It flows to the head. More extremely than ever before.

Christopher Bray

Founder & Engineer

Engineering open algorithms to map the invisible connections of cinema. I build discovery tools that look beyond streaming catalogs to ensure film history isn't lost to the algorithm.

Last Watched: The Wackness (2008) ★★★☆☆

GitHubX / Twitter