YouTube Broke Its Algorithm in Half. That's Actually Good News.
200 billion. That's the number YouTube CEO Neal Mohan dropped in his 2026 annual letter - 200 billion daily views for Shorts. It sounds like a eulogy for long-form video. A slow-motion extinction told in swipeable 30-second chunks.
It isn't. And the reason it isn't tells you something important about how YouTube actually works right now - and what quietly changed while nobody was paying attention.
The Number Is Real. And Also a Little Padded.
In March 2025, YouTube updated how Shorts views are counted. Previously you needed to watch for a few seconds before it registered. Now, any play or replay counts as a view. Each loop is another tick on the counter.
Creators saw view counts jump by up to 30% overnight. Not because their content got better - because the ruler changed.
Daily Shorts views — year over year
Three years ago this number was 15 billion. The growth is real. The ruler also changed in March 2025 - both are true.
I'm not calling the 200 billion number dishonest. It's technically accurate. But "views" on Shorts and "views" on long-form are measuring different things. One measures attention. The other measures a reel spinning until you swipe away.
Where total YouTube watch time goes
Shorts — ~10%
200B views/day, seconds per view
Long-form — ~90%
Where the attention actually lives
Source: DemandSage platform analysis
The blue sliver is doing a lot of promotional work for a very small share of actual attention.
The CPM Chasm Nobody Talks About
Here's the number that puts everything in context. Shorts creators earn roughly $0.01 to $0.06 per 1,000 views. Long-form creators earn $1.61 to $29.30 for the same 1,000 views. That's not a gap - that's a different planet.
To earn $100 from Shorts alone, you need somewhere between 2 and 10 million views. To earn $100 from long-form, a few thousand targeted viewers will do it.
Estimated CPM — revenue per 1,000 views
The Shorts bar is technically present. If you're squinting looking for it - that's the point.
This is why the "Shorts is killing YouTube" narrative has always been slightly off. Shorts is killing watch time metrics the same way listicles killed magazine word counts - by flooding the zone with short units that don't translate to the same value. The platform is still primarily monetized by long-form attention, and YouTube knows it.
Creator Insider's Todd Sherman explains the explore-exploit system that decides if your Short gets 500 views or 5 million. Worth watching if you've ever wondered why some Shorts blow up and similar ones vanish.
The Algorithm Divorce Nobody Announced
Here's the change that actually matters. In late 2025, YouTube fully decoupled the Shorts recommendation algorithm from the long-form algorithm. Two separate systems. Separate signals. Independent evaluation.
If you've used YouTube for a few years, you probably remember the bleed. You'd watch one cooking Short and suddenly your long-form feed was full of recipe channels you never asked for. Your subscription tab started surfacing random Shorts from channels you barely followed. The algorithm was using your Shorts behavior as a proxy for your entire YouTube identity.
That's largely over now. The two formats now use genuinely separate recommendation stacks. Shorts is optimized for swipe-through rate, loop rate, and first-frame engagement. Long-form is evaluated on viewer satisfaction scores, average percentage viewed, and session depth - whether you kept watching after the video ended.
Your YouTube interests are genuinely different across formats. You might binge cooking Shorts while making dinner and watch 2-hour philosophy lectures when you sit down. Those preferences no longer contaminate each other.
The practical effect for viewers: your long-form recommendations should be cleaner. Less noise from your Shorts history. More of what you actually came to watch.
What Neal Mohan's Letter Actually Said
When Mohan cited 200 billion Shorts views, he framed it alongside TV dominance ("YouTube has been the #1 streaming app on U.S. TVs for nearly three years") and creator economics ($100 billion paid to creators over four years). The message wasn't "Shorts is winning" - it was "we have multiple formats that work."
Mohan's 2026 letter in video form. Shorter than the written version and he's a good speaker. The Shorts stat lands differently when you hear it alongside the TV numbers.
That framing is more honest than the headline number suggests. YouTube spent years trying to force Shorts and long-form into a unified recommendation system - training a single algorithm on mixed signals, building features for "creators" as if a 60-second lip-sync and a 45-minute documentary have the same audience needs. The decoupling is an admission that they don't.
For Creators: The Funnel Finally Makes Sense
Shorts was always supposed to be a discovery funnel. Make a Short, get discovered, viewers cross over to your long-form, everyone wins. In practice, it was hard to execute because the mixed algorithm meant audiences weren't reliably crossing formats - the recommendation layer was already blending everything together before a viewer even made a choice.
With separate systems, the funnel has a cleaner architecture. Shorts builds reach and subscriber count with different people, at different moments. Long-form converts subscribers into watch time and revenue. Channels posting both formats are growing subscriber bases 3x faster than single-format channels - but the key is treating them as different tools, not the same tool at different lengths.
The Ratchet Turned the Other Way
I've written before about YouTube's tendency to ship every change that increases the platform's control and quietly deprecate every feature that gives users more of it. The ratchet only goes one way.
The algorithm split is different. It's a genuine improvement to the user experience - one that costs YouTube something (more engineering complexity, less cross-format data blending) to give viewers cleaner, more accurate recommendations. It's rare enough to be worth noting.
200 billion Shorts views and 90% of watch time still going to long-form isn't a contradiction. It's two different products finally being treated like two different products. The headline number will keep going up. Your long-form feed will keep getting better.
Both things can be true. That used to be harder to believe.
Join the conversation