We measured 1,893 YouTube videos. This is how much of them was filler.

This is a census, not an argument. We measure explanatory YouTube videos on how much of the runtime carries information rather than restating, recapping or setting up, and as of 17 August 2026 we had measured 1,893 of them across fifteen published categories.

Below is what that set looks like in aggregate. Two of the results contradict the standard account of why YouTube videos are long, and one of them contradicted ours.

A note on the numbers. Unlike the rest of this site, the figures on this page are frozen at the date above rather than recalculated on each visit. A report people cite should not move under them. The live version of any of these numbers is on the relevant shelf; when we run this census again it will be published as a new edition with a new date.

The headline

The median measured video carries a padding share of 40%. The mean is 41%, the middle half of the library falls between 33% and 48%.

Read that carefully, because it is easy to inflate. It does not mean two of every five minutes are worthless. It means that if the same material were delivered without its structural repetition, the typical video would be substantially shorter than it is. Some of that repetition earns its place, particularly in instructional material where a cue genuinely needs saying twice.

The verdict distribution is less equivocal: of the 1,893 measured videos, 1,493 were worth watching or worth skimming, and 400 — 21% — were not worth the runtime at all.

The useful part starts at 1:03

Half of all measured videos begin delivering within the first sixty-three seconds. The lower quartile arrives at 31 seconds; the upper at 2:08.

The tail is where this becomes practical. One video in ten does not start paying off until 3:56 or later, and those are the videos where knowing the number in advance changes the outcome — not because the opening is dishonest, but because a viewer who already knows why they clicked can skip it and lose nothing.

Result one: longer videos are less padded

This is the finding that contradicted our own expectation, and it is the strongest pattern in the data.

We assumed padding would rise with runtime — that the long videos would be the stretched ones. The measurement runs the other way, monotonically, across every length band:

Videos under ten minutes are the most padded material in the library. Videos over an hour are the least, by eighteen percentage points, and they also carry the highest measured density of any band.

The explanation is not complicated once the result is in front of you. A subject that genuinely requires an hour fills an hour. A subject that requires six minutes and is being delivered in eleven has to find five minutes somewhere, and restatement is the only material available. Length is a symptom of a bloated subject, not a cause of bloat.

The practical consequence is that runtime is close to worthless as a filter, and the very common habit of preferring the shorter of two videos on the same topic is, on average, backwards.

Result two: the spread across subjects is wider than across lengths

Padding share by category, most padded first. Every category listed here is one we publish a shelf for.

Fourteen points separate design from science — a wider gap than the whole length effect above, and it tracks one property: how far the subject’s core idea can be compressed.

Design, productivity and language learning sit at the top because each is built around ideas that can be stated in a sentence. There is nowhere for a twenty-minute video on such an idea to go except sideways. Science and history sit at the bottom because their material chains: the next minute depends on the previous one, and cutting the chain produces a list of facts rather than a shorter explanation.

History is the clean demonstration. It has the second-longest average runtime of anything we measure — over 40 minutes — and the second-lowest padding share. Under the standard account of YouTube bloat, that combination should not exist.

Result three: the title predicts the padding

We also measure how accurately a video’s title describes what is inside it. Grouped by that measurement, the padding share separates sharply:

Two-thirds of the library has an honest title. At the far end, the ten videos whose titles describe essentially different videos carry a padding share of 76% — three-quarters of the runtime is structure.

We are not claiming causation in either direction. The reasonable reading is that both are downstream of the same decision: a video optimised for the click is optimised for the click throughout, and the title is simply the part of that you can see before you commit.

How to read this, and where it does not apply

The sample is not YouTube. It is 1,893 English-language videos in fifteen explanatory categories, assembled to build a recommendation library rather than to be statistically representative of the platform. It skews toward material somebody thought was worth measuring. Nothing here should be generalised to YouTube as a whole, to non-English video, or to the categories we do not publish.

Padding is not waste. It is the share of runtime that does not add. Some of it does real work.

Density is not accuracy. We measure how much a video says per minute, not whether it is true. Every result on this page is about how videos are built.

The figures are frozen at 17 August 2026. Scores are recalculated as the underlying model improves, so the live numbers on the shelves have already moved slightly from these. This edition will not be edited; the next census will be published separately.

Methodology, and the limits of each of the three measurements, is in what value density means.