What "value density" means, and how we measure it

Every page on this site rests on three numbers. They are the same three numbers the product puts on a YouTube page, and they are all any of our rankings use. This page is what they mean.

It is also, deliberately, a page about where they fail. A measurement you cannot argue with is a measurement nobody should trust, so each section below ends with the cases where that number is the wrong thing to look at.

Value density

What it claims: what share of a video’s runtime carries information, rather than orientation, restatement, or recap.

The unit of the measurement is roughly “does this minute contain something the previous minutes did not”. A video that states an idea, gives one example, and stops scores high. A video that states the same idea five times with different framing scores low — not because repetition is dishonest, but because the second, third and fourth statements are not new to you.

Across the 3,486 videos we have measured, the average sits at 59 out of 100. That number is not a grade curve; it is just where explanatory YouTube lands.

Where it does not apply. Density is a measure of information rate, and information rate is not the same as value. A meditation video is not dense. A film essay whose whole point is atmosphere is not dense. Neither is worse for it — they are simply not the kind of video this measurement was built for, which is why we do not publish rankings for categories built on those formats.

Padding share

What it claims: the proportion of the runtime that carries the video forward without adding to it.

This is the intro that recaps what you clicked on, the mid-roll restatement of the thesis, the outro that summarises what you just watched, the section that exists to reach a runtime target. Across the library the median is 40% and the average is 41%.

Padding is the number people find most immediately useful and the one most likely to be misread. A 40% figure does not mean 40% of the video is wasted — it means that if the same content were delivered without the structural repetition, it would be substantially shorter. Sometimes that repetition is doing real work, particularly in instructional material where a cue genuinely needs saying twice.

Where it does not apply. Padding is only defined for videos whose job is to convey something. For a category where the runtime is the product, the number is meaningless and we do not compute it at all rather than print a figure that sounds like a judgement.

When the useful part starts

What it claims: the second at which the video stops setting up and starts delivering.

This is the most practical of the three and the least ambiguous. The median across the library is 1:00. The tail is what matters, though: one video in ten does not start paying off until 3:46, and those are the ones where knowing the number changes what you do.

It is worth being precise about what this is not. It is not a chapter marker, and it does not mean everything before it is worthless — context is often necessary and occasionally the best part. It means that a viewer who already knows why they clicked can start there and lose nothing.

Where it does not apply. Narrative and argumentative videos, where the setup is load-bearing. Skipping to the point of a video whose structure is the argument gets you the conclusion without the reason to believe it.

The verdict

The three numbers collapse into one of three words: worth it, worth skimming, or skip. The rules are ordinary — a video with a high density and an early start is worth it, a good video that takes a long time to arrive is worth skimming, and the skip verdict is reserved for videos where there is not enough underneath the runtime to justify any of it.

About one video in five ends up as a skip. That ratio is not a target and it moves as the library grows.

Four things these numbers are not

They are not YouTube’s opinion. Nothing here is derived from any ranking, rating or recommendation signal on the platform. This is our measurement of the video, and where the two disagree, they disagree.

They are not a judgement of the channel. We publish measurements of individual videos and aggregates over a shelf. We do not publish rankings of the worst channels, and we do not publish a verdict on anyone’s body of work. A channel with one low-density video is a channel that made one low-density video.

They are not a fact-check. This is the limit worth repeating loudest. We measure how much a video says per minute, not whether what it says is true. A confident, well-structured, densely argued video about a claim that does not hold up will score well here. Density and correctness are different questions, and only one of them is one we answer.

They are not fixed. Scores are recalculated as the model behind them improves. A number you read today may have moved by next month, which is why every page on this site carries the date its figures were last recalculated.

Why we publish the numbers at all

Because the alternative is a recommendation you have to take on faith. Every ranking on this site names the measurement it used and the date it used it, so you can disagree with the ordering on grounds other than taste.