Which subjects are worth your time: measured value across twenty subjects

Across the twenty subjects we measure, value ranges from 66.9 in science down to 38.4 in travel — a gap of 28.5 points on a 0–100 scale, between the best-scoring subject in our library and the worst.

Which subject a video sits in turns out to move measured value more than almost anything else we track. Not because one subject is inherently more valuable than another, but because some subjects are structurally suited to explanatory video and some are not, and the difference shows up as a real, measurable gap.

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. When we run this census again it will be published as a new edition with a new date.

Twenty subjects, ranked

Here is the full ranking, highest value first.

Six subjects — science, AI and machine learning, health and fitness, engineering, nutrition and programming — sit above 61. Three sit noticeably below the rest: comedy, gaming and travel, with travel well clear of even those two.

The middle of the ranking is close. Twelve subjects, from history down to business, fall inside a five-point band between 57.1 and 60.7. Read the two ends of the chart as real separation and the middle as a rough ordering.

The subjects at the top chain; the subjects at the bottom compress

Science, engineering, AI and programming all made it into the top six, and they share a property: each minute of material in those subjects depends on the minute before it. You cannot explain backpropagation without first explaining what a gradient is. That dependency structure is exactly what a video is good at — it can only present one thing after another — so subjects built that way translate cleanly into the format.

Travel, gaming and comedy sit at the other end for a related reason. Their core idea usually fits in a sentence — “this place is beautiful,” “this move works,” “this is funny” — and a video built around a one-sentence idea has to fill the rest of its runtime with something else. We measured that mechanism directly in the padding census: travel carries the highest padding share of any subject, at 48.9%.

Value, padding and honesty move together across subjects, not just within them

We have written separately about how padding and title honesty predict a single video’s verdict. The same relationship holds one level up, across whole subjects.

Seven of the top ten subjects by value are also in the top ten for title honesty, and nine of the top ten are in the top ten for the least padding. Science leads the value ranking, has the lowest padding of any subject, and the highest average title honesty. Travel is last on value, highest on padding, and lowest on honesty. These are three separate measurements, computed independently, and they line up anyway.

The likely reading is not that padding causes low value or that low value causes dishonest titles. It is that all three are downstream of the same property of a subject: how well its material actually supports the length and structure a video demands. A subject that supports it scores well on every axis we measure. A subject that does not gets padded, oversold and undervalued together.

What to do with a subject-level number

Use subject as a starting filter, not a verdict on any one video. If you are choosing what to learn on YouTube in general, rather than deciding between two specific videos, the subjects at the top of this list are, on average, the safer bet — science most of all.

Do not write off a low-scoring subject entirely. Thirty-eight-point-four is an average over 64 travel videos, not a ceiling. Some travel videos will beat some science videos; the subject number describes the odds before you know anything else, and a video’s own padding, honesty and verdict scores tell you more about that specific video than its subject does.

Expect the bottom of the list to be the hardest place to find a video worth the runtime. If a video-level verdict said “not worth it” more often for one reason than another, it is more often because the subject made a lean video hard to make in the first place.

How this was measured, and what it does not say

Every measured video carries a value score, and this report groups those scores by the subject a video is assigned to. The reading is ours. Method and its limits are in what value density means.

Four subjects rest on fewer than eighty measured videos — travel, gaming, comedy and DIY & crafts, each newer to the library than the rest. Their position at either end of the ranking is consistent with everything else we measure about them, but read their exact figures as an illustration rather than a tight estimate.

Five subjects do not currently have a public shelf on this site — music, comedy, gaming, travel and DIY & crafts. Their numbers in this report are real and measured the same way as the other fifteen; they are simply not yet linked to a browsable collection.

The sample is not YouTube. It is 3,243 English-language explanatory videos across twenty subjects, assembled to build a recommendation library rather than to be representative of the platform.

The figures are frozen at 26 August 2026. Scores are recalculated as the underlying measurement improves, so the live numbers on the shelves will already have moved slightly from these. This edition will not be edited; the next census will be published separately.