# 121,294 hours: the unwritten archive of long-form YouTube By Tai Nguyen, ExDarkMatter. Published 2026-10-07; updated 2026-10-07. Canonical: https://exdarkmatter.com/notes/the-unwritten-archive-of-long-form-youtube/ **Short answer:** Over 24 months, the 1,284 long-form YouTube channels in The Dark Archive published 121,294 hours of long-form video. The median channel published 72 long-form videos, about 48.8 hours. Nearly all of that work lives only as video: research, interviews and explanation that a reader cannot skim and that answer engines rarely quote one idea at a time. ## What did we count? The Dark Archive studies English-language YouTube channels that publish long-form, script-heavy knowledge work: video essays, history, science explainers, interview podcasts, business breakdowns and similar formats. For every channel in the final sample we listed the long-form uploads of the last 24 months and added up their running time. Shorts and livestreams were left out, so the total measures finished, edited work and nothing else. The result is a single figure for the whole sample. Across 1,284 channels, the long-form output of those two years comes to 121,294 hours of video. The median channel published 72 long-form videos in that window, with a combined running time of about 48.8 hours. The median matters more than the total. A total can be pulled up by a handful of prolific channels. The median says that a typical channel in the sample, not an outlier, produced a body of work that would take a full working week and more just to watch once. ## What kind of material is in those hours? These are not vlogs or clips. The channels in the study were selected because their videos carry knowledge: a historian walking through primary sources, a host interviewing a practitioner for an hour, an explainer that took weeks of reading to script. Much of it is original. Some of it exists nowhere else, because the interview was never written up and the research notes were never published. That is what makes the size of the archive interesting. It is not only a large amount of content. It is a large amount of answers: to questions people search for, ask assistants about and argue over in forums. Each long episode holds many of them, usually spoken in passing, between the introduction and the sponsor read. ## Why does it count as unwritten? Because for almost every channel in the sample, those answers exist only as video. Our [count of channels that keep a written archive](https://exdarkmatter.com/notes/how-many-channels-keep-a-written-archive/) found that only a tiny share of the channels we could measure republish their videos as text on a site they own. The rest of the archive sits on YouTube, and on YouTube alone. Video is a fine way to watch. It is a poor way to look something up. A reader who wants one fact from a long episode has to scrub through it or trust an auto-generated transcript. A search engine can index the title, the description and the captions, but a single answer buried in the middle of an episode rarely surfaces as the source for that question. Answer engines show a similar pattern. They do cite video, and some cite it often. But in [a controlled experiment by OtterlyAI](https://otterly.ai/blog/geo-experiment-video-to-blog/), the same stories were published as videos and as written articles, and the written versions earned roughly 4 in 5 format citations (79.5%). The finding is from one publisher and one set of stories, so it is a signal rather than a law. It points the same way as everything else in this note: text is the format machines find easiest to quote. ## Is a transcript enough? No. A transcript is a record of what was said, in the order it was said. It has no headings, no direct answers and no structure a reader can scan. Pasting one onto a page produces a long block of speech that few people finish and that offers an engine no clear claim to lift. Turning an episode into something quotable takes editing. The useful unit is not the episode but the question inside it. One page per question, with the answer in the first paragraph, a heading that says what the section covers and a link back to the moment in the video where the point is made. That is the shape of a page a person can read in a minute and a machine can attribute to its source. That editing step is also the reason the archive stays unwritten. It takes time, and for a channel funded by advertising and sponsorship the return is not obvious. The hours keep piling up on the platform, and the written version never gets started. ## What can a creator do with an archive this size? The size of the backlog is also the opportunity. A creator who publishes a long video every week already has more raw material than most publishers ever commission. A few practical steps follow from that: - **Start with the back catalogue, not the next upload.** Your older episodes already answer questions people ask today. Pick the ones people still watch or search for. - **Split episodes into questions.** A long interview usually holds several standalone answers. Each one can become its own page, findable on its own. - **Lead with the answer.** Put the direct answer in the opening lines, then add the context, the evidence and the link to the timestamp. - **Publish on a domain you own.** The archive keeps its value when a platform changes its feed, its rules or its terms. - **Keep a steady pace.** A small archive that grows every week does more than a large batch that stops after a month. None of this promises traffic or revenue. Being cited and being clicked are different things, and we make no claim here about income from AI answers. What a written archive does guarantee is that the work exists in a form that people and machines can find, read and credit. ## Where is the full breakdown? This note covers one number for the whole sample. How those hours split across the rungs of the ladder, which kinds of channels publish the most and what the most productive channels have in common is in [The Dark Archive report](https://exdarkmatter.com/report/). For the related finding on access, read [what blocking AI crawlers costs a creator's site](https://exdarkmatter.com/notes/what-blocking-ai-crawlers-costs/). To see how much of your own channel is written down, [run the free Dark Matter Check](https://exdarkmatter.com/check/). ## Sources - [OtterlyAI: video-to-blog citation experiment](https://otterly.ai/blog/geo-experiment-video-to-blog/), retrieved 2026-09-30