How we chose 1,284 channels for The Dark Archive

Why did we filter candidate channels across eight niches?

In The Dark Archive, we set out to study where knowledge from long-form YouTube creators lives across the open web. What happens to the knowledge in a video after it is published depends on whether it also exists somewhere as text.

When an explanation exists only inside a video, it is what we call dark matter: knowledge that exists nowhere else as text, so people and answer engines can read, quote or link to it only by way of the video. Some answer engines cite videos often; others, in the one controlled test we found, cited only the written version (which AI engines cite video).

To study this problem systematically, we started with 14,713 candidate channels. We did not attempt to survey the whole of YouTube. Instead, we focused our investigation on eight knowledge niches where research and substantive explanation form the core product: video essay and cultural analysis; history and documentary; science and engineering explainers; true crime and investigative storytelling; business and case-study breakdowns; interview and conversation podcasts; technology deep dives; and personal finance and economics.

To get a comparable sample, we put every candidate through the same six-part qualification screen.

What six criteria determined which channels qualified?

A candidate channel entered our final sample only if it satisfied all six criteria. If a channel failed even a single test, it was disqualified. Each excluded channel was recorded under the first criterion it failed.

We applied the following six criteria to every candidate:

• Subscriber band: Between 20,000 and 300,000 subscribers at the time of evaluation.
• Language: English as the channel’s primary language.
• Long-form duration: Judged by the median length of the channel’s recent uploads that are not Shorts.
• Active publishing: Uploading through the past year, and recently.
• Mostly scripted: Fewer than half of recent uploads are livestreams. We concede that this was our only format screen: we did not separately screen out music, gameplay, compilation, or reaction channels.
• Independent ownership: Not owned by a media company, a news outlet, or a brand’s content team.

Of the 14,713 candidate channels evaluated across our eight niches, only 1,284 met all six criteria. That represents an overall qualification rate of 8.73%.

The subscriber band removed by far the most candidates. We drew the band around mid-sized channels: past the first stage of building an audience and, in our reading, still mostly run by the creator rather than a studio. That reading is ours; we did not measure team size.

The mostly-scripted criterion screened out channels whose recent uploads were mostly livestreams. It is a narrow screen: it looks at the share of livestreams, not at what a video contains, so a gameplay, compilation or reaction channel in one of the eight niches would pass it if it met the other five criteria. We did not measure how many such channels are in the sample.

What can this sample speak for?

In our reading, understanding the boundaries of our sample is essential for interpreting the study. The 1,284 qualified channels are a specific group, not YouTube as a whole.

Specifically, the sample speaks for established, independent, English-language creators producing long-form knowledge content in these eight niches. With between 20,000 and 300,000 subscribers, they are past the first stage of building an audience. They upload regularly, and most of what they upload is not livestreamed.

Yet few of them publish their videos as text. As our note on how many channels keep a written archive shows, systematic text publishing is rare in this group.

The Dark to Sovereign ladder shows where they sit: from Dark and Satellite, with no website of their own, up to Sovereign, a written archive on a domain they own, which almost none reach. The sample speaks to that gap: established creators whose explanations live mostly in video.

Who does the study leave out?

Just as important as who qualified is who the sample leaves out. We believe research must be explicit about its exclusions.

First, the sample excludes smaller channels with fewer than 20,000 subscribers and larger channels with more than 300,000 subscribers. Our findings say nothing about channels outside the band, in either direction.

Second, the sample does not cover Shorts-first channels or livestream-dominated creators. They follow different formats, and the study does not describe them.

Third, the study leaves out non-English channels and channels outside our eight focus niches. While vibrant knowledge ecosystems exist in other languages and categories, evaluating multilingual content and other genres was outside our scope.

Finally, we excluded channels owned by media companies, news outlets and brands’ content teams. In our reading, those owners usually have websites of their own already. Our study was designed to measure independent creator sovereignty, not corporate marketing.

Why focus on established independent creators?

Why did we draw the boundaries of The Dark Archive around these specific 1,284 channels?

In our reading, these channels represent the segment of the creator economy with the most explanation locked in video and the practical means to act on it.

We think smaller channels often lack the time to keep a website alongside regular videos, and media companies already have sites. Independent channels with between 20,000 and 300,000 subscribers produce a large body of explanation while relying on one platform to host it.

When that work lives only in video, engines that cite only text have nothing of it to cite, and it depends on one platform’s policies. We think this group has the most to gain from owning their words on a domain of their own.

Where are the full numbers?

This note outlines how we chose the 1,284 channels in The Dark Archive and what the sample can and cannot speak for. The exact thresholds for duration and upload frequency, the specific count of channels left at each criterion, and the full qualification funnel broken down across all eight niches are documented in The Dark Archive report.

To see where your own channel sits on the ladder and whether your video explanations are visible to the written web, run the free Dark Matter Check.