Skip to content
Moduloa
← Academy journal
MOD-11 · Field notes

The script is already written

Research

A channel with 22,000 subscribers took 517,000 views in a month on the exact topic of last week's Journal note: sourced humanoid production numbers. Nobody on YouTube runs a dated, falsifiable robot tracker or a scored predictions register, which are the two formats this site already publishes weekly. Research into where YouTube could send traffic here: what is measured to work, which lanes are empty, what the platform's link rules actually permit, and what a video costs when the research is already done.

2026-08-15 · Field notes · 8 min read · By

This note asks one question: if Moduloa made videos, which topics would actually send people to this site? The method is stated up front because the numbers are perishable: every view count below was read directly off YouTube's search and watch pages on 2026-08-15, from a Norwegian locale, and view counts drift the moment they are written down. They are treated here as one day's measurements, not as facts about the future. The platform-rule claims come from YouTube's and Google's own documentation, which is more durable.

The demand

Someone already made our journal note into a video, and half a million people watched it

517kViews in one month: a 22,100-subscriber channel's sourced video on humanoid factory production numbers
6.97kYouTube subscribers of the leading humanoid-news aggregator brand: the tracking lane is empty
956kSubscribers of Asianometry: narrated manufacturing economics with no factory footage
2MViews in a year on "I Tried Starting A 3D Printing Business In 7 Days"
01 · The near-collision that started this research"Humanoid Robot Factories Now Build One Per Hour, Here Are The Production Numbers", a sub-three-minute video from a channel called Jason Lowe on AI, 22,100 subscribers: stood at 517,000 views one month after posting, with primary sources in the description. That is last week's Journal note, almost title for title. The demand for sourced production data is real, measured, and evidently independent of channel size.
02 · What owns the ceiling, and what that impliesRaw reach belongs to the makers and the networks: Unitree's official channel has 2.12 million subscribers and an 8.5-million-view video; 60 Minutes' humanoid segment took 2 million views in seven months; a professionally-produced Figure headquarters tour took 690,000 in three. None of that is replicable by a solo founder in Norway without factory access: which is exactly why the replicable lane is analysis and data layered on top of the footage everyone else publishes.
03 · Narration without footage demonstrably worksAsianometry, video essays on business and economics, narration over slides and archival material, no original footage, has 956,000 subscribers, and its latest month-old video stood at 226,000 views. A faceless AI-news channel took 157,000 views in three weeks on a China-robotics explainer. The format Moduloa could actually produce, a screen and a voice over sourced charts, is the format already proven in this niche.
The empty lanes

The two formats this site already publishes have no YouTube incumbent

Nobody runs a dated, sourced tracker

Searches for recurring humanoid production tracking, monthly updates, dated data, sourced claims, return ad-hoc news videos and nothing else. The Humanoid Hub, the dominant aggregator brand elsewhere on the internet, has 6.97 thousand YouTube subscribers. The "X vs Y" comparison format has proven six-figure demand (a Figure-versus-Optimus comparison stood at 313,000 views after eight months; one comparison Short at 1.6 million), but it is crowded with hype channels and thin on sourcing: a data-grounded version keyed to the tracker's availability ladder and named-customer rule would be alone in its sub-lane. This is an absence claim from search sampling, and the honest-limits section treats it as such.

Nobody scores their own robot predictions

The self-grading format is proven outside robotics, a football channel's "reacting to my 2025 predictions" video took 406,000 views in a year, but a search for graded AI and robotics predictions surfaced no equivalent. The register, with its immutable claim text and its first resolved entry a public miss, is a ready-made recurring format that no one else can copy without first having made falsifiable predictions.

The neighborhood print node is an uncovered angle in a proven niche

The 3D-printing business-journey genre reliably draws views at every channel size: 2 million in a year for a seven-day business experiment, 1 million for a print-farm origin story, 602,000 for a farm tour, 71,000 in a year for a first-week diary from a tiny channel. But searching for the specific model PrintDrop runs, a single neighborhood print node, a service in a home rather than a farm in a warehouse, returns 3D-printed housing and generic farm content. The audience demonstrably exists one shelf over; the specific angle is uncontested, and node 001 physically exists to be filmed.

The thin lanes

Two topics that look right for this site and are not

Makerspaces do not draw

The flagship how-to-start-a-makerspace videos have accumulated 26,000 views in five years and 55,000 in ten; a well-produced Munich makerspace tour took 12,000 in five years. For a note filed under Academy, this is the uncomfortable finding: the academy as a topic is not a traffic channel. The academy's method, teaching in public against real work, is what a channel would be. The traffic lives where the work lives: the robots, the print node, the numbers.

Tool walkthroughs do not create demand

The top tutorial for PCPartPicker, a tool with enormous existing brand search, has 23,000 views in two years; the second result has 292. People search for walkthroughs of tools they already use. The implication for the calculator and the scorecard: never make a video about the tool. Use the tool on screen inside a question someone is already asking, what a print actually costs, how the new robot scores, and let the tool be the method. The one measured contrast: a print-farm company's calculator walkthrough took 8,200 views in seven months as lead generation, while its short pricing questions took 21,000–32,000.

The mechanics
01 · Links are tiered by trustPer YouTube's own documentation: channel-profile links (up to 14) are open to every channel; clickable links in long-form descriptions and comments require "advanced features" verification (phone plus channel history or ID); external links in cards and end screens require the Partner Program (1,000 subscribers and watch-hour thresholds), with cards capped at five per video and visible only on computers. Links in Shorts descriptions and comments have been non-clickable platform-wide since 31 August 2023, with a narrow exception for reported brand deals that does not apply to a founder linking their own site. A new channel starts with the channel-profile links and earns the rest.
02 · Nobody has measured the click-through, and that is the findingNo credible published measurement of click-through from YouTube videos to external links exists, not from YouTube, not from the creator-tool vendors. The figure that circulates, "2–10%," is YouTube's official benchmark for thumbnail click-through on impressions, routinely misquoted as link click-through. Creators see their own link numbers privately; nobody publishes them in aggregate. Any traffic projection built on a claimed YouTube link CTR is built on nothing.
03 · "The second-largest search engine" is folkloreThe claim traces to November 2008 comScore data, 2.73 billion on-YouTube search queries in the US that month, more than Yahoo's, which a TechCrunch article framed as "if it were a standalone site, YouTube would be the second largest search engine." The framing was the journalist's, the measurement was one country's on-site query counts eighteen years ago, and no primary source has re-established the ranking since. It is retired from this site's reasoning.
04 · The structural limit is the objective functionYouTube's recommendation system optimizes what its own engineers call valued watchtime: on-platform satisfaction, measured by surveys. YouTube has publicly treated "external links get penalized" as a myth, and no independent test exists in either direction; but the deeper point needs no test: a viewer who leaves for moduloa.com stops generating the thing the system optimizes. Off-platform traffic is at best a tolerated by-product. Google's video documentation adds the on-site half: embedding a YouTube video on your own page makes your page compete with YouTube's watch page for the same result, and Google promises no ranking benefit for embedded video at all.
The cost side

What a video costs when the research already exists

The best-documented solo practitioner in exactly this format is Asianometry's Jon Y: scripts written in the mornings around a day job, then edit, record, and assets "in the span of three hours" once a script is done, sustained for years at roughly weekly cadence before going full-time. The Journal publishes a deep-researched, fully sourced note every week; the expensive part of a Jon Y video, the script, is the part this site already produces. YouTube's growth team states there is no cadence penalty: each video is evaluated on its own. The costs that remain are real: three-ish hours per video of new work, a channel starting from zero subscribers and zero verification tier, and patience measured in years, CNC Kitchen took four and a half years to reach 325,000 subscribers while its founder kept a research-engineer day job.

What conversion looks like in the best documented case

Asianometry's newsletter, which its author describes as "another way for me to reach you guys outside of the YouTube algorithm", has on the order of 42,000 subscribers against roughly 700,000 on YouTube: about six percent, accumulated over years, in the single cleanest audience-to-owned-channel funnel this research found. Nobody in the niche publishes click-through or revenue-per-viewer figures. Six percent of a successful channel, as a ceiling estimate for what YouTube sends home, is the honest planning number, and it is a ratio from one case, which is why it is called an anecdote here and not a benchmark.

The shortlist

Five formats, ranked by evidence and effort

01 · Monthly humanoid production numbers, sources on screenThe 517,000-view near-collision is the direct evidence. Three to five minutes, screen and voice, every number cited: the monthly compression of what the Journal already writes weekly. Feeds /journal and /humanoid-race. Effort: low.
02 · The tracker, diffed: who moved, who slipped, and whyA screen recording of /humanoid-race with the month's changes highlighted. No incumbent found; the comparison genre around it is proven at 313,000 views to 1.6 million. Effort: low.
03 · Quarterly predictions scoring, misses firstThe register read aloud, on camera or on screen, wins and losses: starting with P-26, the public miss. Format proven outside robotics at 406,000 views; nobody does it for robots. Feeds /predictions. Effort: low.
04 · PrintDrop node 001: a real neighborhood print node, real numbersThe business-journey genre at 500,000–2,000,000 views, the single-node neighborhood angle uncovered, and the node physically exists. Phone-filmed episodes plus the numbers on screen. Feeds /printdrop. Effort: medium, this is the one that needs a camera.
05 · What a print actually costs, calculator inside the questionA cost breakdown of a real order, with the calculator on screen as the method, never the subject. Pricing questions measured at 21,000–32,000 views in Short form; standalone tool walkthroughs measured to fail. Feeds /calculator and /printdrop. Effort: low.
The honest limits

What this note does not settle

Five limits. Every view count is a one-day reading from one locale, already stale; the numbers here date the note the way a price list dates a catalogue. The "empty lane" claims rest on search sampling, not exhaustive crawling: a recurring sourced tracker channel could exist under a title these searches missed. No public link-CTR data exists, so nothing above predicts how many viewers a video would actually send to this site; the six-percent Asianometry ratio is one case, not a distribution. The channel this note reasons about does not exist: zero subscribers, no verification tier, no clickable description links on day one, and The Humanoid Hub's 6.97 thousand YouTube subscribers despite owning the niche elsewhere is a warning that authority does not transfer between platforms automatically. And the real cost is the founder's hours, which no source in this note can price. Whether to spend them is a decision, and field notes are not decisions.

Sources

Where these facts came from

Platform rules are cited to YouTube's and Google's own documentation. View counts and subscriber figures were read directly from YouTube watch, channel, and search pages on 2026-08-15 and are named in the text with channel and title so anyone can re-check them; where a video is named but not linked, that is deliberate: the title and channel identify it, and the number will have moved anyway. The 2008 origin of the search-engine claim is cited to the article that coined it.

YouTube Help: sharing links with your audiences (the current link map) ·YouTube Help: advanced features access ·YouTube Help: end screens (external links require the Partner Program) ·YouTube Help: info cards ·YouTube Help: brand-deal links in Shorts (the narrow exception) ·YouTube Help: impressions and click-through-rate FAQs (the misquoted 2–10%) ·YouTube: on YouTube's recommendation system (valued watchtime) ·YouTube Creator Insider: mini mythbuster on external links ·Search Engine Journal: YouTube growth team on algorithm myths (no penalty box) ·TechCrunch: YouTube disables links on Shorts (August 2023) ·Google Search Central: video SEO documentation ·TechCrunch, December 2008: the comScore data behind the "second-largest search engine" line ·Jason Lowe on AI: "Humanoid Robot Factories Now Build One Per Hour" (the 517k measurement) ·The Humanoid Hub on YouTube ·Unitree Robotics on YouTube ·Asianometry on YouTube ·Asianometry: about the newsletter ("outside the YouTube algorithm") ·Asian Century Stocks: ten questions with Jon Y (the newsletter and subscriber figures) ·Hear This Idea: Jon Y on production time per video ·Make:, maker spotlight on CNC Kitchen's Stefan Hermann ·Slant 3D on YouTube (the calculator-walkthrough measurement)

Discussion

Add to this

Corrections, evidence, and disagreement are welcome: this is knowledge in the open. Anyone can read; sign in with GitHub only to post or react, and it appears here instantly.

Field notes are research, not decisions: dated, sourced, and open to correction. Republish this in full or in part under CC BY 4.0, with credit and a link back. · All Academy field notes →
Ask Datum