DOCUMENTATION
Evidence
Screenshots, source material and explanatory notes supporting the factual claims on the main page.
We believe that everyone deserves to have a voice… assures Youtube
Youtube: the job of our recommendation system is to find that audience
Nothing special. Just an ordinary summer night somewhere in Europe...
From talks with Youtube AI assistant
Sometimes reality forces even a robot to forgot whose side it was supposed to be on...
From talks with Youtube AI assistant
Sometimes reality forces even a robot to forgot whose side it was supposed to be on...
YouTube’s own AI assistant says the algorithm is targeting the wrong audience
Sometimes reality forces even a robot to forgot whose side it was supposed to be on...
The collapse: 48,197 → 2,871 → 1,621 → 931
The first cinematic release received 48,197 impressions in its first seven days. The next three comparable cinematic releases received 2,871, 1,621, and 931.
CTR near 8%. Six months later: a few hundred views.
Two early FIESTA releases showed 7.6–8.0% CTR in Suggested Videos. More than six months later, they had received only 386 and 431 views from that source.
High CTR did not bring meaningful distribution
Multiple videos continued to show strong click-through rates — roughly 8% to 18% in these examples — while receiving very limited exposure.
“Diamonds.” 24 impressions.
YouTube Studio’s own AI assistant called my results “an unattainable dream for most channels” — then called 24 impressions a statistical error.
First appeal to TeamYouTube: evidence attached, manual review requested
I did not ask YouTube for views. I asked for a manual technical review of a specific recommendation/classification anomaly — with metrics, screenshots and the suspected failure mode spelled out.
The audience was there. YouTube kept looking elsewhere.
Several Western European countries watched substantially longer — yet the channel kept receiving more traffic from markets with much weaker retention.
WOW! TeamYouTube’s reply...
Second appeal to YouTube: the anomaly was still there
Weeks after my first request for manual review, the same distribution failure was still visible across geography, impressions and first-hours performance. So I documented it again and contacted YouTube a second time.
$300 spent trying to revive the channel
Google Ads spending for September–October 2026.
Same AI. Millions of views. “Same rules for everyone.”
AI-generated videos were being distributed to hundreds of thousands — sometimes millions — of viewers.
When the incentives point away from the rules
According to my recollection, YouTube’s own AI assistant explained that videos built around familiar existing hits had the advantage because audiences already knew them — while my original music did not.
500K results for my brand name