Numa Numa Dance — Gary Brolsma (2004)

Numa Numa is older than YouTube. Gary Brolsma, then a 19-year-old from Saddle Brook, New Jersey, recorded himself lip-syncing and chair-dancing to O-Zone's Romanian-language single Dragostea Din Tei and posted the file to Newgrounds on 6 December 2004 — two months before YouTube's domain was registered. Understanding that sequence is essential: the video became a phenomenon on infrastructure that had no concept of virality built into it.

Distribution ran on email attachments, instant-messenger links, message-board embeds and the Flash portals that dominated early-2000s internet culture. There was no share button, no recommendation engine, no view counter that anyone trusted. Spread depended entirely on individual people deciding to forward a file to someone else, which meant the clip had to be genuinely, immediately funny to survive each hop. It was: Brolsma performs with total commitment, only from the shoulders up, in a bedroom, with no irony visible anywhere.

By February 2005 it had reportedly been viewed around two million times on Newgrounds alone, and mainstream press followed. The New York Times profiled Brolsma, VH1 and Good Morning America chased him. He responded by refusing almost all of it — cancelling appearances and withdrawing from public view for years. That reaction became its own precedent, the first widely reported case of an ordinary person being made famous by strangers without consenting to it, and the earliest template for the internet-fame backlash cycle that would define the next two decades.

The format Numa Numa established is the more durable legacy. One person, one webcam, one borrowed song, one fixed frame. That grammar runs directly through lip-sync culture on YouTube, through Musical.ly, and into TikTok, where the entire product is a refinement of what Brolsma improvised by accident in his bedroom.

Because the original predates YouTube, its view history is fragmented across re-uploads, mirrors and the official channel Brolsma eventually created. Aggregate counts across the major copies now sit near 200 million against roughly 10 million in 2005 — a 20x expansion, the highest multiplier in this archive, and a direct consequence of a pre-YouTube artefact being repeatedly migrated onto a platform that could finally count it.