Every YouTube market runs the same experiment in parallel and almost never checks the other labs' results. A format that's been refined through fifty iterations in Japan — structure, pacing, the exact ritual that makes it work — can be completely unknown in English. The creators who occasionally port one ("I tried the Japanese studying-stream format") tend to do embarrassingly well, then everyone acts surprised.

Format arbitrage is the most under-used growth move in content, and the barrier is mundane: you can't browse markets whose language you don't speak. The recommendation algorithm won't show you German breakouts; the language wall keeps each lab's results local. It's a wall a transcript layer walks straight through.

deep_search's spoken-language filter opens the other labs: search your content category with the language set to Japanese, German, Korean, Portuguese — detected from the actual speech, not caption metadata — and the results are that market's real native formats, ranked by transcript relevance. The niche-reach dial surfaces their rising formats, not just their established giants.

library_import_youtube_video and get_youtube_transcript get you inside the promising ones: import, pull the transcript, and study the format's mechanics in text — structure, segment lengths, recurring beats, how episodes open and close. You're not copying content (it's a different language; nothing is copyable) — you're licensing-free importing a format, tested on millions of viewers, into a market where nobody's seen it.