The observation. Viral formats cross language markets with a delay. A format proven in one market is a de-risked attention machine; importing it skips the whole cost of format R&D, leaving only localization.
The precedent is an industry: early 抖音 imported musically-era formats; Bilibili's unboxing/reaction/mukbang came from elsewhere; MrBeast's format has been cloned into every language.
The structural theorem that favors small players: the porting industry (dedicated re-upload accounts, MCNs) chases mass formats and closes those windows in weeks. Niche formats — literary, craft, scholarly — are beneath the industry's notice, so their lag windows stay open for months or years. Same moat as the too-small-for-funded-rivals market size, one level up, at the format layer.
The false-vacancy test (run before importing). A format absent in the target language means one of two things: nobody imported it yet (opportunity), or the need it serves is already occupied by a different-shaped native format (a spoken-video quote format may be "absent" in Chinese because 摘抄美学图文 already serves the same need). Importing into a false vacancy is hitting a wall; the secondary play there is a format upgrade, which is a different, harder fight.
Portability screen: structure-borne formats travel (a sentence pattern, a contrast frame); culture-borne formats die in transit (local memes, local celebrities).
As a practice: a weekly scouting ritual — scan rising formats in English (add Japanese/Korean as needed); per candidate: structure-borne or culture-borne? true vacancy or false? localization cost? Survivors enter a candidate pool. Kin: the capability-lag arbitrage (new AI capabilities reaching zh users months after the English rush) is the same principle at the product layer.
The other direction across the same border: audience-language-is-the-market (an asset serves one audience language) and match-routes-by-duplication-cost (pick formats by what a second rendering costs).