Why the same content performs differently on two platforms

The same eleven-minute video: forty thousand views in one place, two hundred in another, posted the same hour with the same title. Creators conclude one audience has better taste. The audiences overlap heavily. What differs is what each system was built to increase.

What everyone sees

Performance is treated as a property of the content, so identical content should perform comparably wherever attention exists. Performance is a joint product of the content and the function that ranks it. One system rewards watch time, another completion rate, another reply volume — and the same artefact scores differently against each without changing at all.

What is actually happening

Bucher’s study of creators working under algorithmic distribution documents how they infer the ranking criteria from fluctuating reach and then reshape their output around the inferred rule — what she describes as an orientation towards the threat of invisibility rather than towards an audience. Gillespie’s account of the relevance of algorithms supplies the other half: ranking systems encode a specific definition of what is worth surfacing, and that definition is a design decision rather than a neutral measurement of interest. By 2026 the divergence is sharper than when either was written, because platforms optimise for retention windows measured in seconds and publish nothing about the weighting. Identical uploads therefore diverge by an order of magnitude, and the cause is not visible from either result.

Why it stays hidden

The asymmetry hides because creators receive only outcomes, never criteria. Analytics report what happened, not what was rewarded. So the low number gets read as a verdict on quality, and the response — changing the content — addresses the one variable that was already held constant.