A book excerpt details how a small team of content curators hired by ByteDance in Mexico City in 2018 shaped TikTok's For You algorithm in Latin America
An exclusive excerpt from Every Screen On The Planet reveals how the social media app's powerful recommendation engine was shaped …
Context & Ripple Effects
The excerpt adds a human, local layer to earlier accounts of TikTok’s recommendation system: prior coverage argued that machine learning could bridge cultural gaps in foreign markets through recommendations adapted across cultures, while TikTok also identified engagement as a primary ranking signal in its own explanation of the feed engagement-led recommendation design. It matters because it suggests Latin American relevance was not solely an automated outcome.
It also sits alongside reporting that internal engineering materials emphasized retention and time spent as core optimization goals. The Mexico City team’s role makes the selection of culturally legible content part of the system’s formative inputs, not just a downstream result.
First-order effects
- The account credits a small Mexico City curation team with shaping the For You system’s early Latin American behavior, complicating a purely algorithmic explanation of how the feed localized.
- It strengthens the case that human content judgment and local market knowledge can materially influence the training, seeding, or feedback environment around recommendation products.
Second-order effects
- Platform teams evaluating TikTok’s regional advantage have a clearer reason to distinguish between ranking-model quality and the local editorial operations that help make a feed feel relevant.
- For creators and media businesses, the history underscores that early distribution in a new market may depend on how platforms classify and surface culturally specific material, not only on universal engagement signals.
Third-order effects
- If this pattern is common, recommendation platforms will increasingly be understood as hybrid systems: automated ranking built on operational choices about what content, communities, and local signals enter the system.
- That framing could sharpen scrutiny of who sets those inputs and how platforms explain regional feed outcomes, especially where engagement optimization and political or cultural exposure intersect.
The trend: This is one data point in the broader shift from viewing feeds as autonomous algorithms to viewing them as socio-technical systems shaped by local human inputs as well as machine-learning objectives.