How 100 Million Children's Drawings Reveal Two Decades of Cultural Trends

Children's drawings act as an unfiltered mirror of cultural shifts, and a new analysis of over 100 million student artworks shows exactly when specific characters, events, and obsessions peak and fade. On September 10, 2026, data analyst Riley Walz published a project charting these trends by scraping two decades of uploads from Artsonia, a digital portfolio platform used by K-12 art teachers.[1]

Artsonia operates as the world's largest online student art museum, currently hosting exactly 138,178,738 published pieces as of late 2026. Teachers use the platform to create digital portfolios, photographing and uploading student work to share with families, while the company donates 20 percent of custom merchandise revenue back to school art programs. Because these uploads are timestamped and tagged, they form a massive, chronological database of what children are thinking about.[2]

Walz processed this unstructured visual data to create searchable trend lines that track the lifespan of cultural phenomena. By querying specific topics, users can see the exact trajectory of a passing fad over the last 20 years. For example, the data shows the rapid rise and subsequent decline of "Baby Yoda" (Grogu), whose appeal among young artists is now waning after a massive peak.[1]

The strength of this evidence lies in its organic origin, which contrasts sharply with other large-scale drawing datasets. Google's "Quick, Draw!" dataset, released in 2017, contains 50 million drawings across 345 categories, extracted from over 1 billion total doodles. However, those users were specifically instructed by an algorithm to draw a "dog" or a "car" within 20 seconds to train machine learning models.[3]

The Artsonia dataset, conversely, is largely unprompted or guided only by open-ended classroom assignments. "Kids shape culture, and their drawings show what captures their attention," notes FlowingData regarding the project's unique value. It captures everyday life, passing fads, and the subjects that children voluntarily choose to spend hours rendering in classrooms.[1]

However, the evidence carries structural limitations that complicate pure cultural analysis. The data is heavily influenced by art teacher curriculums—if a popular lesson plan shared among educators involves drawing a specific subject, that subject will artificially spike in the dataset regardless of broader cultural trends. The signal blends student obsession with teacher instruction.[4]

Furthermore, extracting accurate trends from 100 million images requires relying on the metadata tagged by teachers or students. The scraping process must account for misspellings, varied interpretations of a character, and the inherent noise of unstructured data.[1][4]

Despite these limits, the project demonstrates how massive visual archives can be repurposed for sociological research. By turning millions of individual classroom projects into aggregate trend lines, the analysis provides a quantitative, verifiable measure of childhood imagination over the last twenty years.[4]