Schools can now measure programming skill on a universal scale — and discovered most kids get stuck at the same point

Researchers applied a clustering technique to 2 million Scratch projects and mapped them to a language proficiency scale (CEFR), creating a universal way to measure programming skill that identifies where learners actually plateau. This means schools and online learning platforms can now see exactly which concepts cause students to get stuck, rather than guessing what works.

For years, programming skill assessment has been fragmented — different platforms use different rubrics, teachers grade subjectively, and nobody knows whether a learner struggling in one environment would struggle in another. This framework creates a common language across schools and platforms, which means a student's progress in one place becomes legible to someone teaching them somewhere else. The bottleneck finding is the real signal: 86.7% of Scratch learners never make it past beginner-intermediate level (B2), and the paper identifies exactly which concepts cause the jam: combining logic, timing, and data representation in the same project. Schools now have a specific diagnosis instead of a guess.