A photograph of handwriting has lost what recognizers want most, stroke order and timing; here the app records writing as it happens. Contributors see a word from a 50,000-item prompt bank and write it on the touchscreen at normal speed; path, coordinates and timing are captured, mapped to the prompted word. Scribbles and slow tracing are sent back; words come in batches of 20 per session from contributors worldwide, so breadth of hands wins over depth on a few writers. Data is delivered as JSON stroke sequences, easy to convert to InkML. IAM-OnDB is non-commercial and twenty years old; this is online handwriting you can train on commercially.