OpenRoundup enables journalists to merge multiple tables without coding

OpenRoundup tackles a pain point for data journalists: combining records from multiple independently published sources to build accountability reports. Existing interactive wrangling tools treat a single table as the unit of work, but real-world reporting often involves merging many tables. The tool's interface includes five coordinated panels that implement a schema-first, values-on-demand paradigm. It provides live schema previews, ambient data quality alerts, and a recursive treemap visualization of all wrangling operations. Crucially, the entire system runs client-side using DuckDB-WASM, ensuring sensitive data never leaves the journalist's browser.

To validate OpenRoundup, the authors conducted two studies. First, they replicated 17 previously published programming-based journalist workflows using only the tool's interface, demonstrating expressive coverage. Second, four professional data journalists used it on their own tasks; they confirmed it helped those who understand joins conceptually but lack the programming skills to execute them. An unexpected outcome: the tool proved valuable for data journalism education, making consolidation workflows transparent and teachable. OpenRoundup introduces two conceptual contributions: eager table consolidation (building a composite table early) and a declarative vocabulary of Stack and Pack operations.