TPI’s Senior Fellows on Building the Ideal Broadband Map

Episode Summary

Executive Summary: The episode explains why TPI built a broadband mapping platform: to combine fragmented broadband, census, and USAC datasets, support state broadband planning under the IIJA, and enable analysis beyond simple visualization. The team emphasizes data integration, fast cloud-based computation, regressions, and index-building as essential for actionable policy insights, while warning that new FCC broadband fabric data will be large, complex, and not a perfect solution.

Main Topics: Why TPI built a broadband map: Scott explains the project grew out of repeated, time-consuming broadband data merges and the belief that better tools were needed to combine datasets for policy analysis. How TPI’s map differs from others: Nathaniel describes the map as deeper and more analytical than many existing broadband maps, with multiple datasets, fast updating, and built-in graphs, plots, and regressions. Technical challenges of broadband data integration: The discussion highlights inconsistent missing-value conventions, oversized files, GIS complications, and cloud-platform limitations that make broadband data difficult to standardize and analyze. Why analysis matters beyond visualization: The speakers argue that maps alone do not drive policy decisions; regression analysis, aggregation to different geographies, and contextual interpretation are needed to turn map data into action. Broadband connectivity index and data quality: Scott explains the index approach as a way to combine multiple connectivity measures into a broader picture, identify areas needing attention, and expose missing-data problems. New FCC broadband fabric and future use: The team is optimistic about the higher-resolution broadband fabric data but wary of its size, complexity, update burden, and the possibility that users will still aggregate it to higher geographic levels. Policy transparency and USAC data: Sarah connects the project to her earlier work on E-rate transparency and argues that better data matching can help track where subsidy dollars go and improve broadband spending decisions.

Key Arguments: Broadband policy work requires combining multiple datasets; a single dataset rarely yields useful insights. A broadband map is only a tool; the real value comes from how it is used for analysis, regressions, and decision-making. TPI’s platform is more useful than many other maps because it integrates more datasets and supports analytical workflows, not just visualization. Cloud computing such as BigQuery makes it possible to perform large-scale geographic aggregations in minutes instead of days. Data inconsistencies and missing-value conventions can materially distort analysis if not handled correctly. An index can help summarize multiple connectivity measures and identify areas where more precise investigation or data collection is needed. The new FCC broadband fabric will likely improve resolution but will not eliminate data error, update needs, or the need for aggregation. Better government data practices, including more researcher-friendly USAC fields, would improve broadband accountability and analysis.

Data Points: Data sources loaded into the map: At least 9 - Nathaniel counts multiple broadband-related datasets already integrated into the platform. Geographic types analyzed: 23–24 - Nathaniel says the system supports aggregation across roughly two dozen geographic area types. BigQuery processing time: 5–10 minutes max - Nathaniel says calculations that would take days on a laptop can run in minutes on BigQuery. Decade-scale processing: About an hour - He says a decade’s worth of combinations across geographic areas can be run in about an hour. Census blocks in the U.S.: 11 million - Scott uses this to illustrate the granularity of FCC census-block broadband data. Current FCC 477 data size: About 2 gigabytes - Nathaniel cites this as the size of the current FCC 477 dataset. Missing-value coding variants at Census: 7 different ways - Nathaniel notes the Census Bureau uses seven different conventions for missing values.

Pivotal Quotes: "The first and main part of it is I've built several tools that make it very easy for us to combine the data sets at a specific level into bigger geographic areas so that we can have analysis, not just at the raw level of the data, but at aggregator levels." — Nathaniel Lovin: Explaining the core design philosophy of the broadband map platform. "It's not just the map, it's how you use the map." — Sarah O. Lam: Summarizing the show’s central argument that analysis matters more than visualization alone. "We don't have a national strategy of building a broadband map instead of tools and then we're done. We don't, you know, we don't actually care about that. We care about it to the extent that it's useful." — Scott Walston: On the purpose of broadband mapping as a means to policy action rather than an endpoint.

Implications: Policymakers will need flexible tools, not just maps, to target broadband funds effectively. As new FCC fabric data arrives, the challenge will be converting huge, granular datasets into usable, updated analysis for states and researchers.