TPI adds Broadband Connectivity Index to its broadband map

The Technology Policy Institute is adding a “Broadband Connectivity Index” (BCI) to the TPI Broadband Map. The BCI incorporates information from multiple datasets in a way that makes it possible to compare overall connectivity objectively and consistently across geographic areas. Specifically, the BCI uses a machine learning principal components analysis to take into account the share of households that can access fixed speeds of 25/3 and 100/25 (which we calculate by combining the FCC’s Form 477 data with the American Community Survey), average measured download speed (from Ookla), the share of households that connect at 25/3 (from Microsoft), and the share of households with a broadband subscription (from the American Community Survey). The BCI makes it possible to compare connectivity across regions (counties, in this case) in an objective way. Additionally, it can help identify areas that government might usefully target for assistance. We don’t claim that this index is the perfect indicator of connectivity. In some cases, it might magnify errors, particularly if multiple datasets include errors in the same area. We’re still fine-tuning it to reduce error to the extent possible and ensure the index truly captures useful information. Still, this preliminary exercise shows that it is possible to obtain new information on connectivity with existing datasets rather than relying only on future, extremely expensive data.