The United States of Broadband Map
Different internet speed tests take different approaches. To begin with, “internet speed” isn't a precise phrase, as there is no standard for internet speed measurement (more on that here). This leads to a lot of ambiguity. Should a speed test measure the speed at which the computer accesses content from the internet? If so, what kind of computer should be used to measure the speed? Where in the world should the files be located? Is tracking advertised speeds sufficient?
Measurement Lab (M-Lab) provides one such response to those questions. M-Lab is an internet speed test platform that allows anyone to test their internet connection. (M-Lab was housed at New America’s Open Technology Institute for a decade until early 2019, when it joined Code for Science & Society.) In fact, it is embedded at the top of the Google search results for “internet speed test.” The data generated from these tests is open for all to use and the measurement tools themselves are open source, so unaffiliated developers can also help improve the platform.
M-Lab is designed to determine the speed at which the user accesses data outside the user’s internet service provider’s (ISP) network. This approach provides the most accurate reflection of the user’s experience because most data that users access sits outside the ISP’s network. Other tests mainly measure speeds at which the user accesses data inside the ISP’s network, which is likely to overstate the actual speed that the user experiences.
Currently, the only standardized data collection on broadband deployment speeds is at the Federal Communications Commission (FCC), but its approach is flawed. Through the FCC’s Form 477, ISPs submit data about which areas of the country have access to high-speed broadband and the advertised speeds in those areas, which the FCC puts into a broadband map.
However, Form 477 data is self-serving and inaccurate. First, a census block is considered “served” as long as one address in the entire block has access. This approach leads to overstated availability, especially in rural areas where census blocks are large, and an entire region could be wrongly considered to have access to high-speed broadband. Second, ISPs are required to report only the speeds that they could feasibly provide to households, not the speeds they actually provide.
We need more and better data. The United States of Broadband (USBB) map attempts to fill the gap in understanding between the FCC’s data and the actual speeds experienced by Americans, based on billions of speed tests conducted through M-Lab’s platform. Almost 900,000 measurements are collected per day by Measurement Lab in the United States. You can compare the median upload and download speeds at the census tract, county, zip code, State House, and State Senate levels. You can also compare this data to the median advertised speed data that internet service providers report to the FCC as available.
*Updated 2:45pm on August 28, 2019. The opening paragraph was struck and the final paragraph was updated to clarify that this mapping tool displays broadband speed data—it is not a map of broadband availability. The final paragraph was also updated to provide more insight into the tool’s functionality and to correct the description of exactly what data the tool relies upon, including changing “people” to “measurements” and changing "the datasets" to "median upload and download speeds." We also changed "you can challenge" to "you can also compare" to clarify that the tool does not provide a mechanism for filing challenges.