Broadband Speed: FCC Map Vs. Experience on the Ground
To construct its current map of who has broadband, the FCC relies on the self-reported data of Internet service providers. Another set of data developed through an open-source speed test indicates that the typical Internet user has a much different online experience.
EDITOR’S NOTE: The Measurement Lab (M-Lab) data contains the speed at which users can connect to the internet, as measured by M-Lab equipment. It does not contain information on what rate of connection a consumer pays for. Readers should consider that subscription speeds and other factors could affect the M-Lab data, especially when comparing M-Lab to data from the Federal Communications Commission. While the M-Lab process is designed to measure actual connection speeds of individual consumers, the FCC data shows the highest possible speed (as reported by the ISP) that at least one home or business within a census tract could theoretically subscribe to. So, for example, if a subscriber could purchase 500 megbits per second download but purchased a subscription for only 25 mbps, under normal circumstances that user’s best M-Lab connection speed would be, at most, 25 mbps, while the FCC broadband map would should that 500 mbps is available.
This article discusses broadband availability and usage comparing federal data to data obtained through open source speed test. Small towns and rural areas may lack representation in the open-source speed test. You may participate in the study here.
In February of this year, the Federal Communications Commission (FCC) released the initial version of its Broadband Map, which is focused on the fixed broadband deployment situation across the country (i.e. not mobile). This map and its underlying data represent a change from previous efforts in this area – namely, the National Broadband Map (NBM) that was funded through the American Reinvestment and Recovery Act but is no longer being updated. While the NBM was compiled by different entities in each state with only voluntary participation from providers, the FCC version is based on data from Form 477, which is required from all facilities-based providers (like AT&T, Comcast, and satellite companies) twice each year. The data currently being used are from December 2016, while the NBM still shows June 2014 records. Both the FCC and the NBM provided data at a relatively low level of geographic detail –census blocks, which typically contain 30 to 500 people. There were over 11 million census blocks as of the 2010 Census. Providers are required to detail all census blocks where their service is available, along with the speeds available (from eight possible tiers) and technology type.
On its face, the FCC’s map seems to be an improvement from previous broadband data collection efforts. However, criticism of the map and its underlying data has been fast and furious. Less than a week after the map was released, Citylab released a bruising critique that focused on the imprecision associated with the search-by-address function, the lack of pricing data, and the fact that new Internet providers are left out completely. Motherboard described numerous problems that users faced, including duplicative listings of providers and inaccurate descriptions of the speeds they can provide. A piece in the Daily Yonder worried that defining satellite as broadband was problematic for rural areas. And, a Slate article emphasized that the FCC map only includes fixed (not mobile) data, and that the map does not track actual speeds – only the maximum advertised speeds, which are not necessarily representative of what customers experience.
It is this last piece of the puzzle that we focus on here: the fact that the map’s listing of maximum advertised download speeds does not mesh with true, on-the-ground experiences for many rural areas. This was made clear to us as we conducted focus groups of people who took advantage of rural library hotspot lending programs (we visited 24 small communities in Kansas and Maine with such programs, and conducted focus groups in nine of those. Our research, funded by the Institute for Museum and Library Services, examines the information and connectivity environment for rural areas of Maine and Kansas. The focus group participants commonly pointed out that only relatively slow speeds (less than 10 megabits per second) were available from their local provider, despite the FCC’s broadband map depicting the area as having access to speeds of 25 or even 50 MBPS.
As we tried to figure out this discrepancy, we were alerted to another source of data that would prove useful: Measurement Lab. This lab is a consortium of research, industry, and public-interest partners focused on providing verifiable Internet speed measurements. Essentially, Internet consumers run speed tests through the lab’s website, and the data is compiled by city to paint a picture of “true” speeds that are actually experienced in that location. The data are available for over 87,000 cities across the world – but, as might be expected, the number of tests run is sometimes small in rural locations. This data have been used in several publications – typically to make the point that the FCC’s map is simply not telling the same story as the speed tests, or to measure the actual speeds in public schools. A variety of metrics are available from the Measurement Lab site, but in our opinion the most telling are the median, average, and maximum download speeds, along with the number of tests run (by provider) in a particular city. The M-Lab allows tests from both mobile and fixed connections, but in our experience with rural locations the entries are dominated by the local fixed provider. An example of the summary report available for Clearwater, Kansas (pop. 2,423), is shown below: (As a reminder, the median rate is the point at which half of the speed tests were faster and half were slower.)
We compared the FCC data and the M-Lab data for our 24 library hotspot sites. The results (shown in the table below) are striking. While residents of nearly every city on the list are shown as having at least 25 MBPS available to them, the median speeds of actual connections are below 10 MBPS in all but two cities. In seven cases, the median speed is below 5 MBPS – even though the FCC depicts them as having 25 MBPS available.
FCC vs. M-Lab
Download Speed Comparison (MBPS)
(Population data is from American Community Survey 2012-16. FCC speed is from December 2016. M-Lab data is from June 2017-2018. N/A – speed test not available in that location.)
It is interesting that in most locations, the maximum speed reported is close to (or greatly exceeds) the datapoint reported by the FCC; however, these speeds are far from the typical experience for most users in these locations. For example, in Steuben, Maine, the maximum reported speed was 87 MBPS; however, the median user (out of 119 observations) only experienced speeds of less than 2 MBPS. These discrepancies require further exploration; it may be that the high-speed observation was via a dedicated network (such as a university or library) that is not available to residential customers.
Overall, the M-Lab data is a useful source of information for rural advocates who are concerned that the FCC’s Broadband Map does not paint an accurate picture of the connectivity situation in their area. Their data also should be useful for policymakers. Broadband is a hot topic in the rural arena, with implications for economic development, health care, civic engagement, and education. It is important for broadband activists and policymakers to be aware (and have data to prove) that the FCC’s “maximum speed available” is generally not the same as the “typical speed experienced.” These discrepancies also highlight the need to distinguish between “enterprise” users’ speeds and household access and speeds. Much of the economic development literature correctly recognizes the needs of both businesses and individuals, but fails to distinguish what that means in terms of on-the-ground connectivity options for these two groups.
Brian Whitacre is a professor in the department of Agricultural Economics at Oklahoma State University.
Sharon Strover is a Regents Professor in Communication at the University of Texas, where she directs the Technology and Information Policy Institute.
Colin Rhinesmith is an assistant professor in the School of Library and Information Science at Simmons College.