Rural Download Speeds Are Worse than Reported, Microsoft Study Says
Nearly three-quarters of the downloads hitting Microsoft servers from nonmetropolitan counties are so slow they don’t meet the FCC definition of broadband. Microsoft’s county-level data shows a big gap between what the federal government says is available and what people actually use.
Those of us in the broadband world have read the latest Federal Communications Commission (FCC) 2018 broadband deployment report. This report, using December 2016 data, found that 24.7 million U.S. residents lacked access to fixed broadband at 25 megabits per second (Mbps) download and 3 megabits per second upload, or 25/3 for short. And of these, 78.2 percent or 19.3 million were “rural.”
It is also known that the FCC dataset has multiple issues:
- It is not validated by consumers or third-party entities.
- Entire blocks are considered served if just one business or home in that block has service.
- And the rates are based on maximum advertised speeds, not technical measurements.
About this third point, efforts and tools exist where speed tests can be conducted to get a better idea of the actual speeds available to homes and businesses. However, a national county-level dataset that incorporates actual speeds has remained elusive.
Until earlier this year.
Microsoft released a dataset that includes the percent of people in a particular county using the internet at 25 Mbps. Microsoft calculated this by analyzing its servers’ logs when electronic devices downloaded Microsoft Windows and/or Office updates, as well as when devices used the Microsoft’s Bing search engine and Xbox gaming consoles. However, it is not clear how many download records were utilized per county or the time of day these took place or the duration of the data gathering (e.g., one month, three months, 12 months). Also, it is not clear if these download records include only businesses or only residences, or both ,and if these downloads took place through fixed broadband or mobile internet connections. And finally, we need to bear in mind that the FCC data is the reported maximum speed available, not the speed at which users subscribe to a broadband service (see related article).
Just like the FCC data is not perfect, Microsoft’s isn’t either. But it does provide a different lens through which to look at the digital divide. (And remember, the FCC numbers consider 25 down and 3 up while Microsoft’s methodology does not include a measurement for upload speed.)
Nevertheless, with those limitations stated up front, the differences between the two datasets are startling.
To dive deeper into urban-rural differences, we grouped counties using the USDA-based county urban influence categories. These include 12 categories. The most urban category is large metropolitan areas of 1 million or more. On the other end of the spectrum are remote nonmetropolitan counties that are not adjacent to a metropolitan or micropolitan areas and do not have a town of at least 2,500 people.
The figure below shows the 2017 percent of population, according to the Microsoft dataset, not using the internet at a minimum of 25 Mbps download in metropolitan and nonmetropolitan counties. I also shows the national rate. Overall, according to Microsoft, half of the U.S. population, or about 162 million residents, did not use the internet at a minimum of 25 Mbps download. By comparison, that’s roughly seven times the size of the population that the official FCC data says does not have download speeds of at least 25 Mbps.
Moreover, the gap between metropolitan and nonmetropolitan was also significant: 46% of people in metropolitan counties versus three-quarters in nonmetropolitan counties. A difference of 30 percentage points!
But, remember the USDA codes have twelve categories in total. The table below shows all counties in the U.S. divided into multiple categories based primarily on if the counties were metropolitan or nonmetropolitan and if these were adjacent to metropolitan areas.
Notice how the more remote and smaller the county, the larger the percent not using internet at a minimum of 25 Mbps download. The most remote counties—those adjacent/not adjacent to large/small metro areas and without a town of 2,500 people—had a share close to 90 percent of their population not using the internet at a minimum of 25 Mbps.
The figure below showcases a very large digital divide, as measured by the Microsoft data. The most remote, rural counties were added to the graph. Notice how about 46 percent of people in metropolitan counties did not use the internet at 25 Mbps compared to 76 percent in nonmetro counties and 87 percent in the most remote, rural counties. In other words, the difference between metro and the most remote, rural counties is 40 percentage points!
The graph above depicts clearly that digital parity—the concept of urban and rural having the same level of connectivity, digital skills, and digital mindset—is nonexistent. The concept of digital parity is important to grasp because most digital applications are urban-centered and urban-inspired. What applications would we see rural-centered and rural-inspired if digital parity were a reality? More importantly, how might quality of life in rural communities improve if digital parity were a reality?
This wide digital divide regarding using internet at 25 Mbps speeds is not surprising. Connecting lower density areas is very expensive. Therefore, public-private partnerships are needed to ensure these communities have adequate connectivity.
The main takeaway from this is that accurate data to measure broadband access and use remains elusive. Without accurate and timely broadband data, different pictures emerge (FCC versus Microsoft, for example) of the issue resulting in ineffective strategies and misplaced scarce dollars to truly reduce the digital divide.
Roberto Gallardo (@robertoge) is assistant director of the Purdue University Center for Regional Development and a Purdue extension community and regional economics specialist. He has written extensively on rural development and digital inclusion. He is a senior fellow of the Center for Rural Strategies, which publishes the Daily Yonder.