How many Americans can’t buy home broadband because no ISP offers service in their area? You might think the answer to this question would be straightforward. The FCC releases reports on the state of US networks (wireless and wireline) on a regular basis, including the number of US citizens, principally in rural areas, who lack broadband service.
Unfortunately, there’s good reason to believe that the FCC’s data is wrong. That’s the conclusion of a report, conducted on behalf of the FCC, which studied this problem. The Broadband Mapping Initiative was launched in April 2019 by CostQuest, a research firm working on behalf of US Telecom. US Telecom is an industry lobby group made up of bigwigs like AT&T, CenturyLink, Frontier, and Verizon. The project studied broadband deployments in two states: Virginia and Missouri — to determine whether or not the FCC’s previous method of estimating how many US homes lacked broadband was undercounting the total. The answer: Almost certainly yes.
The Impact of a Lousy Measuring Method
Knowing how many US citizens can purchase affordable broadband is critical to understanding whether citizens can access the tools required for modern life. If state and federal governments don’t know which communities or areas lack service, they can’t create projects to target them for improvement. Unfortunately, the FCC has historically allowed ISPs to report whether a location was served using census block data. If one home or business in a census block could be serviced by the company, the company is allowed to claim that the entire census block can be serviced by the company.
This logically raises another question: How large is a census block? There’s no uniform answer. In cities, a census block might be one city block. In rural areas, they can be much larger. Some census blocks have zero population; others might be entirely populated by a single large apartment building. This approach to data set building might have made sense decades ago, but today we have far more precise tools at our disposal. The image below shows 10 census blocks that would be considered covered under the current categorization system:
Now, here’s how much service is actually provided in each of those blocks:
The term “fabric” refers to the new system US Telecom built for measuring which addresses were actually served for broadband. What the report found collectively is that 38 percent of the rural areas in Virginia and Missouri that supposedly have broadband service…actually don’t. At all. 61 percent of rural locations supposedly served weren’t actually in the proper location. Twenty-five percent of the supposedly served locations were actually more than 100 meters from where they were supposed to be. Twenty-three percent of the locations were attached to the wrong census block.
In Missouri, 9 percent of non-rural locations and 36 percent of rural locations lacked service. In Virginia, 12 percent of non-rural and 39 percent of rural locations were unserviced. Overall broadband availability data was wrong in 48 percent of rural blocks.
It isn’t clear exactly how big the gap is nationwide, but these findings suggest the number of people without broadband could be significantly larger than the FCC’s official estimate of 21.3 million people. A total of 445,0000 homes and businesses in Missouri and Virginia that are currently counted as having broadband access are now estimated to lack it. Expand that nationwide, and it’s going to give a much more accurate picture of who does and doesn’t have broadband. It’s also virtually certain to mean we’ve done a much worse job extending coverage than people think we have.
But if you think about it, this actually makes sense. I’ve known more than one person who moved and suddenly found themselves unable to get broadband access, despite previous assurances of service at the new address from their own ISP. Stories periodically surface of ISPs telling rural homeowners to pay tens of thousands to hundreds of thousands of dollars for new fiber pulls. I’ve met people who can’t buy broadband at their homes and rely on expensive cellular wireless for internet access, or use satellite service with its awful latency. The plural of “anecdote” is not “data,” but these stories surface on a semi-regular basis. It may not be a common issue, strictly speaking, but it clearly is a problem. Hopefully building better data sets will help solve it.