The United States grapples with a deep digital divide in which those who need broadband access the most—the poor in rural areas—are the least likely to be connected. But it’s been a challenge for advocates to understand the full scope of the problem nationally, and for local and regional governments to suss out where their most underserved constituents live. Part of the problem is what advocates have long argued is an undercount of the unconnected population by the U.S. agency charged with overseeing internet access.
According to a new report by the company Broadband Now, the U.S. Federal Communications Commission may have underestimated the number of Americans without access to high-speed internet by 20 million people. The researchers also found that those undercounts tended to be greater in states with a higher rural population, meaning the rural poor aren’t getting the funding they need to get connected.
While there are local funding initiatives, states depend significantly on federal dollars, the distribution of which is determined by the FCC’s measurement of who does and doesn’t have have access to broadband. Most recently, the FCC approved a $20.4 billion Rural Digital Opportunity fund to narrow America’s digital gap, starting with the census blocks that the agency’s data show are least connected. But if the FCC’s count is off, funding will be, too.
And that has consequences. More reliable connection can translate to better job opportunities for working rural poor, more competition among farmers, and better education for kids. “For rural areas to be vibrant, with good remote working jobs and competitive economies, everybody needs access to broadband internet,” says John Busby, the managing director of Broadband Now, which pushes for better transparency between internet service providers and their consumers.
The FCC’s latest deployment report, released in May, calculates that 93.7 percent of the American population have broadband access in their area, leaving only 21.3 million Americans without high-speed internet. But when researchers at Broadband Now crunched their own numbers, their analysis suggest the actual number of Americans without broadband access is 42 million—double FCC’s figure. (And that doesn’t account for people who can’t live in areas with broadband infrastructure, but can’t afford it.)
“Frankly, I was surprised about how big the gap was when I got the data back. I’d assumed it was smaller,” says Busby. “It really sheds the light on the need to have better reporting.”
To get its estimate, the Broadband Now team manually ran 11,663 randomly selected addresses through the “check availability” tool of nine large internet service providers that claim to serve those areas. All in all, the team analyzed 20,000 provider-address combinations. A fifth of them indicated that no service was available, suggesting to the researchers that companies may be overstating their availability by 20%, Busby says. The results also show that 13% of the addresses served by multiple providers didn’t actually have available service through any of them. They then applied these rates across the country to get their final estimate of 42 million people without broadband.
The disparity between their estimate and the FCC’s largely comes from the agency’s reliance on Form 477 reports, in which internet providers self-report the locations they serve. Providers can claim to serve the population of an entire census block if service is provided to just one household in that block. After the release of FCC’s May report, the agency’s Democratic commissioners dismissed the report, berating their colleagues for “blindly accepting incorrect data” and using the numbers to “clap its hands and pronounce our broadband job done.”
A state-by-state breakdown of the data further uncovers a wide range of gaps between Broadband Now’s estimates and FCC’s, depending how rural or urban each state is. "We found that in states that are inherently more rural, there's a much bigger gap between FCC estimates and ours than in a densely populated city or in states that are primarily urban,” says Busby, whose team compared the data to the “urban percentage” of each state’s population.
In Mississippi, where less than half of the population live in urban areas, the FCC’s coverage is over-reported by 20%. Similarly, in Arkansas where 56% of the population are urban, the coverage rate is overstated by 23%. Meanwhile, in states like California and Massachusetts, where the urban percentage is above 90%, the gap between FCC and Broadband Now’s coverage estimates hovers under 5 points.
The results point to why Form 477 data is especially detrimental to rural areas, where census blocks are bigger and more sparsely populated. That means houses are spread further apart, so while one house may be able to reach a wired line, the family next door—a few miles over—cannot.
The FCC did not respond to a request for comment, but in August, it announced that it would improve the accuracy of its count by requiring providers to submit geospatial maps of exactly where they provide service. Those maps, according to the new FCC order, would also be checked against crowdsourced information from the public. But the FCC doesn’t plan on implementing that upgrade until after the first distribution phase of the Rural Digital Opportunity fund, which will allocate $16 billion to census blocks that the agency’s data show is “wholly unserved” by internet providers.
That has states concerned over whether they will get their “fair share” of the fund, as Jeff Sural, who heads North Carolina Department of Technology’s broadband infrastructure office, put it to State Scoop. In North Carolina, for example, where the FCC overestimates broadband coverage by more than half a million people according to Broadband Now, state officials have had to turn to other data sources such as resident surveys to figure out who needs broadband funding the most.
In 2018, Georgia’s community affairs department began making its own county-level broadband coverage map as well. The first phase of the project, in which it mapped three counties, has already proved to be a labor- and time-intensive undertaking that involved working with municipal government to build an extensive database of all residential and business locations, and negotiating with broadband providers to get location data. But it also showed that in those three counties alone, the FCC data vastly overstated coverage.