When You Can’t Trust the Data: Broadband

At the Institute for Local Self-Reliance, we analyze data and explore public policies to empower local communities. Our initiative staff work on varied issues from composting to broadband, but all these issues affect our daily lives and our communities. In the Community Broadband Networks Initiative, we often analyze high-speed Internet service availability using the best data that is publicly available. Some of this data, however, is inaccurate, outdated, and misconstrued.

FCC Form 477 Fails in at Least Four Ways

The most common source of this data is the Form 477. It is designed to be standard, uniform, and provide the Federal Communications Commission (FCC) with detailed information to make sound decisions. The FCC distributes form 477 to Internet Service Providers (ISPs) in order to collect data on their service availability. This form is only accessible online through a government web portal, and it has an accompanying 39-page instruction document. Some of the information is confidential and stripped away before the FCC releases the data to the general public.

The FCC Form 477 may not accurately reflect broadband availability in four main ways: 

1). ISPs may fill out the form improperly. Some ISPs may misplace key information into the form, creating havoc for those analyzing the data. They may input numbers in Kbps instead of Mbps, causing further confusion. For example, a fixed wireless ISP outside of Rochester, Minnesota, offers a maximum speed of 10 Mbps on their website, but the FCC Form 477 states that this ISP advertises a speed of 244 Mbps. Perhaps the ISP meant customers can usually expect a maximum speed of 244 Kbps? Even then, that doesn’t make sense. 

2). The data is out of date. ISPs submit the form twice a year, but the FCC takes time to process this data. By the time we produce maps and research, the underlying data may already be too old to be useful. Mergers may not yet be adequately reflected. For example, at this writing in May 2018 the most recent data currently available is from December 2016. That means the data, the maps, and the research are about a year and a half out of date.

3). The data only includes information maximum advertised download and upload speeds. What the average customer experiences is likely different. They may have bought a lower tier package (see also, broadband adoption) or the technology may get bogged down during periods of peak traffic, such as the early evening.

4). The submitted data itself overstates availability. ISPs can claim an entire census block is served if they could offer service to at least one residence. Census blocks are the smallest unit of measurement for the U.S. Census. They vary in size, shape, and population. Rural census blocks often cover more land area than urban ones.

diagram

Small Errors Add Up to Lost Dollars

We can often compensate for these limitations on the data when the maps and research we produce are small and local. In the case of Rochester, Minnesota, we can remove that ISP that only offers speeds of up to 10 Mbps from any broadband (25 Mbps - download/ 3 Mbps - upload) maps. We can further add caveats to our maps and research that what we present is a best-case scenario.

The limitations of the date can quickly compound when the maps are on a national scale. Rural providers may appear to have a larger footprint than they do because rural census blocks are large. Rural providers, however, may actually only run infrastructure along the roadways in one part of a census block, leaving a majority of the census block unserved. And then we have the real question: how does this affect funding?

The authors of the FCC Broadband Progress Report rely on this data to determine if the country is achieving its broadband goals. The results are the backbone of the FCC’s funding. The FCC's Connect America Fund Phase II Auction came under fire from Jonathan Chambers, former Chief of the Office of Strategic Planning at the FCC, when it dropped 432,302 rural homes and businesses from the program because of the most recent Form 477 data. The data appeared to show that these homes and businesses no longer qualified for the program. The Connect America Fund Auction has $1.98 billion in funding for rural broadband, but because of this flawed data these 432,302 rural homes and businesses will not see any of it.

Remaining Questions

We are left with these questions:

1). How many people are excluded from government funding because of flawed data? 

2). How many people are incorrectly marked as having broadband available? 

3). Why does the FCC not validate and verify the data?

Explore the data for yourself with this FCC database or this FCC broadband map. Submit errors to: broadbandfail@fcc.gov