We supposedly have a decent handle these days of the number of locations that can buy broadband due to the FCC broadband maps and data collection effort. While some folks will argue about the accuracy of the FCC maps, we know a lot more than we did just a few years ago. The maps are supposed to disclose where ISPs are capable of serving but not where they have customers.
But we still don’t have a handle on how many homes have broadband connections – particularly by neighborhood and geographic areas. That’s because ISPs are not required to report in that level of detail. And they shouldn’t be, because the identity of customers is probably the most important trade secret for any ISP.
Policy folks have always wanted to know more about broadband adoption rates. It’s hard to develop state and local policies and programs to get more Internet into homes without knowing specifically who does and doesn’t have broadband, computers, etc.
In September, NTIA and the Census Bureau announced the first results of a joint initiative to produce more granular data on broadband adoption. The Local Estimate of Internet Adoption (LEIA) uses a combination of existing data and statistical modeling to improve estimates of adoption by neighborhood.
Project LEIA got kicked off by making an estimate of the broadband adoption rate in every county in the country. This was done using 2022 data and was released in September 2024. This first trial used microdata from the Census American Community Survey, ancillary data from the FCC, and new modeling techniques to make more accurate estimates.
There is work being considered to change the estimates based on the new realities of the market. Originally the goal was to count landline broadband connections. However, the proliferation of FWA, satellite, and other WISP broadband networks makes it important to count broadband adoption regardless of the network used to deliver it.
There is also consideration being given to expanding LEIA to make estimates by Census tracts. This is a lot more granular since there are 3,144 counties, but 84,400 Census tracts. Broadband availability and adoption has varied significantly by region in every county I’ve ever worked in.
The Census is also considering changes to the questions asked about broadband and computers in the American Community Survey.
“That’s because ISPs are not required to report in that level of detail. And they shouldn’t be, because the identity of customers is probably the most important trade secret for any ISP.”
Interesting that you bring that up, as NTIA’s recent draft Performance Testing Guidance (https://www.ntia.gov/sites/default/files/2024-12/draft_performance_measures_for_bead_last-mile_networks_policy_notice.pdf) would require that level of reporting for BEAD-funded locations. See Sec. IV, p. 14, which would oblige funded providers to “publish their throughput, latency, and reliability measurement results at the location level, as a CSV file linked from their network management practice page…”
Under this proposal, providers would not only have to reveal the specific location of their subscribers supported by BEAD, but also the precise level of service they receive. This raises both competitive and privacy issues.
While NTIA does not appear to have published comments received in response to this Notice (they were due Dec. 19) as of yet, at least NTCA and WTA raised concerns with this provision (https://www.ntca.org/sites/default/files/federal-filing/2024-12/ntca-ntia-bead-121924.pdf and https://w-t-a.org/wp-content/uploads/2024/12/WTA-BEAD-Performance-Comments-FINAL-12.19.24.pdf, respectively).
Yep. Like many other issues, NTIA has gone too far. I suspect this will never happen.
I think it will try to happen, but major providers will pool resources and go after it.
Their overreach might be what starts the battle. In some states, this rule would be outright illegal (customer lists and contracts are legally protected trade secrets) and so state law vs federal agency policy will be a battle.
“But we still don’t have a handle on how many homes have broadband connections – particularly by neighborhood and geographic areas. That’s because ISPs are not required to report in that level of detail.”
Actually, ISPs have long been required to report their customer totals at varied speed thresholds by Census tract, via Form 477. The FCC already uses this data to create national maps of fixed household broadband adoption data by Census tract. See https://www.fcc.gov/internet-access-services-reports.
The FCC’s June 2023 mapping data includes total fixed broadband connections at 25/3 or faster, by tract. It’s only provided in quintiles (0-199 per thousand, 200-399 per thousand, etc.) but there’s no reason FCC couldn’t make it more granular and add speed categories.
This is 100% reported ISP data, updated every six months — not ACS estimates, which have all kinds of problems starting with the total absence of speed information. And changing ACS questions takes years, whereas FCC can unilaterally improve its publication of 477 tract-level adoption data any time it chooses.
National Digital Inclusion Alliance proposed this in a 2022 letter to the FCC 2022 and in comments filed in response to the Commission’s “706” inquiry in 2023. The letter can be found at https://www.fcc.gov/ecfs/document/10128552213969/1.