What is going on with the FCC Wireline data?

The FCC gathers Form 477 data every 6 months, and publishes the results regularly. At a high level, the data documents, at a Census block level, the providers, (e.g. “Comcast”), the technology (e.g. “DSL” or “Cable”), the services (business or residential) and the maximum advertised upload and download speeds.

The data is notoriously problematic: a provider can list a block if even one subscriber has service, and the speeds are variously interpreted. Another issue is that many users want to interpret this data as the actual speeds that consumers and businesses are receiving, or that every house or place of business in the block has the advertised level of service. Neither of those conclusions hold!

For example, for “consumer” services, you get the advertised rates; but for business services, you get the contracted rates. Take the rates with a grain of salt!

That being said, we can use this data in two or three ways —mostly to asses the availability and competitiveness of service providers.

The current definition of acceptable speeds is 25Mbs (download) and 3Mbs (upload). We abbreviate this 25↓/3↑. Prior to 2015, the baseline rate was 10↓/1↑. The FCC adoption data currently uses 10/1 for its base rate.

In practice, we need to be building toward higher speeds: the kinds of application being deployed today in medicine and education use high volumes of 2-way data transfers, and the expected difference between download and upload speeds is expected to narrow from a factor of 8 to a factor of 6 in the near future. A good way to think of this is that

  • download is for consumption, while
  • upload is for production or economic development.

The Benton Foundation’s Broadband for America’s Future: A Vision for the 2020s recommends building toward 100↓/100↑

In the near future, the FCC plans to move away from using Census Blocks as the unit of reporting, in favor of the providers declaring actual map-based regions of service. The claim is that this will be more accurate, but much depends on the accuracy of the shapes submitted. It will also make it more problematic to correlate Census Data with the provider regions. Stay tuned, and remember to read What is the “dirty secret about data?”