FCC ‘looking into’ reported error throwing broadband deployment numbers off by millions

It’s the FCC’s official duty to promote connectivity throughout the U.S., and as part of that it issues a yearly report on improvements to broadband deployment. The latest report, however, seems to contain an error large enough to throw its numbers completely off what Chairman Ajit Pai has already claimed. His office says that they are “looking into the matter.”

The information comes from advocacy organization Free Press, already a thorn in this administration’s side for having pointed out the highly questionable nature of economic claims used to justify the Commission’s new, weaker net neutrality rules.

In a comment (PDF) filed in the upcoming 2018 Broadband Deployment Report’s docket, the organization points out a single huge outlier that vastly, and incorrectly, inflates the numbers of new broadband connections in the country.

These official FCC documents are based on “Form 477” paperwork self-reporting broadband availability, submitted by internet providers abiding more or less by the honor system — which critics already point out is completely an inadequate one on which to base policy.

In the last batch of 477s was one from a company called BarrierFree, an ISP based in the Northeast that was submitting its data for the first time ever. Unfortunately there is a slight discrepancy between the numbers on its form and the numbers in reality.

As Free Press summarizes (very slightly modified for clarity; emphasis theirs):

[BarrierFree] claimed deployment of fiber-to-the-home (“FTTH”) and fixed wireless services (each at downstream/upstream speeds of 940 Mbps/880 Mbps) to Census blocks containing nearly 62 million persons. This claimed level of deployment would make BarrierFree the fourth largest U.S. ISP in terms of population coverage.

We further examined the underlying Form 477 data and discovered that BarrierFree appears to have simply submitted as its coverage area a list of every single Census block in each of eight states in which it claimed service: CT, DC, MD, NJ, NY, PA, RI, and VA.

Further investigation strongly suggests BarrierFree grossly misreported its deployment. BarrierFree claims to offer speed tiers topping out at 940 Mbps/880 Mbps in all of its blocks, using both fiber-to-the-home and fixed wireless services. This speed combination is unique to Verizon’s FiOS FTTH service, and Verizon is the only other 477 filer to claim such a speed tier. But according to BarrierFree’s own website, it does not market fiber-to-the-home service at any speed. Furthermore, the maximum advertised speed for its residential fixed wireless service is 25 Mbps symmetrical.

In other words the company claimed to have gigabit speeds going to 62 million people when really, it has 25 megabit speeds at best going to a few thousand. These enormous discrepancies seem to have heavily shifted national averages in the report.

In a statement to Ars Technica, which has followed the broadband report drama closely (including some good analysis last month), BarrierFree COO Jim Gerbig admitted that “There is indeed an error in the Form 477 filings for BarrierFree, and it doesn’t reflect our current level of broadband deployment. A portion of the submission was parsed incorrectly in the upload process.” He claims the government shutdown prevented correction of this issue.

Unfortunately, Chairman Pai, understandably excited to share good news on broadband, already bruited some statistics from the draft report that, if this massively erroneous form were excluded, would be totally incorrect — and incorrect in an unflattering way to the current administration.

Without BarrierFree’s phantom customers, nearly two million more people than reported lack access to fixed broadband — 21.3 versus 19.4 million in Pai’s press release. This is still well below the 26 million from the previous report, but it’s still a major correction. Of 5.6 million newly served rural broadband customers Pai highlights, 2 million were supposedly on BarrierFree.

And a huge reported increase to people on a sub-gigabit but high-speed tier (250/50 Mbps) would have been largely attributable to these non-existent connections — tens of millions of them.

While there is surely good news to share from this report, it seems that the good news the chairman chose to present may in fact not be nearly as good as he claimed.

Activists and government officials alike have questioned the accuracy of previous reports and warned that the incoming one was likely as untrustworthy as those that came before. But this massive single outlier seems like a new and much more avoidable form of inaccuracy.

It seems that in collating and analyzing the forms submitted by ISPs, it would ring a few alarm bells that an ISP with no presence in 2016 would suddenly be serving more than 60 million people with speeds only offered by decades-old competitors. The error is BarrierFree’s to begin with, of course, although I am suspicious of the “parsing” issue blamed by the COO. But surely spotting an error of that magnitude is the FCC’s responsibility.

When contacted for comment, a representative for Chairman Pai’s office said “we are looking into the matter.”

Others were more verbose.

“Free Press’s allegations are troubling,” Commissioner Geoffrey Starks said in a statement. “The FCC’s maps are frequently criticized for being inaccurate and overstating broadband coverage. The maps and deployment data are becoming a repeat offender.”

“Without getting to the bottom of this, the FCC should not proceed with its current draft broadband report. It is the FCC’s job to have accurate data and to make available maps based on it. Without performing that basic function, we are woefully unprepared to make a number of critical policy decisions that will impact the future of our communications infrastructure.”

Commissioner Jessica Rosenworcel, who has spoken out on the broadband report issue recently and been an outspoken critic of the FCC’s policies of late, also called for closer scrutiny:

“The FCC’s draft report concludes that broadband deployment is reasonable and timely across the country. This is hard to believe when millions of Americans have no high-speed service at home. Now there are allegations that the FCC’s numbers in this report may be based on faulty data,” she said in a statement. “This is not good. It absolutely deserves a closer look.”

While the publication of this report was hitherto expected daily, this issue seems likely to push it out by a few weeks at least — and, though it may be too much to hope — could cause the agency to question the basis on which it is built in the first place.