For the most part, the ReConnect program rules are well constructed and put an emphasis on the right metrics: those areas of the country most expensive to serve and least well-served. In modeling done by the FCC and by those of us engaged in the construction of rural networks, the network economics at a population density of fewer than eight homes per mile becomes difficult without public support. RUS’s approach, by giving points to areas even less densely populated goes a step further and prioritizes the most expensive areas, those with a population density of less than six people per square mile. RUS gets it right when it comes to prioritizing the most expensive areas to serve.
Unfortunately, RUS trips over the first step of the program, which is to identify those areas of the country that are unserved. Congress provided a temporary definition of unserved, those areas where 90% of the households lack the availability of 10/1 Mbps service. That’s also the right metric, because it prioritizes the least well-served areas of the country. Were data available to identify this exact definition of unserved areas, applicants could map out proposed territories with confidence. But, as everybody knows, no such data exists. Not public data nor private data. And yet this is the one place in the ReConnect process where RUS asks applicants for perfection. RUS has stated that even a single household shown in a challenge to have 10/1 Mbps service will disqualify an otherwise perfect application. This is the antithesis of Congress’s 90% statutory language. A threshold of 1 or 0 is a binary calculation, not 90% coverage.
Consider the following scenario.
You examine the most recent publicly available broadband data released by the FCC.
You choose to apply for a grant only in an area where no provider of service claims to advertise the availability of 10/1 Mbps service. You intend to build a fiber network to 1,000 unserved households, as well as schools, libraries, health care facilities and critical community facilities.
You contact the service providers in the area to seek any information about where they provide 10/1 Mbps service, and they decline to give you information for understandable competitive reasons.
You spend hundreds of hours preparing an application, which includes business planning, an income statement, cash flows and balance sheet, TIER and equity projections, and a network diagram prepared by a qualified engineer. You get board approval for such a project and contact your lenders about the availability of financing. You reach out to the community to get twenty farms and fifteen businesses to agree to take future service. You complete all of the other dozens of administrative tasks requested by RUS. Your scoring sheet demonstrates a stellar application of greater than 130 points.
RUS publishes your application area for challenge and an internet service provider contacts the RUS and says that they placed a DSLAM or some radio equipment in the area in the past year and now have service available to a few homes out of the 1,000 in the application area, service that would not have shown up in any publicly available data, because the publicly available data is 18 months old.
A General Field Rep from RUS tests the service and confirms that for at least one location, it is possible to receive 10/1 Mbps service.
The application is disqualified with no recourse other than to try again next year.
Does that scenario seem unlikely? It is the most likely result for every single applicant to the first tranche of grant applications. It is a direct consequence of the unavailability of perfect information and the RUS requirement for perfect information. I understand that RUS uses a similar process for its Community Connect program, but the scale of the ReConnect program is vastly different and will produce different behavior by challengers. There is far more at stake in the ReConnect program for rural America.
Fortunately, this problem is entirely avoidable with one simple decision by RUS to offer applicants a safe harbor against disqualification for otherwise acceptable applications. In fact, the RUS has already taken a similar approach in other parts of the scoring by using national standard data (HiFLD data for critical, education and healthcare facilities). Use of that data in scoring is for all intents and purposes a safe harbor and unchallengeable. Why not extend that level of transparency to the most critical part of the application, the lack of broadband availability?
We propose the following safe harbor procedure, which would be to every applicant’s advantage and no one’s disadvantage:
An applicant uses the FCC’s most recent publicly available 477 data, or similar state government data, and chooses a proposed service area where no ISP indicates the availability of 10/1 Mbps service. The safe harbor procedures apply.
An applicant uses the FCC’s most recent publicly available 477 data and chooses a proposed service area where an ISP claims availability of 10/1 Mbps service, but the applicant has reason to believe that the census block is not served in its entirety by 10/1 Mbps service. The applicant contacts the ISP and seek clarification as to which parts of the census blocks have 10/1 Mbps service and which do not. The safe harbor procedures apply whether the ISP provides such information or not, and the applicant documents its request.
The safe harbor does not protect an applicant from challenge, only from disqualification. In a safe harbor scenario, if an ISP can demonstrate to RUS that households have available to them 10/1 Mbps service then those households and a proportionate amount of the grant request are removed from the application. The modified application would be processed under the current rules.
Safe harbors are particularly useful mechanisms when data is unavailable or extraordinarily difficult to obtain. I implore the decisionmakers at the USDA to remove the uncertainty from the RUS ReConnect program before tens of thousands of hours are spent in preparing applications for this valuable and otherwise well-constructed program.