Network Connection Issues
A partial defense of a broadband boondoggle
Scott Alexander’s most recent post is about the shaky assumptions involved in thinking it’s easy to redirect foreign aid spending into equivalently beneficial domestic programs. It is a good post and I endorse it. But early on he includes this comment:
From there would go to the same kinds of programs that the rest of the pot goes to - like the Broadband Equity And Deployment Program, a $42 billion effort to give rural Americans Internet which, after endless delays, has failed to connect a single rural American.
And then goes on to refer to it multiple times as a boondoggle.
Partial self-dox: I work at an organization that is advising several state and local governments on the BEAD program. I have noticed that the BEAD program has been catching a lot of flack recently, so I thought I would take this chance to explain what is actually behind those delays.
Say you have $42.5 billion to spend on connecting American households that don't have access to broadband (we’re taking for granted that you think this goal is worth the money, though obviously people can debate this. Context: o3 mini estimates that the US spent about $7 billion in 2025 dollars to connect rural households to telephone service. So this program is essentially valuing the internet at 6x the value of a phone line, which doesn’t seem unreasonable to me). How do you decide where that money goes? One option is just to give the money to the states and let them figure it out. But dividing up the pot by population would have left urban states with too much money and rural states with too little. And then how would the states have figured out who to give the money to? If they spun up a grant program for ISPs, and Comcast came to them with a proposal saying they would take $10 million dollars to build out a line to this list of 1,000 addresses, how would the state know that those 1,000 households are actually the ones in need? A lot of people on both the government and industry side were concerned that without any oversight, ISPs would just use grant money to subsidize their expansion into territory already covered by another ISP (something referred to as ‘overbuilding’). Then, the government would have spent billions without actually bringing broadband to any unconnected households.
So, you need some way to know which households aren’t connected to broadband so that you can a) fairly distribute money between the states and b) make sure ISPs only get money if they plan to serve households that genuinely need it. The problem: when the BEAD program started, we didn’t have good data on this. At the time, the standard for broadband mapping was based on Form 477 data, a biannual report that ISPs made to the FCC listing out the census blocks where they provided service. But this data was broadly acknowledged to have all sorts of issues, including the fact that ISPs regularly fudged these numbers, leading to between 3.5% and 11% of households being misclassified as served when they actually had no broadband access (a good summary of some of the issues with Form 477 can be found in this CRS report. For those who want to check behind me, the most critical analysis of 477 is here, while while a more friendly analysis can be found here).
What this all means is that if the government had decided to move quickly with the BEAD program right after the legislation was passed, it would have meant divvying the money up between states according to Form 477 data, then having the states spin up grant programs that used that data to determine which areas would be eligible for grants. To be clear, this wouldn't have been a total disaster. Most of the money would still have gone to households that were unconnected, but there could have been a significant number (4-13 million) of households wrongly ruled ineligible for grants because of data quality issues.
Because of this, the government decided to start from scratch and create a new broadband data collection process and a new map of all households in the country linked to the broadband service available at each address. Creating this map took time (arguably too much, but my point is just to explain what was actually going on during this period):
March 2020 - The bill directing the FCC to create a national broadband map was passed
July 2021 - The FCC issued a request for proposals to solicit bids for actually building the map and the underlying fabric of addresses it would be based on (you would think the federal government would already have a list of all the addresses in the country just lying around, but you would be wrong).
November 2021 - The IIJA was passed, funding the BEAD program
December 2021 - The FCC awarded the mapping contract, then were delayed when (of course) the loser sued the government.
November 2022 - The first version of the map was released, which kicked off a multi-stage validation process where different organizations could challenge the data for inaccuracies (e.g. pointing out that certain “houses” were actually just barns, or alternatively pointing out houses that should be on the map but were left off because state tax records and 911 databases have a lot more errors than you’d expect).
June 2023 - The maps were finalized to the point that the NTIA felt comfortable using them to determine how many dollars each state should get.
December 2023 – The states submitted proposals to the NTIA detailing how they plan to run their grant program and disburse their BEAD money.
March-ish 2024 - Once the NTIA approved each state’s plan, the states opened their own challenge process to do another round of data cleaning (mostly ISPs reporting new areas they were building to that weren’t yet represented, with a small number of public benefit operations like yours truly trying to find places where ISPs were fudging how many households they served).
June-ish 2024 - These challenge processes wrap up and states submitted their results to the NTIA for approval.
November-ish 2024 – NTIA approves the states’ plans, and states open their BEAD grant portals for submissions from ISPs.
December 2024 - States finish accepting their first batch of applications
Now – States are currently in the process of scoring their first batch of applications, with winners to be announced soon
Late 2025 – construction begins on the first BEAD projects once money has been awarded and ISPs get their own ducks in a row to start new builds.
This makes about four years from BEAD’s passage to the construction of the first funded projects. Of that period, about one year is accounted for by the time it took to make the national broadband map, one year for the map to be cleaned and for states to create a plan for how they will run their grant programs, one year for the states’ plans to be approved and for them to run their own cleaning process, and one year for grant applications to be awarded and for ISPs to spin up their own construction projects. Four years definitely feels like too long, and there are plenty of places where this timeline could have been condensed, but hopefully this gives a sense for why things actually took four years.
I should clarify that there are other issues with the BEAD program on top of this. For one thing, the new map we’re using is definitely better than Form 477 data, but they still haven’t solved the underlying problem of being based on self-reporting from ISPs with no outside validation. There is at least a process now to submit challenges to the map if someone discovers that an ISP has misreported their coverage, but there is nothing allowing the accuracy of this data to be verified at scale. In practice this is not exactly the end of the world. The error rate on these maps is probably down to the order of 1%, but it is a bit disappointing that after all of this time and investment we still didn’t solve one of the core problems we had faced with the old data.
And completely separate from the delays, the BEAD program is also facing problems due to its extremely onerous reporting requirements. The administrative burden of complying with this grant has discouraged a lot of even major ISPs from participating. We won’t know for sure until the first sets of awards come out, but it seems very likely that there will be lots of unserved locations that no ISP proposes to build to because it’s just too much trouble. This leads me to two other quick notes:
First, I have seen a lot of complaints about BEAD being delayed because of DEI. I don’t think this is well-founded. I’m certainly no cheerleader for DEI, but to all of us in this space, it is obvious that the main cause of delays has been the multi-step process of doing data collection for the new map, validating the map, and then spinning up grant programs within each state. To the extent DEI/labor requirements are an issue, it will be because they raise the costs and obligations of ISPs who want to get access to this money. Since it seems like a lot of ISPs are already on the fence about this, I could see some of these requirements as being a meaningful factor reducing ISP participation in the program (though I’d hesitate to estimate how big of an effect this will have, and with Trump in power now the question may quickly become moot). Importantly, though, this is different from claiming that DEI has impacted BEAD by causing delays. That seems very clearly wrong from my vantage point.
The other issue is Starlink. A lot of people seem to think BEAD should have been used to just buy everybody in rural America a Starlink terminal. I like Starlink, but from the reading I have done I don't think that plan was ever feasible. The current estimate for the number of unserved households in America is 7.5 million. Starlink does not have capacity to serve 7.5 million households. In 2023 the FCC determined they couldn’t even support 10% of that.
I wish this were not true! It would be great if we could just use Starlink to solve the rural broadband problem. It is definitely much cheaper than building out fiber or cable to every remote household. But there are technological constraints on the bandwidth of satellite internet that make me deeply skeptical this would have worked. I am not a technical expert and I know it’s always risky to bet against Elon on something like this, but I just don't think satellites were ever going to be a broadband panacea.
That being said, I am in favor of recent changes to BEAD that open up the program to satellite providers if there are no other ISPs that are willing to fund areas for a reasonable cost. This strikes me as the right compromise: get fiber and cable to as many houses as you can, and use satellite for all the really expensive, hard-to-reach places that can’t otherwise be served economically.