Wouldn't it be easier to figure out how many people do have broadband and subtract from the population to get a rough guess? ISP's report subscription numbers. It won't be super accurate but maybe that is sufficient.
They are looking for the number of people/households/whatever who don't even have an option for broadband, not just the number who don't subscribe.
Would this handle multi-person households?
Incorporate avg number of people per house...?
Do we really have no record of how many houses are in the US? I hate how absolutely inefficient government is
We have excellent data on this
Then I don't see the issue of why we can't calculate like the parent said to figure out how many houses don't have service. If we know how many houses there are and the ISPs know the homes they serve, then you can figure out the difference.