Creating the National Broadband Map for $3.5 Million Instead of $350 Million

There has been a bit of hub-bub about the provision in the stimulus (ARRA) plan to map broadband in the United States and the $350 million price tag. Specifically the statute calls for “$350,000,000 shall establish the State Broadband Data and Development Grant program, as authorized by Public Law 110-385 *and* for the development and maintenance of a national broadband inventory map as authorized by division B of this Act.”

In a response to an All Point Blog podcast on the topic Archie Belaney said he thought FortiusOne could accomplish the same task for $3.5 million. With gauntlet on the floor it got me thinking about how you would actually pull it off in an economical fashion. It has been a few years since I’ve been deep into the weeds of broadband mapping, so some of my thoughts may not be completely aligned with the status quo.

One of the big challenges, often missed, is that many broadband providers do not know where their fiber is. Large amounts of data still resides in paper maps – in a 2003 court case involving the Maine Public Utilities Commission an expert witness for Verizon states that standard operating procedure for restoring a failed line is to locate the lead engineer for the region, consult paper maps, manually identify an alternate route, and send technicians to wire jump around the outage. The point being that in many cases there is not a nice clean digital database of broadband deployments and routes.

So to build a national broadband map one of the biggest challenges is data collection and aggregation. Therefore the tools in GeoCommons to upload existing digital data is going to be insufficient. To fill this gap there are two key capabilities required digitizing paper maps and collecting data from the field. To solve both problems I think there is a good bit we can borrow from the OpenStreetMap (OSM) project. Broadband fiber lines all have to follow right-of-ways along streets, railroads, pipelines etc. The late 90′s saw a boom in companies whose business models were just finding new right-of-ways to lay fiber across. Mapping these right-of-ways is no different than the streets, railroads, cycle ways and footpaths mapped in OSM.

Since the goal here is to map where broadband is deployed so we can better gauge “universal access” the accuracy of commercial GPS should more than get the job done. Data collection could easily follow the OSM model and a server could be set up inexpensively and leverage any GPS device for data collection. This covers one use case where the data needs to be collected from the field, but is likely to only be a small percentage of data that needs to be created. The larger project is likely to be digitizing paper maps, which I believe is yet another task that can be easily crowdsourced (even if the crowd is not the public).

A good deal of the data in OSM does not come from GPS data from the field but instead from users tracing roads from satellite imagery. This same approach could be used to easily digitize existing paper maps of broadband deployments. To enable this you could leverage Chippy and Schuyler’s Map Warper to pull paper maps into OSM, allow them to be traced and entered into a common system.

I think the combination of OSM and Map Warper would cover the big data creation and conversion issues. You could use GeoCommons to allow the various stakeholders to contribute their existing spatial data on broadband deployments to the system. You could also allow stakeholders to create maps of broadband coverage identify identify need based gaps by overlaying demographic data. Below is an example of cable broadband lines in Vermont overlaid with population count with income between $10-20k.

When you filter the data to locations with more than 150 people in the $10-20k range you can begin to see the mismatch between broadband service and low income population.

While I’m not sure of the exact cost breakdown for what is proposed above, you could deploy everything to EC2 and probably cover software/hardware for half of the 3.5 million. Then cover the engineering labor needed with the other half. Data collection/migration costs would be the biggest variable with lots of possible options – Americorps, volunteers, offshoring (probably unpopular), or leveraging various job works programs. The good news is that it would not require a very technical workforce to accomplish the tasks. It would be a cool project but sadly we’ll probably spend the whole $350 million and still not have a decent product to show for it. On the upside some of the concepts proposed are already being adopted by a few telcos who are trying to crowdsource their coverage maps. RidgeviewTel created a Web site where the public can drop points for where they can’t receive broadband service.

Regardless of how it happens once the data is aggregated we could make maps like these for the entire country:

Or do calculations of broadband deployment concentrations like this (also NYC):