The Penn State Broadband Study

Penn State conducted an intensive study of broadband in rural Pennsylvania. The study was funded by the Center for Rural Pennsylvania, a legislative agency of the Pennsylvania General Assembly.  The results will surprise nobody who works with rural broadband and the study concluded that actual broadband speeds are significantly slower than the speeds reported by the ISPs to the FCC.

The study concluded that there was not one rural county in the state where more than 50% of residents actually achieve the 25/3 Mbps that the FCC has defined as broadband. The study came to these conclusions by conducting more than 11 million speed tests. Residents voluntarily provided an additional 15 million speed test results.

These results are similar to what’s been reported by Microsoft – they measure the actual speeds at which millions of customers download Microsoft software every month. Microsoft says such tests are the best measure of real broadband speeds and that roughly half of all broadband connection in the country are done at speeds slower than the definition of broadband.

Some of the Penn State results are dramatic. For example, in Westmoreland County the FCC maps show the whole county has access to 25/3 Mbps broadband and yet the average download speed for the county was only 12.3 Mbps. Allegheny County also shows 100% broadband coverage on the FCC maps and yet the average download speed in the County is only 20 Mbps.

The study further showed that the difference between actual and reported speeds have been widening since 2014. That’s likely to mean that the FCC maps are showing improvements that aren’t really happening in the rural networks.

I have to point out, in the FCC’s favor, that households don’t always buy faster broadband when it’s available – many households continue to purchase older, slower DSL to save money. However, this phenomenon can’t come close explaining the results in Westmoreland County, where the actual speeds are only 12 Mbps – half the FCC’s definition of broadband. A more likely explanation is that the maps for the County show broadband available in rural areas where actual DSL speeds are only a few Mbps.

CCG helps our clients conduct similar tests on a smaller scale and we’ve seen similar results all across the country. The FCC maps are often pure fantasy. We routinely find rural areas that supposedly have fast broadband where there is no broadband. We often study county seats that supposedly have fast data speeds and yet where actual speed tests show something far slower. The speeds on the FCC maps come from data that is self-reported by ISPs, and some of the ISPs clearly have reasons to overreport the available speeds.

What is really irksome is that the FCC knows all of this already. They know that ISP reported broadband speeds are overstated, and yet the FCC compiles the faulty data and makes policy decisions based upon garbage data. The FCC’s recently published their 2019 Broadband Deployment Report which concluded that broadband is being deployed in the US on a reasonable and timely basis. In my opinion, that conclusion borders on fraud since the FCC knows that much of the data used to reach that conclusion is wrong. The real broadband situation in rural America is much more like what is being reported by Penn State and Microsoft. Rural residents in places like Allegheny County, Pennsylvania should be incensed that the FCC is telling the world that their broadband is up to snuff.

The FCC is starting a multi-year process to ‘improve’ the broadband maps – but this will just push the problem a few years into the future. The fact is that it’s almost impossible to map real broadband speeds in rural America. How can you map broadband speeds when real networks in rural America are in lousy shape? How can you map broadband speeds when two neighbors can experience drastically different broadband speeds due to the nuances in their copper wires? The big telcos have neglected maintenance on copper networks for decades and it’s no surprise that broadband speeds vary widely even within a neighborhood.

The best solution is to throw the maps away. The fact is that every place served by copper ought to be considered as underserved, and locations more than a few miles from a DSLAM ought to be considered as unserved. We need to stop pretending that we can somehow make a realistic map of broadband speed availability – the proposed new mapping might be a little better, but it can never be accurate. Every ISP technician that works in the field will tell you how ridiculous it is to try to map rural broadband speeds.

We need to face facts and recognize that we’re going to have these same issues until rural America gets fiber. There are now enough places in rural America with fiber to show it can be done. The FCC’s ACAM program has shown that fiber can work if there are subsidies to help with the construction costs. We’ve understood this for more than a century since we built the rural electric grids. But we probably can’t fix the problem until we’re honest about the scope of poor broadband. I have big doubts that this FCC is ever going to acknowledge that the real state of broadband is the one highlighted by this study.

Is the FCC Really Solving the Digital Divide?

The FCC recently released the 2019 Broadband Deployment Report, with the subtitle: Digital Divide Narrowing Substantially. Chairman Pai is highlighting several facts that he says demonstrate that more households now have access to fast broadband. The report highlights rural fiber projects and other efforts that are closing the digital divide. The FCC concludes that broadband is being deployed on a reasonable and timely basis – a determination they are required to make every year by Congressional mandate. If the FCC ever concludes that broadband is not being deployed fast enough, they are required by law to rectify the situation.

To give the FCC some credit, there is a substantial amount of rural fiber being constructed – mostly from the ACAM funds being provided to small telephone companies with some other fiber being deployed via rural broadband grants. Just to provide an example, two years ago Otter Tail County Minnesota had no fiber-to-the-premise. Since then the northern half of the county is seeing fiber deployed from several telephone companies. This kind of fiber expansion is great news to rural counties, but counties like Otter Tail are now wondering how to upgrade the rest of their county.

Unfortunately, this FCC has zero credibility on the issue. The 2018 Broadband Deployment Report reached the same conclusion, but it turns out that there was a huge reporting error in the data supporting that report where the ISP, Barrier Free, had erroneously reported that they had deployed fiber to 62 million residents in New York. Even after the FCC recently corrected for that huge error they still kept the original conclusion. This raises a question about what defines ‘reasonable and timely deployment of broadband’ if having fiber to 52 million fewer people doesn’t change the answer.

Anybody who works with rural broadband knows that the FCC databases are full of holes. The FCC statistics come from the data that ISPs report to the FCC each year about their broadband deployment. In many cases, ISPs exaggerate broadband speeds and report marketing speeds instead of actual speeds. The reporting system also contains a huge logical flaw in that if a census block has only one customer with fast broadband, the whole census block is assumed to have that speed.

I work with numerous rural counties where broadband is still largely non-existent outside of the county seat, and yet the FCC maps routinely show swaths of broadband availability in many rural counties where it doesn’t exist.

Researchers at Penn State recently looked at broadband coverage across rural Pennsylvania and found that the FCC maps grossly overstate the availability of broadband for huge parts of the state. Anybody who has followed the history of broadband in Pennsylvania already understands this. Years ago, Verizon reneged on a deal to introduce DSL everywhere – a promise made in exchange for becoming deregulated. Verizon ended up ignoring most of the rural parts of the state.

Microsoft has blown an even bigger hole in the FCC claims. Microsoft is in an interesting position in that customers in every corner of the country ask for online upgrades for Windows and Microsoft Office. Microsoft is able to measure the actual speed of customer download for tens of millions of upgrades every quarter. Microsoft reports that almost half of all downloads of their software is done at speeds that are slower than the FCC’s definition of broadband of 25/3 Mbps. Measuring a big download is the ultimate test of broadband speeds since ISPs often boost download speeds for the first minute or two to give the impression they have fast broadband (and to fool speed tests). Longer downloads show the real speeds. Admittedly some of Microsoft’s findings are due to households that subscribe to slower broadband to save money, but the Microsoft data still shows that a huge number of ISP connections underperform. The Microsoft figures are also understated since they don’t include the many millions of households that can’t download software since they have no access to home broadband.

The FCC is voting this week to undertake a new mapping program to better define real broadband speeds. I’m guessing that effort will take at least a few years, giving the FCC more time to hide behind bad data. Even with a new mapping process, the data is still going to have many problems if it’s self-reported by the ISPs. I’m sure any new mapping effort will be an improvement, but I don’t hold out any hopes that the FCC will interpret better data to mean that broadband deployment is lagging.