Nate Silver’s data journalism site FiveThirtyEight published an article yesterday called “The Worst Internet in America.” The article focuses on Saguache County, Colorado, a beautiful, rural, isolated county, the border of which just so happens to fall about 6 miles south of my house.
The piece starts off by saying:
FiveThirtyEight analyzed every county’s broadband usage using data from researchers at the University of Iowa and Arizona State University and found that Saguache was at the bottom. Only 5.6 percent of adults were estimated to have broadband.
So, it would appear, according to the data, fewer adults use broadband in Saguache than anywhere else.
From this, FiveThirtyEight crafted a piece that is a majestic example of the logical fallacy of ignoratio elenchi, which is Latin for drawing an irrelevant conclusion. Or in this case, conclusions.
First, and perhaps most critically, the headline is flat-out wrong. Saguache County does not have “the worst internet in America.” According to the article itself, anyone in Saguache can get 12 Mbps for $90. And—while not cited in the article itself—if you’re willing to pay more, you can get up to 60 Mbps in Saguache County. That ain’t exactly Google Fiber, and it isn’t super cheap, but it’s not bad for a rugged, mountainous county that’s nearly twice the size of Rhode Island and with a population density of less than one person per square mile. According to this article, Saguache’s available broadband speed is only moderately worse than the national average, and more than ten times better than the internet speed in Northeastern Arizona, which, if they had bothered to check, is actually the worst in the country .
If only five people drive cars in a city that doesn’t mean that the city has the slowest cars in the country. The conclusion has no relationship to the premise whatsoever. As with the data in the article and its conclusion, they are two totally unrelated variables.
The article makes no attempt to compare actual internet speeds in Saguache County to those in comparable rural areas. It just made a single, flip comment about how you could get way cheaper, way better internet in New York City. As if we should somehow be surprised that the technology infrastructure in a county with less than one person per square mile lags behind those in a city with population of 27,000 people per square mile.
Second, the authors seem to discount the possibility that some of the residents of Saguache County might actually choose to live there not in spite of its lack of high-quality of internet, but because of it. Believe it or not, some people might actually enjoy going to beautiful places where they can disconnect. One of the main reasons Saguache falls at such an extreme end of the internet-connectedness distribution for adults is that a significant percentage of the population in Crestone and surrounding areas are associated with the various Buddhist retreats and religious get-aways. There are nearly twenty different religious centers there in a county of 5,000 people (http://saguache.org/spiritual/crestone.html). For those Zen monks, and many others, that disconnectedness is a very conscious decision.
Third, the article chose Saguache because it had the lowest percentage of adults who were connected to the internet, but the crux of the article seemed to be how the disconnectedness hurts educational opportunities for youth. That might be true, but, again, it’s a non sequitor. If that were the focus, FiveThirtyEight should have sought out a data set related to internet access for those under the age of 18 (or age 21). Instead, it picked Saguache for one metric (few adults who use broadband) and then wrote a narrative that addresses a very different issue (educational opportunities for those without broadband).
“Data journalism” can be a great thing, but only if the data actually has some nexus to the journalism. Held to this standard, this article failed miserably.