We found that AT&T, Verizon, EarthLink, and CenturyLink disproportionately offered the worst internet deals to neighborhoods that were formerly redlined, whose residents are lower income and have a higher concentration of people of color than other parts of the city.
We examined actual internet offers to more than 800,000 addresses in 38 cities across the country.
People in disadvantaged neighborhoods would be offered plans as high as $100 per megabit per second, while those in more affluent areas that have more White residents and had the best historical redlining scores were offered plans for less than $1 per Mbps.
The ultimate effect of these practices went beyond fairness: Those in disadvantaged neighborhoods were offered speeds so slow that they were denied the ability to participate in remote learning, jobs, and even family connection and recreation that are ubiquitous to modern life.
A recent Pew survey, for example, showed 90 percent of respondents saying the internet has been “essential or important for them personally during the coronavirus outbreak.”
The Markup collected 848,618 internet service offers made to addresses in 38 major U.S. cities.
The top-line finding is that four companies—AT&T, Verizon, EarthLink, and CenturyLink—offered blazing fast internet plans (200 Mbps download speeds or above) at the same price they offered slow internet plans (under 25 Mbps, which is below the Federal Communications Commission’s floor for labeling a connection “broadband”).
Some researchers have looked at broadband availability, and one, the National Digital Inclusion Alliance, published a report saying some internet providers were advertising varying speeds for the same price, which it called “tier flattening.”
Our investigation is the first examination of actual offers to U.S. addresses, which were gathered from the providers’ lookup tools.
Some of the pricing disparities were extreme. CenturyLink’s offers ranged from $100 per megabit per second (for 0.5 Mbps plans) all the way down to 25 cents per megabit per second (for 200 Mbps plans) in all 15 cities served by CenturyLink that The Markup examined.
The federal government does not regulate internet pricing, so providers can charge whatever they want. However, the FCC is currently engaged in rulemaking concerning “digital discrimination” that could potentially govern this type of behavior.
We used automated techniques to enter millions of addresses across 45 major U.S. cities into each of the four internet service providers’ websites and to save the advertised speed and price offers for each address. We collected data between April 15 and Oct. 1, 2022.
Cities that offered uniform speeds (Baltimore, New York, Philadelphia, Boston, Providence, and Wilmington) were filtered out.
The remaining offers in 38 cities were merged with socioeconomic data: median household income and race/ethnicity from the 2019 five-year American Community Survey, and historic redlining grades from the University of Richmond’s Mapping Inequality project.
In 92 percent of the cities we examined, the areas offered the worst deals were lower income. In two-thirds of cities, the worst deals went to the least-White areas. In every city where there was enough data to do the analysis, the worst deals went to areas that had been historically redlined.
The Markup found at least one of these disparities (income, race/ethnicity, or historical redlining grades) in every city except for Boise and Fargo.
In our Github repository, for each of the following cities, you’ll also find a breakdown of speeds offered to a representative sample of addresses in that city, as well as interactive, address-level maps. You will also find summaries about how internet speeds vary across socioeconomic areas, and code that can be used to reproduce our data-intensive investigation:
Based on which company serves the city you’re investigating, look at the interactive, address-level maps we created to help journalists do further reporting:
You can use the search bar to quickly travel to specific addresses or cities. If you know the areas, this will be immediately useful.
However, if you would like an overlay of any socioeconomic factor in our investigation (median household income, the percentage of non-Hispanic White residents in the area, or redlining grades) we can produce them by request.
These maps should be viewed alongside summaries of how speeds vary across each city and between areas.
For a detailed breakdown showing precisely how we conducted our analysis, along with the full results, check out our methodology.
If you’re interested in using our data for your own coverage but need some more help, here are a few examples of how journalists have used our data in their own stories:
This story has been updated to include examples of local journalism using our data. As we add new articles, we’ll update the date here.
By Leon Yin and Aaron Sankin
First published here
Photo by Alina Grubnyak on Unsplash