When the Federal Communications Commission took up a last year, the reaction from the public was overwhelming: 22 million comments poured into the agency.
What was less clear was what the public actually said.
A number of researchers and data sleuths have concluded that many of the comments, perhaps millions of them, were “bots,” automated submissions spat at the agency, probably from people on both sides of the issue.
A from Stanford University’s Center for Internet and Society set out to examine the non-bot comments, by focusing on about 800,000 unique submissions that were not automatically generated.
The analysis also includes a manual review of 1,000 of those comments, and concludes that 99.7 percent of them, all but three, were opposed to the FCC’s action.
The analysis is meant to help the public “understand the volume and sentiment of the comments to the FCC about its net neutrality repeal.”
The FCC’s order, championed by Republican Chairman Ajit Pai, reversed an earlier policy set during the Obama era that was meant to protect net neutrality, generally defined as ensuring the free flow of content over the internet.
Pai’s order specifically that forbade internet services providers from blocking or throttling content or engaging in “paid prioritization,” assigning content to fast lanes based on monetary arrangements.
Critics of Pai’s order in the K-12 community fear it could to schools, as deep-pocketed content providers pay for faster access to customers. Some education companies also fear that their ability to deliver affordable content to teachers and students will diminish because of the ruling.
Pai, who was appointed to the chairman’s post by President Trump, argued that such fears are overblown, suggesting that the Obama-era policy amounted to over-regulation that stifled internet providers’ ability to innovate.
The Stanford analysis was authored by , a media and strategy fellow at the center. Singel founded , an online engagement platform for publishers, and he opposed Pai’s order, arguing that it would have hurt startup companies like his.
Singel’s analysis builds on earlier by data scientist Jeff Kao, who used machine-learning methods to break down the flood of FCC comments to determine which were virtually identical with each other. That process allowed Kao was to filter out 800,000 “unique” comments, which were outliers in terms of their wording and structure.
Not all of the non-unique comments sent to the FCC were fake, notes Singel. Many people submitted form letters created by interest groups, and Kao’s research confirmed that those individuals in fact acted to submit those comments.
The FCC has not responded to a request for comment on the Stanford center’s findings.
Singel also breaks down the 800,000 unique comments on a political map, matching them with congressional districts and states.
He concludes that while support for net neutrality was strongest among traditionally Democratic-learning urban districts, GOP districts backed the concept too. The average number of unique comments from congressional districts was 1,489. In those represented by Democrats, it was 1,846, while in Republican ones, it was 1,202.
(Those comments are from a pool of 646,000 that Singel geo-coded to congressional districts. They are from the same pool of unique comments that are 99.7 percent percent pro-net neutrality, meaning they are almost all pro-net neutrality, he said in an e-mail.)
Singel also found that many commenters had a nuanced understanding of federal regulations tied to net neutrality, such as Title II of the federal Communications Act. He also said his data indicates that rural commenters were deeply concerned about losing net neutrality protections.
“Rural commenters who have slow and expensive internet service with few, if any, choices of providers,” he writes, “express strong concerns about the ability of their provider to unfairly interfere with their choices about what they do online.”
See also: