FCC Comments

Based on an extremely non-scientific analysis, there seem to be two astroturf anti net neutrality campaigns, along with one potentially organic net neutrality campaign (CFIF)

Overview

Since late April, the FCC has been taking comments on Docket 17-108 Restoring Internet Freedom. This has garnered over 1M comments, a large portion of which were in the past week, following John Oliver’s call to action on May 7th. The FCC website was so overwhelmed that it went offline for some time immediately after the segment aired.

Content Concentration

There’s mounting evidence that some of these comments aren’t organic users; in fact, more than half of all comments are verbatim repeats of 3 anti net neutrality comments, all found on anti net neutrality websites. The top comments are shown below (some truncated for clarity/spacing)

Or, to make Tufte truly irate, an entirely superfluous ink-heavy graphic

Timing

I think this is the interesting piece. A timeline of the hourly number comments posted for the top three most common comments, alongside a view of the other (apparently organic) comments posted in green.

The green shows a pattern roughly as expected – an initial gap due to the fcc.gov outage followed by a peak the following evening and a decay (with a somewhat circadian rhythm to it), along with a few gaps, likely corresponding to when the fcc website was overloaded/slow.

The red and blue comments (from the Taxpayer Protection Alliance and freeourinternet.org) do not follow this or any reasonable pattern. To quantify it, their posting volume is highly correlated with each other (pearson correlation of 0.7 for the week shown above) and uncorrelated with the baseline ‘Other comments’ (between 0.0 and 0.1).

The behavior of the CFIF message looks almost human-like, at least on this hourly basis. There appear to be gaps/trickles whent he website was having issues and the ramping up and down seems similar in pattern to the baseline ‘Other comments’. It is most correlated with the baseline ‘Other comments’ (correlation coefficient of 0.5), and the hourly autocorrelation coefficient is similar to the ‘Other comments’ (highly autocorrelated at the first and second lags). While the distribution is very different (when the bulk of the comments were made), most of the ‘Other comments’ were made soon after the John Oliver segment.

Conclusion

While the ‘Obama’ and ‘Soros’ messages seem pretty fishy, I woulnd’t be surprised if the CFIF message was genuinely posted by anti net-neutrality advocates (although likely through some intermediary mechanism, like a form on another webiste).

Notes

1. This is entirely unscientific. There’s a decent chance that all of this is incorrect and I make no claims about the accuracy.
2. The ‘sources’ of these comments are guesses, at best, based on a combination of reddit and googling.
3. The FCC is surely tracking more detailed behavior and hopefully will choose to identify which comments are true and which were posted without the listed commenter’s knowledge/permission.
4. The actual reporting being done is way more important than this very high-level summarization with numerous assumptions. I wish they’d note which suspect comments they were following up with (i.e., whether they’re the CFIF or one of the other comments), but either way their actual research is valuable.