Last week I was intrigued when this paper came out: Trust, tribalism and tweets: has political polarization made science a “wedge issue”?. In it, "Helmuth and his Northeastern colleagues analyzed the Twitter accounts of U.S. senators to see which legislators followed research-oriented science organizations, including those covering global warming. Democrats, they found, were three times more likely than Republicans to follow them, leading the researchers to note that “overt interest in science may now primarily be a ‘Democrat’ value.”" Interesting approach, I thought. I was very pleased to learn that the data sets for the article are available at Northwestern University's data repository (yay open science!).
Then for fun I thought I'd check the datasets to see whether any US senators followed Edward Snowden. None did! Then I realized that's 'cause the data sets are from February 2015, and @Snowden didn't appear on twitter until the fall of 2015! Doh!
Now I'm on a rabbit hunt to figure out how to compare @Snowden's 2.12 million followers to see if any of them are US senators. Since I had the list of senators from the aforementioned data sets, I thought it would be easiest to just get a list of all @Snowden's followers and compare the two. Surely somebody has a utility that will allow me to download all 2.12 million followers, right? Not so much :-(
I started with something in php I found on Github that sounded absolutely perfect, but it flat out wouldn't work, and so far no feedback from the developer. Next checked with Nick Ruest to see if twarc might be able to do it, and he suggested I look at tweepy (python) instead. That looked really promising, except I couldn't figure out how to work around the rate limits imposed by Twitter over gathering such a large follower list. Next up, a random kind stranger suggested a program written in ruby called simply "t". I now LOVE this program, in part because it's very simple and very well documented. But it still falls short because of the rate limits. :-( And then my knight in shining armour, Ed Summers, came through with some real life working examples in tweepy! I'm now 3 hours in to gathering the full list of @Snowden's followers, and napkin math suggests I'm only halfway though. Fingers crossed for when I come back in on Monday!
I'd already spent too much time on this, though I've learned tons along the way, and will be able to utilize a lot of this in the future. But I still wanted to find out which US senators followed @snowden. Back to t and I find the command "t does_follow". Hell, there's only 100 senators, so I ended up doing it manually, which in the end only took about 10-15 minutes. And here's what I got for each and every one:
ppival$ t does_follow chriscoons snowden
No, @chriscoons does not follow @snowden.
Talk about an echo chamber! I guess if he's still officially a traitor it looks bad for a senator to be acknowledging him? Sure, if he says something useful I'm sure someone would share it with the VIP, but c'mon, you can't even hear what he has to say or acknowledge his existence by following him on twitter? That's pretty lame, IMHO. btw, neither Hillary nor Donald follow @snowden either. Neither do Hillary or Donald follow each other. Whatever.
Thanks also to John Brosz for helping me work though code, and Andrew Pasterfield for the same.