Recently got a ruling in my six-year-old (!) FOIA lawsuit against the @US_FDA. Mixed result, but I finally got one interesting document that:
a) implies that a @NYTimes reporter is so in the pocket of the agency that she not only gave them quote approval, but also... [thread] /1 ... allowed @us_fda flacks to vet entire articles, seemingly without the knowledge of higher-ups in the paper.

b) shows that the FDA was willing to fight for six years to keep this fact from the public.

c) demonstrates not just (IMO) egregious abuse of FOIA exemption 5... /2

Here's a quick Benford's Law primer... and an explanation of why deviations from Benford's Law in precinct-level election results isn't an indicator of fraud.

My credentials: Ex-mathematician; not a statistics specialist, but have used Benford's law before. 1/ (And I invite anyone more expert than I, e.g. @jengolbeck, to comment/critique this thread!)

I. What is Benford's Law?

It's a rule that states that in certain distributions of decimal numbers, the first digit is more likely to be 1 than 2, 2 than 3, 3 than 4, etc... 2/

OK, I'm crunching through some precinct-level data and in this thread I'll post some observations that I find interesting as I come across them.

First, less than 20% (19.6%) of the presidential votes were traditional election-day votes. Early voting was 54%, and absentee 26%. Fewer, not less. :)

Day-of voting went to DJT, 61.6% to 38.4%.
Early voting to JRB, 53.1% to 46.9%.
Absentee most lopsided to JRB, 65.2% to 34.8%.

I'm noticing unusually high numbers of provisional balloting in a number of counties; Spalding County is off the charts.

In the age of big data, I can not BELIEVE how bad the COVID-19 statistics are. Pardon a brief twitter rant, but I 'm annoyed that it's amateur hour when it comes to pandemic data published by national and state public health authorities. /1 To track the progress of the pandemic, you want to know, for example, how many people die each day.

So, typically, every day, a public health authority will announce something like: there are 489 new COVID deaths today, May 1. On May 2, they announce 189. 138 on May 3. /2

Thread: Mapping a @Monsanto-loving octopus. Let's start with @ScienceBlogs (whoa... just realized they blocked me!) was a popular blogging site ca. 2008 that was bought out several times and eventually died. Then @science2_0, a nonprofit org, bought the moribund site & rights. Some former bloggers expressed concern about the purchase by @science2_0. Rightly so, and not just because they posted embarrassing Nazi-sympathetic and antisemitic nonsense.

https://twitter.com/JoshRosenau/status/1016456396160884737

For those who asked: here are a few more graphs. One important point: they're not always easy to interpret. @ callouts to famous people (e.g. @realdonaldtrump, @pontifex, @hillaryclinton), get them incorporated into the graph, even though they're not really involved. 1/ Graph 1: Biggest thread from July 11 (the @jacobawohl one from Friday was #2.) More a dogwhistle/distortion than straight misinformation: Pope pushes "New World Order" ideology & open borders. @amike4761 is main actor here. Note inclusion of @pontifex. newspunch.com/pope-francis-u… 2/

@JacobAWohl Happy to explain. I'm a journalism prof interested in, among other things, misinformation. On July 11, I was running a program to monitor a fake news site to see if I could figure out what was responsible for spreading its falsehoods. In this particular case, you were. 1/ @JacobAWohl The misinformation in question was the statement that joining Antifa had just been outlawed and punishable by imprisonment -- an obviously and provably false claim. And in this case, it turns out that you bore the prime responsibility for spreading this lie. 2/

Before @Twitter took them down, I captured a bunch of the #MAGABomber's tweets and did some quickie analysis... the results were pretty interesting. All the following are based upon tweets from @hardrock2016 and @hardrockintlent accounts. 1/ First two plots: Plotting the times of #MAGABomber tweets on a 24-hour radar plot. Note the time when it tends to go silent every day. If the time is set right (or even if we're on GMT), he doesn't sleep when most East Coasters do. 2/