Thread by @benimmo

Looking into Twitter traffic around the #IowaCaucus mess.

Lots of Americans trolling each other, and some accusing others of being Russians.

Always useful to look at the evidence at times like this.

(Hashtags below, not wanting to add to the trends...)

This hashtag, about Mayor Pete, got the most mentions of the ones I looked at - 105k earlier this evening (over 120k now).

Some people were saying it must have been driven by Russian bots, but...

... these were the first two tweets. Both from verified users.

This was the most-retweeted post on the hashtag, calling for Democratic unity, but also suspecting Russian propaganda accounts.

Yes, division was one of the Russian IRA's goals, but there's no proof of it here that I've seen yet.

The sad fact is that, on Twitter at least, America is so polarised you don't need Russians to feed division. Americans are doing it.

Here's the next hashtag. A bit over 50k mentions, but spiked on January 31.

The most-retweeted posts on that hashtag are interesting: Jill Stein, Dan Scavino, and "BernieWon2016".

This one really didn't go well. Under 5k tweets. That's barely a pimple on the face of Twitter.

Again, a mixture of pro-Trump, pro-Sanders and "Russians are driving this" comments.

Great little tool here from my colleagues at @Graphika_NYC. Summary stats for this traffic.

"CTM" is the coefficient of traffic manipulation. A normal score would be 12 or under. Heavily gamed and bot-manipulated traffic would usually score 24+.

12.8 looks pretty organic.

Definitely some active posters in the mix, but the number tails off sharply (in serious bot spikes, I've seen dozens of accounts that posted the same hashtag 200+ times each in a couple of hours).

Again, doesn't look like the whole flow was massively gamed at this stage.

Upsum: this doesn't look like "Russian bot" activity.

It looks like polarised American users jumping on a chaotic moment to push hyper-partisan lines.

Not a great outlook for the rest of the year.