Thread by @wiczipedia

THREAD. A few quick takeaways from today's new info ops dataset that @TwitterPolicy released, including 418 #Russia-connected accounts. A collab brought to you by me and @APavliuc.

More on the dataset: blog.twitter.com/en_us/topics/c…

These accounts were created mostly in 2014-2015, with a few outliers in 2016-2018.

The oldest account was created in 2010. According to its location data, it was based in Kaliningrad, Russia. (2/)

These were low follower, low engagement accounts.

Average # of followers = 384 (compared to 1664 in last year's dataset)
Average # of replies = 0.04 per tweet
Average # of likes = 0.62
Average # of RTs = 0.39

So, uh, what exactly were these accounts doing?

(3/)

They were mostly amplifying other accounts. In this image, the green nodes are the only original content. All the pink stuff is RTs. (4/)

When you cross reference the content with the year the accounts were created, you get something interesting: almost all that original content came from accounts created in 2017 (green) and 2018 (pink). Strange. (5/)

Only 9 of the accounts had over 5000 followers:

One was a Syria news account (Inside_Syria) with 19,845 followers.

In 2nd and 3rd place: accounts that use hashtags like "DeportAllMoslems" and "InfidelForLife" in their bios

4th place: an "unapologetically black" account

(6/)

The lower-powered accts in the top 9 had pretty anodyne bios ("Outgoing, social, open, does not like being alone" 😂) and low tweet/engagement counts which makes me think they were lying in wait to be activated for another use.

(7/)

Across the dataset, the following terms came in among the top 100 (along with a bunch of common words). In order, from more to less frequent:

Trump
Islam
MAGA
ReleaseTheMemo
Obama
Hillary
America
Clinton
IslamIsTheProblem
Muslim
ISIS
Qanon
StopImportingIslam
BanShariaLaw

(8/)

So, as usual, we're seeing Russia weaponize preexisting fissures in our society, not create new ones. This dataset seems to use mostly right-leaning issues, but I'm sure there is some content on the left in the Syria and BLM accounts that didn't show up in the top 100 terms. (9/)

The accounts were not particularly effective at creating new audiences/generating engagement, but that wasn't really their goal; it was to amplify hot-button issues and create the impression of support for these discussions through retweets. (10/)

That's all for now! Big ups to @APavliuc, who helped answer my burning questions and was my partner in Nerdville 🤓 Let us know what else you'd like us to look into and we'll do some digging over the weekend. (11/11)

Whoops, and that's @Policy. The 10pm hour is not really my prime tweeting time. Or thinking time. I am a curmudgeonly old person and I like to go to sleep early.