Twitter’s update on retrospective review of Russian-related 2016 election activity

by Chris Zappone

Twitter recently IDed 50,258 automated accounts as Russian-linked and Tweeting election-related content in 2106.

However, since Russia information war strategy is not just about engineered outcomes, but rather the ideas that can be laundered across domains, it’s likely Twitter’s count doesn’t get at the true scale of the operation.


It doesn’t for example get into individuals located around the world tweeting themes propagated by Russia. Twitter would likely not have the capacity to generate such a number. And yet, confecting and amplifying themes is part of Russia’s game plan.

In any case, here is Twitter’s update on its retrospective review of Russian-related 2016 election activity, delivered to the United States Senate Committee on the Judiciary, Subcommittee on Crime and Terrorism.

It was published on January 19, 2018.

One interesting point, it makes is the spread of conspiracy material from Wikileaks. As I recall, Guccifer 2.0 was much busier as a clearing house of “scoops” for journalists than a Twitter account.

We also reviewed engagement between automated or Russia-linked accounts and the
@Wikileaks, @DCLeaks_, and @GUCCIFER_2 accounts. The amount of automated
engagement with these accounts ranged from 47.5% to 72.7% of Retweets and 37% to 64% of likes during this time—substantially higher than the average level of automated engagement, including with other high-profile accounts [my emphasis]. The volume of automated engagements from Russian-linked accounts was lower overall. Our data show that, during the relevant time period, @Wikileaks Tweets were Retweeted approximately 5.65 million times. Of these Retweets, 196,836—or 3.48%—were from Russian-linked automated accounts. The Tweets from @DCLeaks_ during this time period were Retweeted 6,774 times, of which 2.47% were from Russian-linked automated accounts. The Tweets from @GUCCIFER_2 during this time period were Retweeted approximately 24,000 times, of which 2.32% were from Russian-linked automated accounts.

We also analyzed data concerning Tweets promoting the #PodestaEmails hashtag, which originated with Wikileaks’ publication of thousands of emails from the Clinton campaign chairman John Podesta’s Gmail account. We found that slightly under 5% of Tweets containing #PodestaEmails came from accounts with potential links to Russia, and that those Tweets accounted for less than 20% of impressions generated within the first seven days of posting. The core of the hashtag was propagated by Wikileaks, whose account sent out a series of 118 original Tweets containing variants on the hashtag #PodestaEmails referencing the daily installments of the emails released on the Wikileaks website. In the two months preceding the election, around 64,000 users posted approximately 484,000 unique Tweets containing variations of the #PodestaEmails hashtag. Our automated spam detection systems identified in real time approximately 25% of those Tweets, hiding them from searches. Based on information we had available at the time we submitted our written testimony, we know that approximately 75% of impressions on the trending topic within the first seven days were views by U.S.-based users.

A significant portion of these impressions, however, are attributable to a handful of high-profile accounts, primarily @Wikileaks. At least one heavily-Retweeted Tweet came from another potentially Russia-linked account that showed signs of automation.

Follow Chris Zappone on Facebook