Report Outline

In the following sections, I will attempt to lay out the following:

  1. Evidence of sockpuppet use
    • And the connection with Shareblue
  2. Which voices are being amplified
    • Verified and unverified accounts
  3. What messages are being pushed
  4. The extent of the astroturfing operation

Sockpuppet tech support

The best evidence of Sally Albright using sockpuppets is the fact that she literally tweeted it out. In March of 2017, Sally was concerned over some changes to Twitter automation software called Buffer. In the process of troubleshooting the issue with Buffer support over Twitter, she tweeted a screenshot of the the Buffer browser extension. The screenshot clearly shows dozens of Twitter account avatars signed in to the service and ready to lay down astroturf. Buffer allows for creating groups of different accounts in order to schedule tweets (or retweets) for different purposes or communities. At the time of the screenshot, Sally had 24 accounts in a group called “Weird Twitter”, 25 in “Hillary”, 6 in “Group 1”, and 14 in “Tier 2”.

Getting red and mad and nude online trying to troubleshoot sockpuppets with Buffer support.

Getting “red and mad and nude online” trying to troubleshoot sockpuppets with Buffer support.

Connection to Shareblue

In the above screenshot, avatars of three dozen accounts are clearly visible. Six of the avatars were easily recognizable and matched photos from accounts that commonly retweeted Sally. Importantly, all of these accounts were tweeting links to multiple times per day in late January, 2018. The screenshot of account avatars, the Shareblue articles shared via Buffer, and the fact that the official @Shareblue account also uses Buffer, offer circumstantial evidence that Sally’s sockpuppet army is either working with or adjacent to Shareblue. Shareblue has a reputation for not disclosing its employees, so there is little chance that they would admit to their involvement in this astroturf campaign.

The tweets in the above image can be found in this collection. In general, tweets sent with Buffer can be searched for on Twitter by including ‘source:buffer’ along with the desired query. To find the Shareblue tweets sent from Buffer, search: shareblue source:buffer.

In the event that tweets are deleted, they have been archived here:


Tweets from 64 suspected sockpuppet accounts were collected using the rtweet package for R. Up to 3200 tweets (includeing retweets and replies) were gathered from each target account. The total number of tweets collected was 194377. Data regarding the clients used and individual sockpuppet accounts can be found at the end of the report.

Which accounts are being amplified?

Unsurprisingly, Sally Albright was the most retweeted account overall. She retweeted herself 30939 times from 63 unique sockpuppet accounts. This amounted to 15.6% of the 190k total tweets collected. In the first six weeks of 2018, she managed to retweet herself an average of 559 times per day.

Verified Accounts

Eric Garland was the second most retweeted account overall, with 2678 retweets originating from 59 unique sockpuppet accounts. Eric Boehlert was the most retweeted Shareblue employee, with 1883 retweets originating from 60 unique accounts.

The wordcloud below shows the 100 most frequently retweeted verified accounts. Font size of each screen name corresponds to the number of retweets received relative to other verified accounts.

In addition to actively amplifying conspiracy theorist Eric Garland, “the Breitbart of the left” also engages in artificially inflating it’s percieved reach and authority by retweeting it’s own writers.

Unverified Accounts

In the following analysis, @SallyAlbright’s massive number of self-retweets were omitted for ease of visual interpretation.

What Are they Retweeting?

Most Commonly Retweeted Words, Overall

Most commonly retweeted positive or negative words

Most commonly retweeted positive or negative words in tweets mentioning Bernie Sanders

What Narratives are Being Pushed?

Smearing Sanders Supporters as Racists




Bots and Trolls Hypocracy




Attacking the Unity Reform Commission



Defending Her Honor At All Costs



Bad Political Takes



General Bernie Hate



Dynasty Worship and the New Feudalism


Frequency of (re)tweets

99% of the posts made by the sockpuppet accounts were automated retweets using Buffer. The following plot shows the retweet frequency of the sockpuppet accounts. In the past month, the total number of retweets was 132773, for an average of 4200 per day. However, the number of retweets frequently reached over 6000 per day and went up to a maximum of just shy of 8000 retweets per day. Overall, Shareblue’s astroturfing campaign has the capacity to produce upwards of 50k retweets per week. It is important to note that this is all done in-house with a single person using software that costs $200 per month and fake accounts.

Original Tweets Linking to

While the primary purpose of the astroturf sockpuppets is retweeting real accounts, 1.04% of the tweets collected were original tweets that linked to articles on Shareblue. The existence of these tweets is evidence that Sally Albright’s astroturf operation is indeed connected to Shareblue.

The following plot shows the frequency at which original tweets with links to were sent via Buffer.

Individual sockpuppets

The plot below shows the retweet frequency timeseries separated by individual accounts during the last month.

Retweets of verified accounts

The following plot shows the rate at which eight commonly retweeted verified accounts were retweeted over the past month.

Retweets of unverified accounts

The following plot shows the rate at which eight commonly retweeted unverified accounts were retweeted over the past month.

Clients Used

About 95% of the tweets collected were made using Buffer automation software. The next most common clients were RoundTeam (7k tweets), Web Client (800 tweets) and Hootsuite (400 tweets). RoundTeam and Hootsuite are other types of Twitter automation software.

Sockpuppet Accounts

Tweets from each of the 64 accounts were primarily sent with Buffer except for 2 accounts which used RoundTeam.