Online disinformation campaigns. A deep dive.

A few days ago I published a little summary of online disinformation campaign for the Spanish far right [1].

I tried to highlight the emerging patterns of how these amplification machines operate. Shoving retweets distribution for users rt-ing one of the top-talker accounts.

I am now going to share a larger archive of tweets. To obtain this archive I compiled all the tweets from users that have retweeted the account @ldpsincomplejos.

Here is how I obtained the tweets for the account @ldpsincomplejos was obtained:

def get_tweets_search(key, client):
    tweets_endpoint = "https://api.twitter.com/1.1/search/tweets.json?q=" + key + "&count=500&include_entities=true&result_type=recent"
    response, data = client.request(tweets_endpoint)
    return json.loads(data)

This function will return 500 tweets, including both tweets by the account and RTs to the account. For each tweet I extracted the username and got the latest 200 tweets user timeline.

def get_timeline(username, client):
    timeline_endpoint = "https://api.twitter.com/1.1/statuses/user_timeline.json?screen_name=" + username + "&include_rts=true&count=200"
    response, data = client.request(timeline_endpoint)
    return json.loads(data)

Now I identified a few top-talkers and computed the distribution of RTs to Accounts.

Here are some results:

Twitter distribution for users RT-ing @okdiario: who are they RTing?

distribution_okdiario full image

Twitter distribution for users RT-ing @hermanntertsch: who are they RTing?

distribution_hermanntertsch full image

Twitter distribution for users RT-ing @alvisepf: who are they RTing?

distribution_alvisepf full image

Here is the thing. If you look at the distribution of comments instead you get a different result.

Twitter distribution for users commenting-ing @okdiario: who are they commenting?

distribution_okdiario full image

Twitter distribution for users commenting-ing @hermanntertsch: who are they commenting?

distribution_hermanntertsch full image

Twitter distribution for users commenting-ing @alvisepf: who are they commenting?

distribution_alvisepf full image

This means in practice that there is a list of accounts RT-ing sources on the right and commenting on accounts from the left. The anticipation machines objective is in fact to spread division and hate.

Finally here is a visualization of the graph of accounts RTin OkDiario:

plot_graph_okdiario full image

If we isolate the accounts doing more than 100 RT, we find a group of approximately 20 accounts doing most of the tweets.

plot_graph_okdiario full image

Here is probably a team of accounts acting to promote the tweets from OkDiario and a group of other accounts.

Would you like to have access to this data?

I have a metabase instance setup on heroku.

You can have a look at the data used in this study.

If you get in touch I can give you access.

REFERENCES

[1] https://www.hiro7.eu/blog/2020-04-05-online-disinformation