Category Archives: Twitter

Using Extremist Politics to Make Money on Twitter

I was searching for “Sessions” and “Russia” on twitter and found a series a similar posts published within minutes of one another as you can see below from the time stamps.

I then clicked on some of these and it turns out that not only these post are near identical, but that they have similar features: same posts, same links and same number of tweets (around 10.7k).

Crucially, if you click on the link viid.me…. it will take you to an advertisement. So it seems as though all these accounts are automated. They relay highly charged political news and bait people into clicking on those links to make advertising money.

Here is a video I recorded that summarises all of the above.

Is This Twitter Account Fake? Here Are The Features You Should Be Looking For!

As I was reading through the tweets with a #trident hashtag I found a succession of tweets from different accounts that were identical and posted within seconds of one another.

#trident twitter accounts.png

What is the likelihood that a dozen different people with distinct first and last names post the same thing at almost the same time?

So I went on to investigate a little further and here is what I found.

  • These accounts follow between 70 to 100 people.
  • They in turn are followed by 50 to 70.
  • In terms of post, some like Josh Wallace and Sarah Bentall have posted 12.5k times while others likeLayla Hare and Mandy Blume. have posted very close to 5800 times.
  • They tend to follow one another, so for instance, Mandy Blume, follows Jaime Frazer who follows Layla Hare who follows back Mandy Blume.
  • They post only news links in the same order at the same time.

//platform.twitter.com/widgets.js

//platform.twitter.com/widgets.js

It is difficult to believe that a dozen humans would have such a coordinated pattern of posting. More like this is the result of algorithmic posting. This is not unlike a similar pattern I had found on Linkedin earlier this year.

This suggests that if we wanted to find out a way to identify non-human twitter accounts we should look for the following features:

  • small to negligible standard deviation in terms of
    • followers
    • following
    • number of times
    • time of posting
    • date accounts created
  •  90% or more similarity in posts
  • accounts link to one another to make the numbers and influence analytics

Soccer Is Gaining Ground in the US

I was browsing for books on football at the Brookline Booksmith earlier today. One of them caught my eye:

The United States of Soccer: MLS and the Rise of American Soccer Fandom

The book traces how football finally took off in the US. And indeed, the most recent data confirm the main premise of the book.

By the end of 2016, the average attendance at MLS games stood at 21,692. This is 1,000 more than French Ligue 1 as of December 2016.

Meanwhile NBC made a calculated bet by broadcasting live premiership games. Very quickly cautious optimism gave way to outright euphoria. According to Comcast,

Nearly 10 million viewers have tuned in to NBC Sports Group’s Premier League coverage since Aug. 17, leading to a 67 percent daily viewership increase for NBCSN, marking the largest cable sports network growth over that span. Meanwhile, mobile viewership continues its rapid growth; fans have streamed nearly 35 million minutes via NBC Sports Live Extra over that span.

Even considering US trending tweets, today’s game between Manchester City and Arsenal was in consistently in the top 10 trends for the duration of the game. See the hashtags #MCFCvsAFC.

screen-shot-2016-12-18-at-17-46-34

Data from “https://trends24.in/united-states/

Tweet Ducasse: “Pas de Génies dans les Cuisines”