The synchronized fake race 2016

The synchronized fake race 2016

Twitter audience growth of the presidential candidates in 2016

On January 1, 2016, candidates have approximately an equal number of followers. Hillary Clinton reached 5M, Donald Trump was very close to 5.5M.

As of November 1. the number of followers of State Hillary Clinton exceeded 10M, and Donald Trump had 12.83M

Most of this growth consists of new users joined in 2016.

Total growth
of followers
joined in 2016
Percentage of followers,
joined in 2016

H. Clinton




D. Trump




The quality of both Clinton’s and Trump’s audiences is very low

41% and 42% respectively has no tweets.

73% and 67% respectively have abnormally high ratio following to followers.

90% of both sets of fresh followers have less than 10 tweets.

According to the STAQ methodology, more than 65% percent are definitely bots Level 1, that is primitive bots followers. You can verify this yourself by checking any of the new followers of any candidate at any time.

DonaldTrump`s followers

HillaryClinton’s followers

The research each account individually very important for analytics, but the study of the audience action can certainly lead to more interesting conclusions.

We decided to consider an audience growth of both candidates in each quarter by day.

The results looked quite normal until we have built both graphs in the same coordinate plane. So look the first quarter of 2016.

Q1 2016

Out of the 50 first days of the year, 40 of them have the ratio of new followers, created in the same day is 63.4% (+/- 4%). After February 23 the curves diverge, but there are some similarities. But that was only the beginning of an anomaly.

Q2 2016

If not for these two strange spikes, we could say that these curves are parallel. We see two obvious downturns of Trump new audience, and then the same sudden rise, compensating all the losses in full.

Q3 2016

In this period, we see not only the synchronous oscillation within a certain area but also the simultaneous abnormal fall and subsequent rise.

The most explicit anomaly of Q3 can be divided into three stages by 3 days

1. Falling from 4 to 6 August.

2. Stagnation in the area of low activity from 7 to 9 August.

3. Recovery of growth from 10 to 12 August.

The question is whether the two different audiences behave the same way, repeating all the trends, or it may be a common core of these audiences? Let's add a line for common followers of both candidates on the chart.

714,000 of common followers of both candidates joined Twitter in Q3 2016.

It is 52% of the 1,370,000 Trump followers, and 72% of the 995,000 Clinton followers.

We see a simultaneous increase in the number of followers of the two candidates during the election campaign. In addition, we see the unexplained repeated synchronous drop and following synchronous growth. Consequently, the number of followers is regulated from a single center. Is it possible that Trump team and the Hillary team coordinate their actions? No chance. That means there is a third party that takes care of both candidates growth audience.

Who can simultaneously add followers to both candidates? It can be any large botnet owner. The problem is that to work on the two campaigns at the same time is very difficult. In addition, the bots during the race should be very active, but there is 90% of dormant accounts.

Who is interested in the statistical growth, and who do not care about the inactivity of this audience?

Instead of conclusions

For the first three-quarters of 2016, the number of unique followers of both candidates joined the tweeter in 2016 amounted to nearly 5 million users (taking into account common 1.5M followers).

According to official stats, the total increase in monthly active users (MAU) for this period amounted to 7 million, including 2 million US users.

Analogies in Science

This year two scientists, Konstantin Batygin and Michael E. Brown, showed us an evidence for a distant giant planet in the solar system. The study is published in the Astronomical Journal January 20, 2016. Their proof is based on the discovery that the orbits of distant Kuiper Belt objects (KBOs) cluster not only in the argument of perihelion but also in physical space.

In more simple terms, we can not see the planet, but we can find out how to change the force of gravity the same trajectory of thousands of objects on the distant orbit of the solar system. The chance that they accidentally changed its trajectory, according to scientists is less than 0.007%. Here we are dealing with something like this.

The appendix. Twitter audience growth of the presidential candidates in 2015