If you haven’t read Part I yet, you should. What follows is built off what was covered previously. Fair warning, the following is going to be rather extensive, because there’s much to unpack.
Every time I open up Twitter, the content and design is meant to influence me. For instance, the infinite scroll function has been called “behavioural cocaine” by the guy who designed it. There are ads strategically placed to get my attention. And of course, users are trying to influence me with their content.
When genuine people try to get a topic to trend, they do it to draw more attention to the message. When most users send out tweets, they are hoping to get that digital applause and get people’s attention. There is nothing wrong with any of that, generally. This is a natural extension of the social dynamics we engage in offline.
As mentioned in Part I, almost every activity people engage in can be done for positive purposes or malign purposes. Sometimes the only difference is what the doer is aiming for. You can swing a bat while aiming to hit a ball over a fence or you can aim that swing at a person.
Every morning, a number of Twitter users in America are rolling out of bed, starting their morning routine, and are likely glancing through Twitter. Some will glance at the trending list and notice a hashtag or phrase there that’s rather provocative. Before they opened Twitter, they likely weren’t thinking about that. But now, they may well just start thinking about it. They might even tweet about it.
Yesterday, there were trending topics when users woke up in the morning. The same was the case the day before that. The Twitter audience has largely become accustomed to this scenario. This is the status quo.
A year ago, I decided to question that status quo. I had already been investigating other aspects of Twitter and other sites. In my earlier investigations, I had begun noticing that certain accounts had a habit of heavily promoting hashtags. When combined with the information that had been revealed about some of the Internet Research Agency’s Twitter accounts, a question started bouncing around in my head.
I wondered, “Where do babies come from?” Then I wondered, “Where do these trends come from?”
I know some readers have asked this same question. I’ve seen so many tweets essentially asking, “Why is this trending?” I’ve seen the confused and relieved tweets when some folks find out that nothing bad happened to a celebrity that was randomly trending. (I’ve even seen many hostile actors that have copied that relieved message for their own purposes.)
In order to get to the bottom of it, I had to truly get to the bottom of it. When you click on a trending item in Twitter, you are taken to a search for that item. At this point, you are offered a few viewing options. The default is ‘Top’ but by clicking on the next tab over, you’ll view it in ‘Latest’ order, as seen in Fig. 1. Then it is a matter of scrolling down to when it started being heavily mentioned.
Catching a trend early was the ideal scenario for these investigations, but that wasn’t often the case. The earlier I was able to begin the search, the less there was to scroll through. Worth noting, where the trend started may not be the end of the search. There will often be sporadic mentions of the phrase or hashtag in the days or weeks prior to it trending.
During these searches, one thing I suspected became clear. Many trending topics are not what one might call ‘organic.’ They are mostly promoted in some way. Some by ‘blue check’ accounts with plenty of followers they’ve inspired. Others by a different sort of crowd.
Some readers undoubtedly already came to this conclusion before reading this. Some others might find this makes sense. I suspect there are a few who might like more information, and I’d love to break this down further.
When news breaks about a renowned person, some part of their name will naturally trend as people start posting about it. Their name is a constant. It is a known.
For instance, Twitter folks might recall when Neil Degrasse Tyson trended on Twitter for a controversial tweet he sent out in early August. In the trending list, it was listed as ‘Neil.’ As I’m typing this, ‘Lara’ is on the trending list, a reference to Lara Trump. Numerous users are responding to a contentious tweet she sent.
When an event happens in a city, often the name of the city will trend as users react to it. If a major sports team wins a championship, that team’s name will frequently end up trending. ‘Dorian’ trended when that hurricane was inbound.
These are known terms that people will naturally write about without needing to coordinate the message. As people react, these terms hit the list, which drives further reactions as people see those terms. More people start thinking about and looking into the terms, in response.
Conversely, there are trends that contain unknowns. Over the summer, the hashtag #GamersAreGood trended. This was in response to some remarks by Trump. But it originated from a ‘blue check’ user, who prompted his followers to tweet that hashtag. Many of the subsequent tweets were rather wholesome. Quite a few users explained how being a gamer had made their life better. This was not organic, but it still had a positive purpose.
When Jimmy Fallon would promote some hashtag on his show to get responses from Twitter users, that wasn’t organic but it also wasn’t malicious.
Returning to the #GamersAreGood trend, there are three unknown terms here. ‘Gamers’ ‘Are’ ‘Good’ are all unknowns. Without coordination, individual users who want to express this sentiment would likely choose differing synonyms or even different phrases to express that idea.
One user might say #GamingIsGreat. Another might write #GamingChangedMyLife. Yet another could go with #GamersAreTheBest. It often takes some form of coordination somewhere along the line for a uniform message to gain traction in these inorganic scenarios.
So far, we’ve covered organic trends and inorganic but non-malicious trends. Now comes the time to cover the inorganic and malign trends. But first, there’s something you should know. According to the most recent stats, there are 68 million US-based Twitter users and 330 million worldwide users. Have you ever wondered what those 262 million international Twitter users are up to?
I certainly have. And as I wondered about that, I realized that no one is checking out how these accounts are set up. A user could set up their account in any way they wish. Who would be the wiser? A guy from Asgard could pretend to be a guy from Midgard, and only he would know. This is the basic premise of anonymity on the internet, but it is still important to ponder. How many of those 262 million international accounts could be posing as Americans right now? As far as I can tell, nobody knows that.
For the record, I don’t believe that all 262 million international users are suspicious. I’ve seen plenty of legitimate Canadian accounts, British accounts, etc. There are clearly plenty of users who are from abroad that aren’t contributing under false personas. The concern about these false accounts is not just isolated to the US, either. In Part I, the issue of Brexit and the hostile actors therein was a subject of discussion.
That aside, the trending list is a vital component for hostile actors on Twitter. When getting to the bottom of their trends, there’s an entirely different world there. There are two general types of characters here. There are the workhorses and their accounts are usually husks, with very little information. The typical features are: No user description, either no picture or a nondescript one, no personal tweets and tens if not hundreds of thousands of tweets/retweets/likes.
Then there are the chatty promoters. Whereas the others are designed to stay off the radar and seem to have no interest in gaining followers, the chatty ones are far different. They want to be noticed, they want followers, and they can converse with ease. By now, you might be guessing the reason for these differences. The chatters are fluent, the workhorses are not.
The chatty accounts are dressed up ‘to the nines’ and are meant to be presentable. And just like those workhorses, they usually have tens if not hundreds of thousands of tweets/retweets/likes. High volume is the name of the game here. Whereas the top 10% of authentic users only tweet 138 times a month on average, these accounts reach that figure in less than 2 days, typically.
Based on the math, one of their accounts is doing the same amount of work that 15 of those top users would do over the same timeframe. I’m not sure you’d even want to hear how that compares to the median users’ tweet frequency of 2 per month. For the record, that research linked above intentionally did not include commercialized accounts, because those are an entirely different category of user.
You’ll find these types of accounts at the bottom of every malicious trend. While the trend itself may or may not have a provocative message, the chatty accounts don’t stop at just writing the hashtag or the phrase. They’ve got much more to say and share. And they want you to see it.
They ride their manufactured trend so that what they are saying and the images they post will get more visibility. The workhorses and the chatters work together to push the chatters’ messaging at average users.
Once the trend has started and real users start talking about it, the trend carries on from there and the chatters work their magic. In the case of non-provocative or mildly provocative trends, they’ll start incendiary arguments. This is often among themselves and done in a public manner. One of their methods is to latch onto a ‘blue check’ users’ tweet and create a divisive argument in the thread below it.
If the trend itself was extremely provocative, it is already working to influence users before they even see what the chatters have said. It is already driving arguments and division. Sometimes, as seen in the case of the #CivilWarSignup hashtag in the earlier figure, the term itself is so divisive that they play it off as “just a joke, bro.”
That #CivilWarSignup hashtag itself and the trends that preceded it were more than enough to get what they want. It goes without saying that Civil War is one of the worst case scenarios for a divided nation. Yet, if someone digs deeper, they’ve created a bit of plausible deniability there. “Oh, these accounts are just joking about bringing some food to the 2nd Civil War, as if it’s a picnic. This obviously makes sense?”
I’ve discussed earlier how subtle and effective these influence operations are. If you are accustomed to seeing this, you might not question or think too much on it. Time is and has been on their side. We are at a point where they can sneak several Civil War hashtags into the trending list and virtually no one questions it. This did not happen overnight. They’ve been working at this for years, gradually upping the ante, bit by bit.
Five years ago, they would not have been able to get those terms into the trending list without activating alarm bells in the heads of most audience members. Firstly, because the human embodiment of divisiveness wasn’t in the White House, throwing out those kinds of terms for them to latch onto. He is a current event that they consistently use in their operations. And he gives them more plausible deniability.
The second means by which they managed this was by using our biases against us very gradually. These operations are continuous. At least one of their manufactured trends will be in the trending list at all times. I have yet to investigate and discover that none of the top 5 trends are backed by hostile actors. There’s always been at least one.
Once the audience believes that what those actors trended and what they’ve posted is legitimate, the influence operation is a raging success.
Even if they don’t get a single person to promote or repeat their messaging, it doesn’t matter. Their material stands side by side with other legitimate material as an alternative to reality. And that would be enough. But they go much further than that. That particular discussion will have to wait for Part III, as this piece is quite long already.
The more people that know about how these actors operate, the less effective these tactics can be. That’s the purpose of this writing.
I’ll be back for Part III. I hope you’ll be back, as well.
P.S. By publishing these details in Part II, it could result in a change in tactics and appearance by these hostile actors. For that reason, there are some details that I’m leaving for later. However, the main identifying feature of these actors is the pushing of division. Seeing as that is their entire aim, I’m not too sure they can do much to alter that aspect. Time will certainly tell.
In weighing the pros and cons, the decision to publish rested upon the fact that if people don’t know about this, nothing will change and things may well get worse. Accordingly, publishing seems like the right move.
If you wish to support the work I’m embarking on here, consider contributing to my Patreon. This site is new but it has been in the works for quite some time. I’ve been investigating certain happenings since early 2017 and am committed to bringing them to light. I’ve saved up for this and have essentially robbed Peter to pay Paul. I’ve set aside the 9-5 job and have pumped it all into this. Because, in my estimation, these are matters of great import.
Do not feel obliged to contribute, as I will make it happen no matter what. Might take longer to do so, but I’m in it for the long haul. Thank you for reading and I hope you’ll consider passing along this information to others.