Whose Tweets are More Controversial

In this post, I compare “The Ratio” of tweets by both Donald Trump and Barack Obama—via their personal twitter accounts—to determine which one is more controversial.

But first, an important note:

The numbers in this text were all valid as of around 1:15pm CDT on August 28, 2017, when the data was first downloaded.
It’s not unusual for replies/retweets/likes to disappear. That happens all the time: as users get deleted, they clear out their profiles, they get banned, they change their account to be protected, or Twitter just “samples” data for displaying details rather than showing every bit of data they have.
In other words, don’t be surprised if the numbers don’t seem to match what I have in the text, or if you can’t see replies even when there are replies counted.

The methodology and “The Ratio”

“The Ratio” is a nebulous thing, I’ll admit—but it can be a useful indicator of the potential controversy caused by a tweet. In order to understand what it means, let me start out by giving you an illustration. Here, you can see a recent tweet by Donald Trump (embedded as an image, for annotating):

Trump’s recent tweet on “THE WALL”

I’ve highlighted two numbers (both rounded for display by Twitter): The number of replies, and the number of retweets. These aren’t exact, of course, but that doesn’t mean that the exact numbers aren’t available. More on that in a few paragraphs…

That ratio, though. That tweet got 55,935 replies, and just 26,059 retweets (those numbers are likely to have changed by the time you read this). If you divide the number of replies by the number of retweets, you get “The Ratio”.

“The Ratio” for Trump’s tweet about the wall.

2.15 to 1. That’s “The Ratio” for this tweet. What does it mean?

We Twitter users are pretty lazy, for the most part.

That’s not to say that we’re all lazy, just that we tend to retweet things we agree with more often than we reply to them. The river of tweets on our timelines flows quickly, and a quick retweet is how we show our followers that we agree with a tweet.

When we have something to say, though… we don’t waste time. We come right out and say it. Sometimes in mega-tweetstorm-threads, sometimes with a commented retweet (“quote tweet”), but we’re not afraid to put fingers to keys and reply when we disagree with what someone says.

Side note for all of those people who say “retweets aren’t endorsement”: you’re lying. That’s exactly why you retweeted it, because you agree with the message. If you had something to say about it, you’d reply to it, or you’d “quote tweet”.

“The Ratio”

How often have you reached out to a company to tell them how good a product or service is? How often have you mentioned to others about a product or service by that company? Finally, how often have you complained — either to the company or to others—about a company?

When a tweet gets someone stirred up so much that they break their timeline-scrolling stride to actually write something, it’s rarely a good thing. The more comments a tweet receives, the more potentially controversial it is. Especially when that post has comparitively fewer retweets.

The higher the ratio, the more controversial the tweet tends to be. Note, I didn’t say that as an absolute! Not every tweet with a high number for The Ratio is a controversial one. Sometimes, it’s because a question was posed to other Twitter users, and feedback was therefore requested directly. In this post, you’ll see several tweets like that—where the author was seeking comments, and the Twitter audience responded by replying.

When The Ratio is high for a given tweet, some have taken to calling that a “Ratio Tweet”. It’s a good phrase, I think. Since there’s no official definition, I’m going to give it one:

Ratio Tweet: A tweet where the number of replies exceeds the number of retweets by more than 25%.

Collecting the data

So, let’s get back to finding the actual non-rounded numbers for these tweets, shall we? Firefox’s Web Developer Tools to the rescue! I can inspect the markup used to create the reply count by right-clicking and choosing “Inspect Element” in Firefox.

The element showing the number of comments received

Looking at the source code for this element, I can see the real number behind the approximated value used for display, in an HTML attribute called “data-tweet-stat-count”:

The HTML source code for the “Reply action count” shown for this tweet.

That’s 55,935 replies to that tweet. The retweet count shown is 26K, and the actual value (which is in a similar place in the HTML source code) is 26,059.

Trump has—as of about 1:15pm on Monday, August 28, 2017 (when I retrieved the data I used to write this post)—written 32,530 tweets since his first one, on May 4, 2009. That first tweet was, as you might guess, a self-promotion. But that’s okay, that’s what Twitter is for, right?

Darn, I missed it.

Barack’s first tweet—out of the 13,512 sent as of 1:15pm on Monday, August 28, 2017—was sent on April 29, 2007. Ready for it?

He was a US senator from Illinois back then.

Because I know you’re curious, here’s my first tweet. You’ll note that I tweeted more than a year before Donald Trump did. I’m so proud. 🙄

Spoiler alert: Yes. Yes, I will.

Now, I’m not about to spend my day using my browser’s inspector tools to read 46,042 tweets between these two accounts and write down their exact numbers, so I wrote a program that crawled the search results for all of their tweets. All of them. The Twitter API doesn’t return the number of replies for a tweet, so writing a crawler was the only way to get that number. I wrote the crawler in a programming language called Python, and used one of the best web scraping/crawling libraries out there: Scrapy.

I’ll post a technical write-up another time; I’ve removed it from this post to help slim it down. 🤓

The analysis

I threw out all the tweets that had a zero for either replies, retweets or likes (which Twitter calls “favorites”). This resulted in a total of 30,629 tweets for Trump, and 12,399 tweets for Obama, after removing those with zero activity in one of those three areas.

Let’s see some basic stats on their tweets, shall we? First, the controversial status of their tweets. As a simple pie chart, the difference in these men’s styles is startling.

Note: I’ve only included Trump’s tweets since he declared his candidacy on June 16, 2015.

Comparing the percentage of “Ratio Tweets” from Donald Trump and Barack Obama

Let’s see what their most “controversial” tweets are about. I’m going to share the top 10 tweets, sorted by “The Ratio” from highest to lowest.

Donald Trump’s Most “Controversial” Tweets

Let’s just look at tweets since he announced his candidacy. He wasn’t in government before then, and I’m interested in “presidential” Trump. I excluded tweets that were themselves replies, retweets, or quote tweets.

What’s his most controversial tweet since he announced his candidacy on June 16, 2015? Let’s see:

No, wait, that’s not controversial at all! He asked a question. It got lots of replies, and that’s to be expected—or, at least, hoped for. I think it’s safe to say: It’s not a “Ratio Tweet”.

He seems fascinated with bleeding women ¯\_(ツ)_/¯

89,585 replies (remember, you may see a different number, since the embedded tweets are “live”) and just 20,784 retweets. “The Ratio” of this tweet?

That’s a hell of a Ratio Tweet, Donald.

4.31 to 1. That’s crazy. Definitely a “Ratio Tweet”. For every person opting to retweet this post to their followers, more than four others replied to it. Now, not all of those replies are negative, mind you. There are automated accounts (“on both sides”, legitimately) that reply to every one of his tweets, either with support or with shade. There are supporters there, defending whatever he said. There are the “troll” types, who show up to say controversial or hateful things in reply to other comments. But the point remains: When you get a group of millions of users—who mainly click their way through their timeline—to stop and write something…Well, you’ve struck a nerve, haven’t you?

His next most controversial tweet so far?

Little known fact: Andrew Jackson was the illegitimate son of Miss Cleo, the famous telephone psychic.

I’ve used the amazing Airtable to create a table of Trump’s top ten tweets, sorted by “The Ratio” in descending order — it includes the non-controversial tweet about the debate, because it (technically) had a high ratio.

Trump’s 10 most controversial tweets since June 16, 2015

Barack Obama’s Most “Controversial” Tweets

Since Obama was already a member of Congress when he started tweeting—back in 2007—his entire timeline is fair game to find controversy. Without further adieu, Obama’s most controversial tweet, according to “The Ratio”:

Wait a second. It has a crazy ratio of 20:1! But…it’s clearly not a controversial tweet. Only one retweet (poor guy) and twenty replies. What’s weird is that I only see seven replies, and they’re all from the same person, made several years later. Maybe Twitter lost some data, or there’s something to their calculation of replies that I’m not getting. Or there were a lot of replies made by people with “protected” accounts, so we won’t see them.

Either way, it’s not controversial. Maybe it’s like Trump, and his next one really got people riled up:

Well, geez, I’d probably talk about how you don’t actually seem to cause Twitter controversies. Obviously, not a “Ratio Tweet”—he’s directly asking for comments—even though the ratio is 7.9 to 1.

Maybe the third tweet’s the charm?

Barack, dammit, why won’t you rock the boat with your tweets? A ratio of 6.57 to 1, and it’s also directly asking for replies. Not controversial, even if the replies may be.

In fact, nearly all of his top ten “Ratio Tweets” are asking for feedback. I had to go to the last in the list to find another one that wasn’t directly calling for replies:

A ratio of 4.48 to 1, that one. I don’t see what the big deal is, though. Doesn’t seem particularly controversial to me.

Here are Obama’s top 10, embedded from Airtable:

So, What Have We Learned?

In nearly every case where Barack Obama’s tweets garnered more replies than retweets, it was because he was asking for feedback. He reached out to his fellow Americans in a public forum, seeking their equally public responses. A world cloud generated by using the text of all of his “controversial” tweets — the ones with The Ratio higher than 1.25—is here:

Word cloud of Barack Obama’s tweets with a high reply to retweet ratio.

Donald Trump, on the other hand, doesn’t seem to care about feedback from people. In fact, his only requests for replies came from before his candidacy, and were mostly asking for a response to something that happened on Celebrity Apprentice. He seems to create controversy pretty reliably with his word choice, as seen in this word cloud of all his “Ratio Tweets” since announcing his candidacy on June 16, 2015:

Word cloud of Donald Trump’s tweets with a high reply to retweet ratio.

It doesn’t look like his controversial approach to social media is going to wane any time soon. And why would it? Nobody has any authority over how he behaves on Twitter. As long as “fake news”, things that he “will” do, and things that are “great” continue to energize his base, he’ll continue to use those key words whenever he tweets.

So, what do you think?

You can leave a response to this post, or share it on Twitter with your feedback. I have other insights I’m experimenting with, and if I find anything interesting, I’ll be sure to write up a new post.

To quote Trump: “Enjoy!”

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.