clock menu more-arrow no yes mobile

Filed under:

Monday was the saddest day ever recorded on Twitter

The Las Vegas massacre moved the needle on the “Hedonometer,” a tool that measures happiness on Twitter.

Las Vegas Mourns After Largest Mass Shooting In U.S. History
“It’s common for terrorist attacks or natural disasters to move the needle of this instrument,” said Chris Danforth, the University of Vermont mathematician behind this tracking tool, “but this is the lowest measurement we’ve ever had.”
Photo by Drew Angerer/Getty Images

If what we say on Twitter is any indication of how we’re feeling these days, Monday, October 2, was a very, very bad day.

Mathematicians and complex systems scientists at the University of Vermont have been tracking sentiments on the social media site since 2008 with a program they call the “Hedonometer.” And Monday, the day after the massacre in Las Vegas that claimed 59 lives, was the saddest day they’ve ever measured on Twitter.

The Hedonometer uses an algorithm that scans a random 50 million (or 10 percent) of all global messages on Twitter written in English. It then tracks the most frequently used 10,000 English words on a happiness scale the researchers devised and throws away neutral filler words (like in, and, or of) to arrive at an average measure of our collective joy or sorrow.

On their scale of 1 to 9, 9 is pure happiness, and on the average day the mood on Twitter hovers around 6 to 6.1 — relatively high numbers the researchers attribute to the fact that there are more positive than negative words in the English language.

On Monday, the scale dropped down to 5.7. As you can see in the chart above, Twitter users were using very somber words like tragedy, victims, gun, dead, evil, and killed on Monday — language the algorithm tagged as negative.

“It’s common for terrorist attacks or natural disasters to move the needle of this instrument, but this is the lowest measurement we’ve ever had,” said Chris Danforth, the University of Vermont mathematician who co-founded the tracking tool with his colleague Peter Dodds.

Monday, October 2, was the saddest day ever recorded on Twitter.

The Hedonometer has found that happiness — at least the kind expressed on Twitter — generally comes in pretty smooth waves: The lows typically follow natural disasters, the deaths of beloved celebrities, and terrorist attacks (though the election of Donald Trump marked one of the saddest says the group measured as well). The highs center on holidays (Mother’s Day, Christmas) and other joyous occasions like a royal wedding.

Of course, the meter isn’t perfect — Twitter users aren’t necessarily representative of the general population, and we may not always say exactly what we think on a public website — but Danforth says it tends to track pretty closely with moods expressed in opinion polls.

Interestingly, people tend to be happier on the weekends (especially Saturdays) and more pessimistic on Mondays and Tuesdays, Danforth says. That’s why he thinks the timing of the shooting may have compounded people’s sorrow. “It’s the biggest mass shooting in [modern] US history, and it happened on one of the days of the week that tend to be sad to begin with,” said Danforth.

There’s more bad news: On average, we seem to be less happy than we were, at least according to the Hedonometer.

When you zoom in on the past 18 months, you can see the happiness scale trending down, especially following the terrorist attacks in Orlando and London and the US election last year.

The past 18 months have been “more of a roller coaster” emotionally, according to our tweets.

“[Our happiness cycles have] been incredibly regular for eight years until the last year,” said Danforth. “Now the signal is jumping down a lot more, and the regular weekly cycle has fallen apart. It’s more of a roller coaster now than it used to be.”

So it’s not just you. The past several months have been rough, if Twitter sentiment is to be believed. “We have a lot to think about these days,” Danforth said. “The news feels progressively worse and more stochastic than it used to be.”