A team of U. Vermont scientists, led by applied mathematician Peter Dodds, analyzed over 46 billion words from 63 million individual users on the website Twitter over a 33-month period to conclude that happiness is not a trending topic.
Through Amazon Mechanical Turk, an online workforce that provides human intelligence, people rated the “happiness” of the 10,000 most popular words on a scale from one to nine: one being sad, five being neutral and nine being happy.
Words such as “laughter” rated an 8.5 and “food” rated a 7.44, while “funeral” rated a 2.1 and “terrorist” rated 1.3.
These results indicate that happiness is on the decline in the world of Twitter, according to a study published on PLoS ONE, a website dedicated to peer-reviewed science publications.
“Words are the atoms of our language, the building blocks of our sentences,” Dodds said. “We kind of forget about language, since we use it so much. Now we have social networks that provide a wealth of knowledge that we can look at and collect data [from].”
The study revealed a response between bad news and unhappiness in tweets.
Events such as the bailout of the U.S. financial system, the swine flu pandemic, natural disasters in Chile and Japan all produced relatively low happiness. The largest single-day drop occurred with the death of Michael Jackson.
Cultural and sporting events such as the season finale of “Lost” and the results of the 2010 World Cup when Germany defeated England also demonstrated recognizable drops in happiness.
Dodds said that because Twitter contains such an enormous amount of data, it pushes the limits on what scientists can do.
“Twitter is a great tool to see the reflection of movements,” he said. “We can see how movies, books and even diseases take off and slowly build over time by word of mouth.”
The research team is not the only group interested in studying social mood patterns on Twitter.
“The CIA has been following Twitter for signs of social discontent,” Dodds said. “There is a worry that big governments are following people, but as democratic nations-we can only hope it is for good of the people.”
Chris Danforth, an assistant professor in the department of mathematics and statistics, said that social media has introduced a whole new method of research for scientists.
“Before, researchers could only collect data from a few hundred people. But now, we can do it on a global scale, since Twitter is nearly universal,” he said.
The research group said they benefited greatly from the Vermont Advance Computing Core.
“We were receiving roughly 30 million messages a day, taking up 50 gigabytes of memory,” Danforth said. “Messages that come from Twitter were pulled out [by the ‘Super-Computer’] with their relevant information such as location and time. We wouldn’t be able to parse that data without it.”
The team of researchers is currently working on a website where users would be able to play around with the information gathered from Twitter.
“For example, people could search tweets about ‘pancakes’ and find out how happy those tweets are or the demographics of the tweets,” Danforth said.
But not all news is negative.
The study also showed that people were happier during the weekends and holidays, with Christmas Day reported as the highest level of happiness, followed by Christmas Eve.
Yet Dodds also said that too much happiness may not be the best goal for society.
“We wouldn’t want to live in a place like Brave New World—where everyone is high on soma, wandering around because everything is beautiful,” he said.