Concepts such as Blue Monday (saddest day of the year), which is celebrated on the third Monday in January; or Yellow Day (happiest day of the year), which is celebrated every year on June 20, are pseudoscientific and advertising concepts, without any scientific basis. There is no evidence that during a certain day of the year people are more or less happy, not even evaluating more or less objective variables, such as the weather.
However, we could qualitatively assess the level of satisfaction of a group of people at a specific time through, for example, the analysis of the content of the tweets issued by a community. This is precisely what an artificial intelligence tool has done, which has concluded that the two weeks after March 26 were ‘the saddest time of the year for Twitter’. The tool, which has managed to quantify ‘global happiness’ on social media, recorded an unprecedented drop in mood starting in late May, as shown in the graph below. Twitter users were sadder than ever at the end of May.
Chris Danforth and Peter Dodds, applied mathematicians at the University of Vermont at Burlington, have been tracking the sentiments of Twitter users since late 2008 using the ‘hedonometer’, a tool that analyzes words in 10% of randomly sampled tweets each day, and rates them as positive or negative.
In mid-March, when the COVID-19 pandemic swept the West, the tool traced a deep and sustained drop in global mood. Then, global protests following George Floyd’s assassination in May triggered a second wave of negative sentiment on the platform, the researchers report in the journal Nature .
How does it work?
The hedonometer compares the language used in tweets with a database of more than 10,000 common words, taken from Google Books, The New York Times, song lyrics and Twitter itself, which have been rated by 50 people on a scale. of 9 points; neutral words and those that receive a wide variety of scores are excluded from the analysis.
Certain events, such as mass shootings, controversial government policy announcements, and deaths of well-known figures, like that of Michael Jackson in 2009, show up on the hedonometer as one- or two-day drops. “Typically, these are quickly drowned out by the background noise related to music, sports and celebrities,” says Danforth. “But with the pandemic, we see that the signs of sadness are prolonged.”
Other events, on the other hand, are typically positive, and the messages of happiness are repeated cyclically year after year, such as the Christmas period, as can also be seen in the graph.
In addition, the volume of tweets in the last 2 weeks is almost double the daily average since 2019: “Both the pandemic and these protests have been much more coherent in terms of collective attention than anything we have seen,” the researchers point out. .
