Edoné in Greek means pleasure, therefore to measure happiness one could think of using an instrument called for example a hedonometer. Here, the hedonometer really exists: he created it the University of Vermont using an algorithm that can measure the mood of public opinion in real time and around the world. And how do you measure happiness, which is a subjective state of mind, through the icy objectivity of a computer? Easy: analyzing what people express on Twitter, one of the most used social networks in the world, with almost one billion users, of which around 200 million are active daily.
"The United Kingdom and Twitter are not the same thing," said former British prime minister David Cameron after winning the election against the most popular - at least on social networks - Ed Miliband. But it is true that The University of Vermont Hedonometer scans about 50 million tweets a day, and something true will have to come out. For example, it turns out that 2020, due to the Covid pandemic, is by far the worst year since 2008, that is, since the first data was collected. In fact, precisely in March, when the emergency began (in the Western world), an initial peak of "unhappiness" was reached, even if it must be said that the absolute record was reached in May, when, in addition to all the worries that gives the virus the murder of George Floyd and the resulting Black Lives Matter protests have been added.
This last datum in particular gives an idea of how slightly "Western-centric" the hedonometer is, even if it is not devoid of scientific basis. It was the American university computer scientist Chris Danforth ad teach a machine to understand emotions behind the analyzed tweets (no human being could ever read them all): this process, called sentiment analysis, has made great progress in recent years, finding multiple applications. “In the social sciences we tend to measure simple things, like gross domestic product. Happiness is an important element but it is very difficult to measure,” admits Danforth.
But how does the hedonometer actually work? In fact, understanding language is one of the most complex problems in the field of artificial intelligence. However, within a written text, there are abundant emotional cues that computers can recognize even without understanding the meaning of words. Meanwhile, counting the number of positive words and subtracting the number of negative words. A better measure can be obtained by weighing the words: the term "excellent", for example, carries a more intense feeling than "good". The weight of words is usually established by flesh-and-blood experts, and plays a role in the creation of lexicon dictionaries, those that translate words into emoticons (and vice versa) and are widely used in the analysis of feelings.
Yet words can be misleading and sometimes the mere algebraic sum of adjectives can miss the meaning of a complex sentence, perhaps said as a joke or ironically. But today, experts swear, the algorithms are so sophisticated that they can even grasp the hidden meanings and be able to correctly connect words to each other. The new frontier of sentiment analysis by algorithm is no longer just the measure of mood, but also the prevention of potential depressive disorders, anticipated by what people write on social media. It already happens with Facebook, which uses an algorithm capable of identifying users who appear to be at risk of suicide. A team then examines the reported cases and decides whether to invite users to contact specialized assistance.
However, that 2020 was a nefarious year so far perhaps there was no need for Twitter to confirm it. Just as it is quite obvious that people are less enthusiastic when it rains, yet analyzing two billion interventions on Facebook and a billion tweets in 2016, it came out that three centimeters of rain caused a 1% reduction in the happiness expressed by users , a percentage that doubled in case of temperatures below zero. But while they were at it, Vermont computer scientists put it to the test too Monday's reputation as the saddest day of the week. And here instead there was the twist, because social media users are more "negative" than on Tuesday. Friday and Saturday, however, were the happiest days.
And for the record, the second "unhappiest" year globally, according to the hedonometer, was 2017, while the "best" period for general mood was between 2015 and 2016.
