A few years ago, two Cornell University sociologists, Michael Macy and Scott Golder, studied more than 500 million tweets that 2.4 million users in eighty-four countries posted over a two-year period. They hoped to use this trove to measure people’s emotions—in particular, how “positive affect” (emotions such as enthusiasm, confidence, and alertness) and “negative affect” (emotions such as anger, lethargy, and guilt) varied over time.
