When Fidel Castro passed away on November 25, how many people read his obituary? While the obituaries of political leaders may not be the most exciting texts to peruse, they are integral for researchers at NYU who are exploring how death affects politics.
Ralph Grishman, Professor of Computer Science at New York University’s Courant Institution, and Alastair Smith, Chair of Politics at NYU, are building a database of the health and obituaries of political leaders using web-scraping, natural language processing, machine learning and handcoding to extract relevant information from articles, documentation, and obituaries.
After completing the database, they hope to use the database to test various hypotheses about levels of political protest, coups, and political liberalization. For example, one prediction they want to test is whether the expectations about specific leaders’ health have significant influence on the evolution of political institutions.
Consider the scenario where an autocratic leader’s health is declining. “In autocratic systems the loyalty of supporters is maintained by the expectation of the flow of private benefits that a leader provides. Leaders cannot commit to continue rewards from beyond the grave,” Grishman and Smith explained. A leader’s approaching death would therefore undermine loyalty, since supporters become unwilling to continue supporting the regime, and the political elites begin searching for replacements.
Grishman and Smith also said that not only were protests more likely once a leader’s declining health became known, but their game theory model also predicted that “leaders’ responses to declining health generally involve creating more inclusive political institutions and liberalizing.”
Grishman and Smith’s project was awarded an NYU Data Science Seed Grant by the Moore Sloan Data Science Environment at NYU earlier this year. In an era of increasing political instability across the globe, their work is crucial for our broader understanding of the particular social and cultural conditions that spark volatile political events.