Can Data Science Save Our Democracy?

The two audio books I’ve listened to most recently are How Democracies Die by Steven Levitsky and Daniel Ziblatt, and Facism: A Warning by Madeleine Albright. Not exactly spiritually uplifting choices, I’ll admit. Levitsky and Ziblatt explain how the erosion of political norms and the failures of democracies’ gatekeepers enable autocrats to seize power. They cite several historical examples in which democracies fall or have been severely tested when the people who are supposed to protect democracy by following unwritten codes of conduct abdicate their role. In her book, Albright chronicles the rise of several fascist regimes over the past century, starting with Mussolini and Hitler and continuing through more recent examples like Chavez, Duterte, Maduro, Orban, and Putin. In most of these cases, an outsider identified a nascent thread of anger and discontent, offered the aggrieved an alternative that won their affection and support, gained acceptance by – or, at least, cooperation from – the establishment by persuading them that they could not afford to distance themselves from the growing din of his political base, and then, once let in, seized power. The way the two books present it, the rise of authoritarianism and fascism seems to follow a playbook.

The availability of personal data could make it even easier for a fascist to rise today. Indeed, the importance of data adds a whole set of tools to the fascist’s strategy. He who controls the data sets the conversation. An autocrat could use all the data that social media, e-commerce, and search platforms collect on us constantly, cluster us into distinct political groups based on that data, and promulgate messages that unite certain of those clusters against others to solidify a base of support and create divisions that amplify that base’s voice. Or, the would-be fascists can tailor messages to each cluster, similar to how Netflix tailors movie recommendations, so that he can tell each group what they want to here, when they want to hear it, and thus win them over. Once smitten with the rising star, we’d surely be more malleable and forgiving as our various freedoms are stripped away. All the demagogue needs is the data. The tools to analyze the data already exist, thanks to the fast-growing field of Data Science.

So far, I have painted a very bleak picture of technology’s role, because I have focused solely on how data analysis tools can empower those who seek to damage our democracy. Of course, that same technology can do considerable good, provided it is used to produce clear, accurate, succinct signals that people can understand.

Here’s a project I challenge budding data scientists to take up. Working with historians, analyze speeches, newspaper articles, letters, political commercials, and other artifacts from the rise of previous fascists and autocrats. Using natural language processing, text mining, and clustering techniques, create a model that helps characterize and identify fascist tendencies. Train that model using the fascist-era data set as well as similar artifacts from the eras of non-fascist leaders to enable it to predict as reliably as possible which current or potential leaders most clearly exhibit fascist or autocratic tendencies and which do not. Design the predictive model to provide a very simple output: a percentage likelihood that, based on speeches and writings and decrees and question-and-answer sessions, this person is or is not a fascist-leaning leader.

In other words, create the Fascism Index.

Provided that people pay attention and care about the rise of fascism and the decline of democratic values, such a model could be extraordinarily useful. The great value of history is that it takes the surprise out of most contemporary developments, as what happens now likely resembles what happened yesterday. Sadly, most of us know too little history or have compartmentalized it as a subject we studied back in school that doesn’t have much bearing on our lives today. In reality, you can’t overstate the value of history to clarify what we currently see. A well-constructed, thoroughly vetted, and reliable data model could produce an easy-to-understand number to make the meaning of what we are seeing abundantly clear.

That is, of course, until the autocrat starts screaming, “Fake data model! Fake data model!” Then, I suppose, we’ll need some other tool.

About Ray Klump

Professor and chair of Mathematics and Computer Science Director, Master of Science in Information Security Lewis University http://online.lewisu.edu/ms-information-security.asp, http://online.lewisu.edu/resource/engineering-technology/articles.asp, http://cs.lewisu.edu. You can find him on Google+.

Leave a Reply

Your email address will not be published. Required fields are marked *