Jacob Javits of New York is the first United States senator to become fully automated,” the Chicago Tribune announced in 1962 from the Republican state convention in Buffalo, where an electronic Javits spat out slips of paper with answers to questions about everything from Cuba’s missiles (“a serious threat”) to the Cubs’ prospects (dim). “Mr. Javits also harbors thoughts on medical care for the elderly, Berlin, the communist menace,” and more than a hundred other subjects, the Tribune reported after an interview with the machine. Javits may have been the first automated American politician, but he wasn’t the last. Since the nineteen-sixties, much of American public life has become automated, driven by computers and predictive algorithms that can do the political work of rallying support, running campaigns, communicating with constituents, and even crafting policy. In that same stretch of time, the proportion of Americans who say that they trust the U.S. government to do what is right most of the time has fallen from nearly eighty per cent to about twenty per cent. Automated politics, it would seem, makes for very bad government, helping produce an electorate that is alienated, polarized, and mistrustful, and elected officials who are paralyzed by their ability to calculate, in advance, the likely consequences of their actions, down to the last lost primary or donated dollar.