How Computers Started Predicting Election-Night Results

A 1959 photo of the Universal Automatic Computer (UNIVAC), one of the first used to predict election-night outcomes. CREDIT: Courtesy of the Library of Congress, LC-USZ62-118471 View full size image

One of the world's first computer-aided predictions of an election outcome came with an oddly human voice. "IT'S AWFULLY EARLY, BUT I'LL GO OUT ON A LIMB," read a 1952 printout from the Universal Automatic Computer (UNIVAC), which the U.S. Census Bureau called "the first successful civilian computer." 

UNIVAC ultimately predicted President Dwight D. Eisenhower's landslide win against Illinois Gov. Adlai Stevenson after Americans had cast about 3.4 million votes. In an Oct. 31 segment, National Public Radio told the stories of UNIVAC and other early computers used to forecast U.S. presidential elections 60 years ago. That same year saw the first nationwide TV broadcast of election night, NPR reported. 

Now, computing divining is a staple of election-night news on TV and in other mediums, but at the time, news stations were wary of UNIVAC and its compatriots' numbers. While the machines were reliable at crunching numbers, the math equations they used for their predictions were still very simple, NPR reported. UNIVAC was a "sideshow" for CBS in 1952, while in 1956, ABC ran a "Man versus Machine" contest that pitted a pollster and the network's reporters against a computer called Elecom, made by Underwood Corporation. The humans won. 

Things changed in 1962, when CBS hired Lou Harris for that year's midterm election coverage. Harris was a pollster who worked with IBM to create election prediction software. His computer-generated forecasts allowed CBS' Walter Cronkite to call races earlier than any other news organization.

"This was a huge triumph for the then newly created CBS News election unit," Martin Plissner, formerly the political director for CBS, told NPR. 

Since then, newsrooms of all kinds have relied on computer predictions.

Source: NPR's "Morning Edition"

This story was provided by TechNewsDaily, a sister site to LiveSciene. Follow TechNewsDaily on Twitter @TechNewsDaily, or on Facebook.