Introduced March 1951
The immediate post-war years saw the first viable commercial and scientific computers hitting the market, such as the LEO I and Ferranti Mark 1 in the UK and the UNIVAC I in the United States. These huge, heavy, power hungry, expensive and slow computers are very primitive by today’s standards, but they found their niche in corporations and laboratories.
|UNIVAC I in use at the US Census Office|
Rather like the British LEO I, the UNIVAC I was a business computer made up of around five thousand vacuum tubes, weighed about 8 tons and sucked in 125kW of electricity. Primary memory consisted of 1000 12-character words stored in mercury delay lines (similar to the LEO), but one novel feature the UNIVAC had was magnetic tape drives for secondary storage. These UNISERVO I drives were the first ever commercially available computer tape drives, and they used heavy nickel-plated phosphor bronze tapes. Other rival computers tended to use punched cards, and initially the UNIVAC I lacked this option.
It was an eye-wateringly expensive system, costing the equivalent of around $7 million in today’s money. This tended to put it out of reach of many universities and into the realm of government departments and large corporations instead. Given the price and complexity of the UNIVAC I, they wouldn’t always go straight to the customer but would instead be used for a while as demonstrators.
|UNIVAC I mercury delay line memory|
Once such UNIVAC I was bound for the US Atomic Energy Commission, but took a detour to detour to CBS to help predict the results of the 1952 US Presidential Election. Programmed by the legendary computing pioneer Grace Hopper, the UNIVAC had a complex model built up of data from past elections. With a sample of 5.5% of the vote, the UNIVAC came up with a prediction – out of the two candidates of Dwight D Eisenhower and Adlai Stevenson, Eisenhower would win in a landslide. The problem was that the UNIVAC’s prediction was completely at odds with pollsters who were predicting a win for Stevenson. So the UNIVAC’s analysis was downplayed by CBS.
Even if you don’t follow US presidential elections of the 1950s, you’ve probably heard of President Eisenhower and not President Stevenson.. that’s because the UNIVAC turned out to be accurate and Eisenhower did indeed beat Stevenson, by a similar margin to the computer’s prediction. Eisenhower might have won the vote, but the UNIVAC won when it came to the star of the election night count.
|Grace Hopper working on a UNIVAC I console|
Demand for UNIVAC machines boomed… but Remington Rand were struggling to build them. So ironically, it was rivals IBM who actually benefitted with their IBM 701 mainframe and its successors as they could build them in the quantity customers wanted. As for Remington Rand, they were taken over by Sperry in 1955 which in turn merged with Burroughs to create Unisys in 1986. Unisys is still around today, and it still makes computer hardware such as the Intel-based ClearPath Forward systems among a muddle of consultancy services and resold products.
Although the UNIVAC I was only a moderate success in sales terms, it is socially significant for its role in the Presidential Election where it offered a glimpse into the future of computing, only seven years after the end of the Second World War. It’s astonishing to think that all this was still more than a quarter of a century before the launch of micros such as the Apple II…
U.S. Census Bureau via Wikimedia Commons – Public Domain
Tiia Monto via Wikimedia Commons - CC BY-SA 3.0
Smithsonian Institution via Wikimedia Commons - CC BY 2.0