1
Fearsome Contraptions

Nobody saw it in real time. How could they? The history of American
elections changed profoundly on the night of November 4, 1952. The
candidates were at the center of attention, of course, but they were not
central to this change. We can see its significance now with decades of
hindsight. Something never tried before was launched live and untested
on national television. Viewers witnessed an outside-the-box approach
to predicting the winner from early returns using something new—
computers. Like exhibits in a freak show, the computers were referred
to as “electronic brains” and “mechanical monsters.” That’s the way it
can be with innovations—considered laughable before they become so
locked-in that we cannot imagine life before them. This innovation would
help fuel an obsession with numbers as a way of understanding—and
shaping—politics. It would also engender controversy right down to
our own time. And it would herald a future in which the public square
would go digital. Yet on that night, this path was in no way guaranteed
or even envisioned. Americans did not yet have computers at home or at
work. Most had never seen one. The thinking in broadcast news circles
was that the flashing lights of these newfangled gizmos as they spit out
forecasts might provide an edge. Television news executives hoped to
get a jump on the competition both in calling the election and attracting
viewers. But the whole enterprise of reaching outside the box with
something so new was an enormous gamble. And it is important to
understand—especially in our own time—that this gamble was under-
taken in response to a crisis of credibility. This crisis was the fallout from
election-night forecasts four years earlier that turned out to be upside
down. What’s more, a lackluster presentation of returns on television
in 1948 was at odds with television’s promise as a new visual medium
for news. What transpired four years later, in 1952, is a complex tale of
innovation in response to crisis, along with reactions that ranged from
enthusiasm to ridicule to outright resistance. The events surrounding
that election night in 1952 also offer us a way to think about another
more recent inflection point—the vexing events of 2020—and the future
of election nights in America.

At 8:00 p.m. on election night in 1952, cbs launched its coverage
from a cavernous studio above Grand Central Terminal, the massive
rail hub in Midtown Manhattan. A camera perched aloft on a swivel
in the middle of the studio slowly surveyed the scene and zoomed in
toward the anchor desk. The zoom lens itself was a recent invention.
Viewers found themselves immersed in the hubbub. Walter Cronkite
was anchoring his first election night on television. The audience could
see for themselves what he described as a teeming beehive. The studio
was packed with people, cameras, cables, telephones, typewriters, adding
machines, and all manner of other equipment. A massive banner that
read “cbs Television Election Headquarters” was installed along one
wall. The click-clacking of teletype machines signaled a steady flow of
wire service dispatches delivering a rising mass of vote counts. Beneath
the banner, and in view for the first few seconds, was an exotic-looking
device about the size and shape of an organ console. It had a keyboard
and rows of blinking lights. Viewers would learn later in the broadcast
that this was the so-called “supervisory control panel” for a Universal
Automatic Computer, known simply as univac, one of the pioneering
computer models in a quickly growing field. The control panel was not
actually connected to a univac computer. Rather, the panel had been
installed on the cbs set as a prop. Its blinking lights were sleight of
hand, produced by something akin to a Christmas tree light circuit.
The actual UNIVAC was a behemoth. It weighed in at more than eight
tons, too big to move to cbs for the broadcast. So it would be generating
forecasts from the place where it was built, one hundred miles away
in Philadelphia. Tensions were running high behind the scenes at the
univac home base.4 And the stress was only going to get more intense.
The first predictions soon generated by the computer would be so far
from the expected outcome that they would not be released to the network.
Instead, the computer’s keepers would scramble for a solution to
this mystery. How could their eight-ton baby be so wrong, they wondered.
The UNIVAC programmers and engineers had built redundancy
into their systems to guard against errors in data entry and calculation.
But they had not considered what they would do if the computer’s first
forecast diverged wildly from expectation—as, in fact, it did.