Sunday 5 February 2023

Jill Lepore: IF THEN

I was lured in by the initial impression this book would have stories of old-timey punch card computers and their operation. Not really. This is more of a political story revolving around computers and their application in behavioral science-led questions. And all the better for that I think.

If Then: How One Data Company Invented The Future, from the historian Jill Lepore, weaves together politics, university life, zeitgeist and popular culture, in an attempt to explain how computer prediction and data analysis came to impact on politics and public perception in the 1960s.

The prologue and backdrop is the famous event of televising the 1952 presidential election, doubled with a televised UNIVAC prediction of the outcome. For some inventive minds, the event suggested that instead of using computer to report vote results faster, a computer might influence who could get elected.


Through the era of Dwight Eisenhower, John Kennedy, Lyndon Jonhson, Vietnam, Martin Luther King, emerges an unlikely narrative of the Simulmatics Corporation, a company that used behavioral science-based computer modelling to predict people's actions. A "people machine". 

The Kennedy election in 1960 was followed by a bold claim from Simulmatics that their reports, based on number crunching and 480 voter types, resulted in policy suggestions that landed Kennedy the presidency. Through mist of assertions, denials, counter-denials, fiction and non-fiction, it is probably impossible to say what the real impact was.

Among the cast of characters, the author details the involvement of Eugene Burdick, now a largely forgotten Hollywood-celebrity author of political thrillers with a sci-fi bent. His book The 480, was a fictional account of Simulmatics, and not from a neutral perspective either. 

Although the claimed influence resulted in a sort of a scandal in itself, the novel probably ignited more debate and public opinion about computers than anything else. The fear of computers predicting what people will do and government reacting to it before people themselves knew what they were about to do, seemed to become real rather soon. Whatever the truth, the idea of a candidate led by a computer entered into the popular culture big time, reducing voters into punch-card caricatures with no agency of their own.

Simulmatics banged their drum with the Kennedy report, also basking in the (somewhat indirect) infamy of The 480, but didn't have it easy with subsequent clients. Politicians and the administration were understandably not too keen to rely on the company, and other industries didn't have that great data or didn't want to give it away. 

Rather quickly, advertisement companies with long history and large databases could do their own computing, and Simulmatics was left in the sidelines in the field they helped create.

Vietnam war provided a life extension for Simulmatics, as they set up shop in Saigon and attempted to collect data for predicting insurgencies and find opportunities for propaganda and soft warfare. Simulmatics was hardly alone in this, but at least in this telling it looks theirs was the most incompetent and shoddy research setup, an exemplary of how not to conduct social research in another country.

In meantime, administration had changed and Nixon hardly had use for an outside predictor company, as voter segmenting and balancing segment responses to policy changes with estimated votes, were already old hat. (It's unclear whether Simulmatics offered to politics anything new in the first place, except more speed and "real time".)

The book then continues exploring the latter days of Simulmatics, its entanglement to the civil right questions and the black vote issue, a supposed linchpin of the Simulmatics report for Kennedy. Back in USA, Simulmatics engaged in predicting not rural insurgencies in another country, but riots in the urban areas. In fact it seems "predicting riots" was something of an ongoing sub-genre in the area of behavioral science applications.

The author is then able to weave and knot the story of the corporation to the moment when the two political parties began to polarize in the way we now recognize. This is perhaps more zeitgeist represented by the story of Simulmatics, rather than Simulmatics or computers being a direct root cause for it.

What strikes as both funny and terrifying is the unprincipled nature, even corruptness of it all. In absence of data, Simulmatics would cobble up some, patch it with educated guesses, then present results to clients. Again, this speaks likely more about the time than about Simulmatics or the field of computer-based behavioral science. But I can’t avoid having flashbacks to university innovation and research activity.

Of the central cast, to me the only relatively familiar figure is Ithiel de Sola Pool, whose name became tarnished through the Simulmatics/Vietnam association. MIT students protested vocally against defense funding of research projects. The opposition was even extended to the openly presented and rather innocent Arpanet, the consequences of which hardly anyone understood at the time.

Here these two developments at least touch each other in history. Simulmatics, representing the more graspable idea of government control over people through mainframes, and "reducing people into punchcards", was still a name brand at these debates. It was soon forgotten from history whereas the developments that grew out of Arpanet grew into global importance.

Later, Pool was able to predict in great accuracy the consequences of a world-wide internet: in principle anyone can have personalized newfeeds and therefore live in reality-bubbles of their own, or get behind fringe ideas by networking in ways that were impossible before.

Recounting the legacy of Simulmatics, Lepore appears to liken Media Lab as a continuation of the practice of milking government money for grand promises of technical revolutions, a "Simulmatics scheme" of sorts. Not sure if I got this correctly, or perhaps this is a case of Harvard bashing MIT or something.

Looking Back

The book makes the always gleeful and enjoyable demonstration that a technology we thought was new, was actually used a long time ago and largely abandoned. Perhaps even for moral and ethical reasons. And then as time passed it came to be done anyway, by the Cambridge Analyticas and the Facebooks of our age.

This requires the insight that the mainframe-as-predictor, and the large-scale internet practices of user data manipulation can be juxtaposed, and to make say something about each other. What could have been a historical footnote becomes more important through the lens of the present.

Apart from mentioning IBM computer models and bit about FORTRAN and punch card tech, the hardware is not meticulously explored. The algorithms are simplified into kind of IF-THEN evaluations (hence the title), so as not to confuse the more casual reader. Then again, the US election system is not explained in any great accuracy either, so the text is able to cast an overview of a web of events without boring anyone greatly.

From today’s perspective, Simulmatics was a fledgling attempt at best, and would seem naive if it wasn't acted out against the backdrop of presidential campaigns, civil right questions, the Vietnam war, and a world on brink of nuclear destruction. Certainly the story is a precursor to the data analytics practices later embraced by any large-scale commercial or political internet operation.

It's also darkly funny, this story of the would-be Merlins in Kennedy's collapsed Camelot. The writing is infectious, making me feel like I'm suddenly an expert in areas I had no knowledge of prior to reading this book. Perhaps the author also borrowed a leaf from Eugene Burdick's supposedly lurid and unsubtle prose, as if to show the pen is mightier than the punch-card after all.

No comments:

Post a Comment