Navigate to News section

The New American Blindness

How the data-driven political journalism of Ezra Klein, Nate Silver, and others, applies the techno-elitist values of Silicon Valley to flatten political reality

Sean Cooper
August 06, 2020
Photo illustration: Kurt Hoffman
Photo illustration: Kurt Hoffman
Photo illustration: Kurt Hoffman
Photo illustration: Kurt Hoffman
This article is part of The Decline of the American Press.
See the full collection →︎

“Attention—Attention—Attention,” the hallway loudspeaker booms above the heads of six city residents, these the first batch of arrivals installed in a defunct insane asylum. “The Government regrets having been forced to exercise with all urgency what it considers to be its rightful duty, to protect the populace by all possible means in this present crisis, when something with all the appearance of an epidemic of blindness has broken out.”

Thus begins the inept response by anxious officials to a contagious disease that soon consumes nearly the entire population in José Saramago’s Blindness, his 1995 allegorical dystopian novel. For the six gathered in the asylum, the newly improvised laws that sound out from the speaker codify the shift to a new social reality. “Leaving the building without authorization will mean instant death,” the official declares. Meals will be delivered to the door, garbage shall be burned in the courtyard. If fires spread on the grounds, no government help will arrive.

In Saramago’s story, just as for us, the entire population suffers from the effects of a disease, and in both instances from a form of blindness: If I cannot see you and how you live then I do not know you, and likewise, if you cannot see me then you cannot know me. Blind to one another, our once intact social bonds, stripped of their core meaning, disintegrate rapidly, and we spiral into abysmal suffering. Similar to other ongoing phenomena accelerated by the pandemic—the late-stage collapse of local American retail, of print and digital media, of higher education, of failed government agencies unable to provide basic protections to citizens—the affliction of this reciprocal blindness has spread rapidly and widely, like an outbreak in parallel with COVID-19.

Over the past four months I’ve tried to imagine the distant future and what, looking back to now, I will most vividly recall about the pandemic and lockdowns; increasingly I believe that the strongest memories of the lived daily experience will be tied to my overwhelming sense of cultural isolation. While I’ve lived through the quarantine with my wife and daughter, and appreciate the unique density of the time we’ve spent together, I can’t remember ever feeling a sense of being this socially afar from the larger American culture. Propelled by a deep anxiety to surface the larger forces that were shifting the ground beneath my feet, I kept up a near-constant collection of information for details relevant to our physical safety as well as to know how others had experienced this change—information, in other words, that might help shape my awareness of a shared reality.

Yet no matter how much more information I took in, or how much more effort I exerted to increase my proximity to others, I observed an inversely growing distance between myself and the thing I was seeking out. It would oversimplify my point to say that now, rather than people, what I encountered over the past four months was almost exclusively social media profiles run by people—but it serves at least as a first position.

Was there no recollection of what happened to Hillary in 2016?

Though there were some rare, welcome exceptions, it was striking, as I read the volume of essays, dispatches, and written documentation of personal accounts of the pandemic, or of any subject matter at all produced during this time period, how near universal the commitment was to a sensibility and point of view grounded not in how a person lived but rather in how a person could orient his or her representation of the world to accommodate social media. It was as if the tension and stakes of each piece were engaged above all else to please and strengthen bonds online, to anticipate a reader as she existed not through the text but through the representation of a reader online. Said another way, there was a hostage quality to the prose. The author, so often, seemed to have been coerced to write it by their social media persona, in the timid, ingratiating language one might use to satisfy a captor, in homage to allegiances and allies and shared political grievances as they exist in circles of affiliation on social media platforms. This was certainly true, to a degree, well before the pandemic, and it has always stood out the way content marketing or propaganda stands out; but all at once it was as if social bands of people making culture had disappeared en masse and been replaced by social media avatars.

Sometime last month, the media resumed making narratives about the presidential election. It was here, oddly, in the political coverage, and in the data journalism in particular, that the obliteration of the individual continued apace.

There were the usual suspects on the major news broadcasts and the leading figures across the data political outfits like Nate Silver running the numbers at FiveThirtyEight. On CNN, I saw clips of Harry Enten, the Whiz Kid, half-shrieking about the polls that projected Biden bludgeoning Trump in a landslide. The New York Times said Biden had consolidated the entirety of the left behind him in a united front. But it was June, and the election was months away.

Was there no recollection of what happened to Hillary in 2016? Worse, there was a flimsy toxicity to the return of the strictly political when everything else in the past four months had rapidly been tainted by cultural partisanship and polarization. The premise of a return to politics as usual didn’t make any sense. What was I even looking at?

A strange mood and sudden energy infused the media landscape in 2014. It was an odd time for anyone to express excitement about journalism. The industry as a whole had never recovered from the recession. Mass layoffs and newsroom contractions rose sharply that year as print-ad dollars were sucked up by Google and Facebook profiting millions on content they didn’t create. Magazines were dying, newspapers were dead, and editor and reporter apprenticeships became antiquated curiosities of marginal value for journalism’s new leaders. Among this narrow cohort there was enthusiasm, and understandably so, with some $300 million of venture capital seed money poured into FiveThirtyEight, Vox, and other digital-native outfits that promised sophisticated applications of technology were foundational for the advancement of journalism. Only a more savvy integration of web design, distribution, and analytics would transform the stale opinion and reportage of prestige outfits which could no longer make sense of the world to its audience.

Moving over from The Washington Post to launch Vox, Ezra Klein explained, “We were badly held back not just by the technology, but by the culture of journalism.” Much like the Silicon Valley monopolies that were ravaging traditional media business models, Klein sought to disrupt the notion of readers and advertisers supporting journalists who go out into the field to gather facts and information. Klein and his staff instead relied on social media and news reporting done by others to not so much break news but rather filter the gross sum of online content, repackaged in layers of hot takes and explanation. There was no embarrassment when the home page of Vox was made up of headlines built almost entirely upon the reporting done by other journalists. Indeed, such cost efficiencies were good for business. Reflective of the entities that funded their new endeavors, Vox and other data-centric digital media properties eschewed not just the tone but the reportorial disposition of their legacy outlet competitors, taking on the lean development sheen, beta-launch language, and iterative design values of Silicon Valley startup culture.

As something of a niche enterprise under the umbrella of this new genre of media, the rise of data-driven political journalism was no less influential on shifting the priorities and ambition for a field suffering an existential crisis. At the helm of his FiveThirtyEight blog and arguably the preeminent face of political data reporting, Nate Silver rose to national prominence on the strength of FiveThirtyEight’s accurate prediction of 49 of 50 states in the 2008 presidential election and all 50 states in 2012. In terms of polling accuracy and those who valued that kind of information it was an impressive back-to-back feat. Silver’s outfit was housed at The New York Times, then acquired in 2013 by ESPN, making Silver one of Time’s 100 Most Influential People and the author of a bestselling pop stats book.

Silver’s success argued for both the cultural relevance and profitability of mainstream data political journalism that presented horse race campaigns in a quasi-narrative mode of discrete winners and losers favored by socioeconomic groups of citizens. Even better, these models predicted the future. Seven or nine months before an election, a reader or a potential voter could log on to FiveThirtyEight, read a graph or blog post, and know with some degree of certainty that they were looking at a mirror which reflected back their image as one that could be safely identified within a definitive demographic body, with shared political feelings, and where small adjustments in shared opinion were tracked over time. Race, gender, and beliefs could be synthesized into a quantified unit of measurement, like a credit score or retirement fund.

Data political journalism promised a simplified, utopianist interpretation of a confusing world to demographically divided contingents of reader-voters—finding the signal through the noise, as the title of Silver’s book put it. It was a clean and clinical worldview that segmented human variability into margins of error while distilling the complexity of politics into a set of probabilities that corralled like-minded groups into echo chambers that could vote alike—and, of equal importance, be marketed to and sold on similar sets of products and services pegged to the data trails scrapped while readers were told how best to absolve themselves from anything like contradictory emotions, or a sense that their political views might extend beyond the bounds of quantifiable results from focus groups.

Particularly appealing for investors, data political journalism, much like Vox’s explainer journalism, dispensed with any of the burdensome overhead of traditional field reportage—travel and expenses, painstaking development of leads, spending time in places being written about, understating human perspective embodied by geography, layers of fact-checking and editorial shaping. “The conceit of ‘data journalism,’ at least as I see it, is to apply the scientific method to the news,” Silver wrote. With a vision sharpened by seven- and eight-figure investments, the new media mongers saw the capacity for data to become hooks which pulled heaven right down to earth, and served as field guides in this digital Eden of numerical analysis and statistical certainty that relieved anyone from their overwhelmed sensory experience of deciphering the chaos of a news and media environment radically altered by new technologies. If the disease originated in Silicon Valley, then so did its cure.

The phenomenal misunderstanding of Donald Trump’s 2016 presidential campaign by both the data political journalists and the prestige outlets that overvalued the news worthiness, cost reduction, and predictive power of polls and projections to explain the moods and feelings of a nation is today remembered as a singular failure. In fact, it was two failures, both of which were engendered by the same faulty tools and reportorial instruments. First, there was the belief that one can regularly predict the future. Enten’s analysis of the primary took on ingratiating tones of elite condescension by using his data to conclude that Donald Trump has “a better chance of cameoing in another Home Alone movie with Macaulay Culkin—or playing in the NBA Finals—than winning the Republican nomination.” That narrative of pending failure was substantiated with definitive FiveThirtyEight charts that placed Trump’s probability of winning the nomination at between 2% and 7% for months in the run-up, topping out around 13% as New Hampshire prepared themselves for their primary.

By way of an explanation, Silver said, “No, I didn’t predict that the Republican Party would lose its fucking mind” on Trump’s nomination. A prediction, Silver emphasized elsewhere, that wasn’t based on a statistical model in the traditional sense, but rather “‘subjective odds’—which is to say, educated guesses. In other words, we were basically acting like pundits, but attaching numbers to our estimates.” Scientific method indeed.

Though the widespread myopia across the political data journalism sector provided an opportunity for careful reflection upon the applicable social value of ceaseless quantification of people’s thoughts and feelings, FiveThirtyEight’s fellow travelers at The New York Times instead decided to double down on their callow, failed—yet appealingly cheap—methodology. “With Mr. Trump heading for the general election, news organizations will get a second chance to rethink how they approach the race still to come,” they said. The New York Times, FiveThirtyEight, and other prognosticators dutifully spent the next several months churning out thousands of data visualizations, charts, tables, and front page headlines that placed Hillary Clinton’s chances of winning anywhere between a 70% and 99% certainty.

For Enten, who was then at Silver’s FiveThirtyEight, the lesson learned here was one of tone and decorum. “We need to be a little more cautious of exactly how we present things,” he said about the 2016 race.

In his own evaluation of where the data analysis faltered, Silver emphasized that however poorly he imagined the future he at least didn’t do as badly as his former New York Times colleagues. “To some of you, a forecast that showed Trump with about a 30 percent chance of winning when the consensus view was that his chances were around 15 percent will self-evidently seem smart,” Silver wrote on his website.

I’d mostly put aside the 2016 race, and didn’t think of it again until the pandemic and how strange and unusual the return of the presidential campaign data journalism was in the context of the social-mediazation of daily written culture. On June 25 The New York Times ran the front page headline of a story written around a single survey: “Biden Holds 14-Point Edge Over Trump in a New Poll: Voters Signal Disapproval over President’s Handling of Pandemic and Protests.” Elsewhere, Trump’s path to reelection was deemed impossible, or nearly so. And perhaps it is—particularly if he’s measured by voters on his presidential fitness. Yet, in the middle of a historic global pandemic and unprecedented geopolitical volatility, there is plainly something off in the gravity and scale of this coverage.

The assuredness of Biden’s trouncing of Trump by pollsters and data quants several months away from the general election was of a confidence, I felt, that like much of the daily cultural coverage, programmatically sought to obliterate the individual, and neuter possibilities for idiosyncratic emotions. The point was to reduce human beings to abstractions that fit neatly into enclosed social media echo chambers, or pollster demographic projections—all to make people malleable to the whims of those who owned the data.

If nothing else media’s commodification of people as data sets and demographics assured audiences that the world was best explained not by writers who engage with lived experience but by people who spent all of their time in front of screens, like hedge fund managers and supply chain specialists. Generations spent developing the craft of news gathering or analysis could be distilled to their residual empirical essence in charts and probability snapshots that shared virally on Facebook and Twitter, where there was some scraps of ad money left over for publications.

Moving up from FiveThirtyEight to CNN as an analyst, Enten was on the national broadcast explaining that the latest statistics on Biden versus Trump left him struggling to communicate the severity of their implication. “There’s a huge wow factor to looking at these polls.”

“Right now [Biden’s] up by six, seven, and nine points in Florida, Arizona, and in North Carolina,” Enten said of key states that Trump won in 2016. “Simply put, John, [these] are huge, huge leads. This is a huge story.”

“Now, Harry, one of the things you’re going to hear … oh, it’s just one poll. This doesn’t mean anything” the anchor said.

“And truth be told, that would be a BS line,” Enten said.

Here I was, watching Enten sell the severity of a summer poll’s implication for a November election to a national audience during a pandemic lockdown. There was great value in Enten’s affectation, born as it was, I suspected, from Enten’s obsessive interest in old election coverage on YouTube and the cultivation of his anachronistic style. It was an attempt at something representative of political data journalism if seen in its most flattering yet entirely inauthentic light, a resurrection of an idea of an objective shared reality that exists as a faint echo of a midcentury moment when Cronkite and others patiently and paternally rendered the happenings of the day into what the audience believed to be coherent understanding of the world which resonated across a hundred million living rooms and kitchen tables.

Push beyond the nostalgia and one finds the platonic ideal of political data journalism taken to its logical extreme. Here, finally, political prognostication could be 100% right every time. With enough historical data points and computation modeling power we should strive to envision the final televised broadcast of any presidential election, with the results all in, which then segues into the cable news studio’s data team explaining that four years later the incumbent or the challenger of a particular race, sex, age, and political track record was without a statistical doubt assured to win the following election. Like Deleuze’s bodies without organs, citizens would be alleviated, at long last, of the nuisance of having to worry about what they see.

Sean Patrick Cooper is a journalist who has contributed narrative features and essays to The New Republic, n+1, Bloomberg Businessweek, and elsewhere. His first book, The Shooter at Midnight: Murder, Corruption, and a Farming Town Divided will be published in April 2024 by Penguin.