The photograph jumps from my timeline, instantly familiar and banal. A low-res satellite image, made of drab dappled greens, mottled browns, and a full palette of grays, cut with a digital jigsaw and set against a flat field of white. Census Tract 20, DeKalb County, Illinois. Barely a title, cryptic and blunt.
Click through and there are thousands more. Census Tract 304.01, Columbia County, Georgia. Census Tract 2264.10, Los Angeles County, California. Census Tract 9558, Trego County, Kansas. The Twitter bot @everytract has posted one picture like this every half hour since the morning of February 20, 2018. Working its way through the nation’s 74,134 census tracts, the bot should finish up around the time of the midterm primaries in 2022 — the first elections to be held in new voting districts based on the census now underway. But @everytract is not interested in electoral politics. Not explicitly. What it offers, instead, is line and shape and color, and a common place where we can gather to share these streaming mysteries.
Working its way through the nation’s 74,134 census tracts, the Twitter bot @everytract should finish up around the time of the midterm primaries.
Follow the mysteries back far enough, from screen through internet connection, deep into the place where pictures unspool into lines of script, traveling up a stream of electric ones and zeroes, over wires spun of copper and glass, and you will eventually reach the keyboard of Neil Freeman — New York City artist, urban planner, and prolific architect of Twitter bots. 1 His @everylotnyc sends out Google Streetview photos for every property in the city’s tax records, alongside sibling accounts in Chicago, San Francisco, and Los Angeles. Another bot, @buoypix, tweets views from 55 buoys owned by the National Oceanic and Atmospheric Administration. Then there’s @combinedsewer and @therealgowanus and @newtown_creek, reporting sewer overflows and industrial tide tables in New York. And @obligatoryair, which retweets posts containing the phrase “obligatory airport tweet.” There are more — somewhere around 40 bots in total, which together have tweeted well over half a million times. 2
Freeman is a virtuoso of the archive; every line of text, every image has to be sourced from somewhere. A generous one: following the utopian ethic of many hackers, he documents his scripts and shares his tools on GitHub. 3 But, above all, Freeman is a seer, the sort of person who looks at the commonplace and finds, just beneath the surface, ordered wonder. Did it ever occur to you to visit the geographic center of New York City? (It’s somewhere in Bushwick, just beneath the J train; Freeman can take you there.) 4 Did you know that there are 19,468,490,047,488 unique four-color combinations of the 50 United States? (Freeman’s website, Fake Is the New Real, will display every one of them, but only once.) 5 That if you lay out every London street alphabetically on a grid, you get an extraordinary picture that looks like a felted elbow patch, but if you do it for Los Angeles you get a circuit board? 6
“The world,” writes Sarah Sentilles, “is made. And it can be unmade. Remade.” 7
Freeman’s map Every US local government explodes the country into its most basic political components, while works like SP weather report; All the streets, centered; and Circled states find geometric patterns in abstracted data, all empirically gathered in some real-world place, now reshuffled by a logic that reveals startling beauty. To gaze upon this clairvoyant madness is to see, just for a moment, the ley-lines of a different world cohere. 8
Apart from the philosophical and political questions that confronted the framers of the United States Constitution was a practical one: who lives where? Without an answer to that question, without hard data, there is no hope for a representative government, and so a decennial census was prescribed at the top, just a few lines down from “We the People.” The census plays a role in apportioning federal funds, it determines each state’s number of congressional Representatives, and it helps mark who is and who isn’t “American.”
The first survey, in 1790, sorted the country into broad categories: free White males (minor and adult), free White females, other free persons, and slaves. American Indians were left out. Totals were reported haphazardly, at the level of state, county, or town; there were no census tracts. A century later, the survey had grown to over 30 questions, covering race, childbirth, home ownership, literacy, and disability, and the results filled 21,410 pages, a triumph of Progressive-era social statistics. 9 Yet still, the picture was blurry. The government wanted a closer view of the cities and towns (and even the blocks) where Americans made their lives, to see, intimately, how the population changed, which neighborhoods thrived and which did not, who was healthy and who was miserable; all of this at high resolution, with high fidelity — to see precisely who lived precisely what sort of lives where, precisely.
As typhoid fever swept the Northeast in 1906, Walter Laidlaw, director of the Population Research Bureau of the New York Federation of Churches, advocated for the creation of small, permanent census tracts, to monitor the spread of disease. He was persuasive enough that, for the 1910 survey, officials tested his idea in New York and seven other large cities. In the following decades, census tracts sprouted in city after city, mapped by local organizations in partnership with the Census Bureau. By 1934 the feds had outlined standard criteria. Each tract should have a population between 3,000 and 8,000, and be defined by stable physical boundaries: rivers, railroad tracks, the centerline of streets. (These criteria are largely the same today, although the lower population bound has been changed to 1,200, to better fit rural areas.) And each was to contain “a population reasonably homogenous both as to racial characteristics and as to economic status or type of living conditions” — a bright baseline against which change, and especially the mixing of race and class, could be measured. Once a tract surpassed 8,000 inhabitants, it would be subdivided. After staking out the cities, the Census Bureau turned to suburban and rural areas. It was not until the 2000 survey that the entire United States was divided into census tracts, those obscure, powerful units which most of us cannot locate, but which structure our lives. 10
I write at night, after my day job is over. After I’ve taken my turn at cooking dinner and helping our oldest child with homework and reading to our youngest and sending them both off to bed, after the precious hour my wife and I share once we have finished the night’s chores, after she heads to our bedroom and turns out the light. I’m tired when I finally settle in to write, in tract 415.01, Franklin County, Massachusetts, and too often distracted, or else looking for distraction, so I head over to Twitter, looking for something I know will never be there.
One especially agitated night, I found myself staring at a picture of some suburban somewhere: Census Tract 111.14, Yuma County, Arizona. I didn’t know yet that I had landed a dozen miles north of the U.S./Mexico border, in the Fortuna Foothills neighborhood, which springs from the dust on the edge of an Air Force Range just beyond the Gila River’s green coil. I didn’t know that you could see the black Fortuna Hills rising up beyond RVs parked in the driveway of mobile home lots, that the average housing cost is $301 per month, that 83 percent of the population drives to work. 11 None of this mattered to me. I had been drawn in by the photograph’s beat, its alternating colors, never quite regular, but not chaotic either: white, tan, brown, a southwestern Mondrian, not Broadway Boogie Woogie, but Corrido Fortuna.
I clicked to see if there was anything behind the timeline, and the image unfolded into a rhythmic silicon stele stretching across my screen. The major axes brought air into the picture, while hundreds of tiny capillaries shuttled my eye around this photograph with no center, no focal point, no sense of perspective, no place to begin or conclude. It was a very late night, and I was supposed to be writing, but instead I spent my time here, in Yuma.
Census tracts do not describe a territory, nor are they really about people. They are statistical frames filled with facts, pinned to the land to give form to columns of demographic data. We rarely see them mapped. But tract boundaries do exist; they are published as an enormous digital folder of shapefiles that report the longitude and latitude of points on the edge, accurate to within a millimeter. 12 To make @everytract, Freeman purchases satellite images from a commercial provider, then slices them according to those shapefiles. And here, at the moment of inscribing a line, is where the photograph loses the precision of the purely statistical. Where it begins to blur. “The question of where a line is,” Freeman told me, “is what I’m very interested in, a line in a piece of map data.” 13
Here, at the moment of inscribing a line, is where the photograph loses the precision of the purely statistical.
Every blade has a kerf. Run a saw through a plank: the space lost to sawdust, the void created, is the kerf. Since Freeman is working with pixels — square units of color that cannot be cut — his digital blade has an eccentric kerf, slicing a ragged path along the edge of a census tract. And since the size of a tract is determined by its population, there are extreme differences in scale. The 147,805 square miles of Alaska’s Yukon-Koyukuk County are covered by four census tracts, while the 33.77 square miles of New York County are covered by 288. When those images show up in my timeline, a single square brown dot of Alaskan earth might represent 50 acres, while a gray dot of Manhattan is only 5 square feet. Digital sawdust floats down to Freeman’s feet, the piles deeper for rural America, though the result is the same no matter where he cuts: something real is lost.
I asked Freeman about the kerf of his blade, about how closely his pictures match the tracts defined by the Census Bureau. “I can’t show you a census tract,” he replied, “because a census tract is. … I don’t know what a census tract is.”
In 1958, when Hannah Arendt bent her mind toward “understanding the nature of society” in the modern age, she began with the launch of Sputnik, the world’s first artificial satellite. That event — more important, Arendt thought, than even “the splitting of the atom” — was met not with a sense of pride or awe, but with “relief about the first ‘step toward escape from men’s imprisonment to the earth.’”
The relief worried her. In The Human Condition, Arendt traces a history of decline where, by degrees, a world-weary public embraces an automated, predictable, curated existence, a life of convenience and comfort and novelty, and so becomes vulnerable to demagogy. It’s a history of retreat from engagement with the earth, with one another, and with each of our own creative capabilities; a history that ends in the triumph of a technocratic society, sophisticated in its ways of acquiring data, solving equations, and engineering solutions, but which lacks the capacity for reflective thought and deliberative action, a society that values only labor and production and consumption — the physical processes of life, merely. “It is quite conceivable that the modern age … may end in the deadliest, most sterile passivity history has ever known,” she wrote. 14
A satellite rising. This is the age-old dream of flying high above the messiness of pedestrian life; a dream of achieving the total, perfect view, from god’s own vantage point. Geographer Denis Cosgrove calls it the “Appollonian eye,” “a vision of unity … synoptic and omniscient, intellectually detached” that stretches back to Plato, at least. The very conception of a globe bears within it the ideology of globalism, the notion that the earth and everything upon it belong to one universal system. 15 If we could simply see this, if we could engineer our society according to aerial truths, we’d be freed from the responsibility of judgment and the risk of error.
Of the 2,666 operational satellites in orbit today, roughly a third (that we know of) are dedicated to earth observation.
Sputnik was the first step toward that point of detachment, although it transmitted radio beeps, not pictures, back to earth. The great human leap to Appollonian vision occurred a decade later, when astronaut Bill Anders made the iconic photo Earthrise. In our time of Google Earth, this view is familiar, even comforting, but in 1968 it was revolutionary. People thought it would change the world. “Men’s conception of themselves and of each other,” wrote the poet Archibald MacLeish, on the front page of The New York Times on Christmas Day, “has always depended on their notion of the earth.” Until Anders tripped his camera’s shutter, human vision had been constrained by the horizons of state and nation, with catastrophic effects; now there was hope. “To see the earth as it truly is, small and blue and beautiful in that eternal silence where it floats, is to see ourselves as riders on the earth together, brothers on that bright loveliness in the eternal cold — brothers who know now they are truly brothers.” 16
Of course no picture will change the world; only we can do that. Readers today may cringe at MacLeish’s naiveté, as the “riders on the earth” settle in to watch a globalized spectacle of satellite-driven communications: Netflix and Sirius, remote learning and telemedicine, embedded war reporting and drone flights and self-driving cars. Satellite vision is a $300-billion-a-year business, more or less. 17 It is also an instrument of power. Of the 2,666 operational satellites in orbit today, roughly a third (that we know of) are dedicated to earth observation, including hundreds owned by militaries, intelligence agencies, and private security contractors. 18 Until the 1990s, most earth-observing satellites belonged to governments, and access to high-resolution images was strictly controlled. But since President Clinton privatized the American satellite industry, Appollonian vision has been available to anyone who can afford it. 19 Today, the best commercially available imagery has a resolution of 30 to 40 centimeters per pixel, but it’s expensive: a 179-square-kilometer photo of New York City, for example, costs $2,866 from one vendor. 20 At that rate, a complete set of images for the United States would run more than $157 million dollars.
Yet we are not our technology, and our lives needn’t be reduced to data. “The world’s potential salvation,” Arendt wrote in On Revolution, “lies in the very fact that the human species regenerates itself constantly and forever” — humans are born beginners. To retain our “natality,” our capacity to take action and shape the world, we have to resist the lullaby of Sputnik’s radio beeps. And so the political theorist turned to art and poetry, where she found the “condensed concentration” necessary for beginning anew. “The storehouse of memory,” Arendt wrote, “is kept watch over by the poets, whose business it is to find and make the words we live by.” 21
For a couple of days last September, @everytract went crazy.
There’s surely a technical explanation for the aqua haloes irregularly cut around Hawaiian islands, maybe an error in the retrieval of satellite images, or a glitch in the way Freeman’s bot interpreted the shapefiles. There’s an answer to be found, if one wants it. But I’ve never been the sort of person who cares to know how the magician makes her assistant disappear.
Besides, Alaska awaits. These are my favorite works in the timeline: miraculous pictures shimmering with an austere mystery, where the landscape diffuses into the white of oblivion, leaving me to wonder what’s there and what’s not.
When I interviewed Freeman for this essay, I was struck by his mode of speech, by its skittering precision and careful finality. Long pauses interrupted each clause, and then words tumbled out with tremendous economy, as if he had surveyed thousands of linguistic possibilities to find the exact fit. His speech was concentration condensed into the space of a syllable, and I, a wanderer by nature, found myself listening intently, straining to keep pace.
I wasn’t sure what questions I had anyway. In the months since I had begun following @everytract, my expectations had been thwarted, consistently. The photos were factually explicit, seemingly self-evident, yet they refused to hint at what they meant or even how they were to be understood. “One of the most common kind of responses,” Freeman told me, “is something along the lines of, ‘that looks like a duck,’ or ‘that’s a man with an ax.’ It’s the same pleasure of looking up at a cloud and guessing the shape, and saying the shape of the cloud.”
Outraged at Trump’s distortion of the census, I wanted the comfort of seeing my anger reflected here.
Other artists have used satellite images to make work with an explicit social edge: James Bridle’s Dronestragram depicts landscapes where American drones have executed human targets and bystanders, and Mishka Henner has framed toxic feedlots and oil fields so that their visual beauty clashes with the evidence of violence. These photographs “torment visibility,” as Teju Cole put it; they distort how we see, calibrating our vision to register exploitation, desecration, and murder. 22
The politics of @everytract are harder to discern. For a long time, I tried to read the account as subversive commentary on the superficial notion of democracy-as-spectacle. Then I dismissed it as good-natured geekery, in the vein of Freeman’s @longestword, which spent a year tweeting the full chemical name of the protein titin (it has 189,819 letters). Still later I wondered if it was all an elaborate put-on. “A Rorschach test,” Freeman told me. “A psychological game. And games like that are meant to tell you about yourself.”
When I started following @everytract last summer, the Trump administration was fighting to add a citizenship question to the census, as part of a scheme to disenfranchise immigrant communities. On June 27, the ruling came down from the Supreme Court: There would be no citizenship question on the 2020 census. Except that Trump defiantly announced, via Twitter, that there would be. I remember sitting in my car, a mile from our house, when the news came on the radio. The rickety bridge over the Connecticut River, linking Turners Falls to Greenfield, Massachusetts — tract 407.01 to tract 411 — was under construction, down to one lane, and so I idled behind a dozen other vehicles, listening to a debate about population and density and citizenship and data. “The News Reports,” tweeted the President, “about the Department of Commerce dropping its quest to put the Citizenship Question on the Census is incorrect or, to state it differently, FAKE!”
On that same day, @everytract was tweeting out pictures of Bibb County, Georgia, a solidly Democratic county in a purplish-red state. Since it is also a predominantly Black county, it has been targeted by the state’s Republicans as they purge voters from the rolls. 23 But I didn’t see any of that when I looked at @everytract. Outraged at these episodes of racist authoritarianism, at Trump’s distortion of the census, I wanted the comfort of seeing my anger reflected here, but all I saw were baseball diamonds and running tracks and railyards, forests and right-of-way clearcuts and, I’m pretty sure, a wastewater treatment pond. I wanted politics, but got the prosaic: a census tract shaped like a bird’s beak (one follower replied with a gif of a pelican eating a fish) and another that pictured a cemetery (which someone said holds the graves of les brers Duane and Gregg Allman).
There are pictures that torment visibility, but there are also, Teju Cole wrote, “images that use clarity in an ambiguous way.” 24
The pictures that erupt into my timeline live in a context unique to me, surrounded by other tweets, arranged at the algorithm’s pleasure. Because meaning leaks across each picture’s frame, no two followers of @everytract will see the same thing, not really.
“Where is meaning created?” Freeman asks. “Is it created by an author? Is it created by the author in collaboration with an audience?” Despite @everytract’s superficial clarity, its automated aura of objectivity and digital precision, the pictures refuse to show much at all. After following the account for a year, I’m still not sure if Freeman’s intent is puckish or punk.
My inclination is to see the pictures aesthetically — even, as Susan Sontag put it, erotically — attuning myself to how they strike my senses; to ignore what they represent in favor of what they are: line and pattern and color and shape. 25 It’s the fern-like structure of tract 9501, Bracken County, Kentucky, that attracts my eye; the vivid angularity and burnt dust of tract 9660, Seward County, Kansas. There is a common beauty in the way they exceed the limits of data and so condense my concentration.
Some followers, especially in the planning community, thrill to the pictures as evidence. They reply to the tweets, identifying industrial facilities, infrastructures, and housing types, discussing the beauty or folly of certain landscapes. They consult Freeman’s density.website — How close to each other should people live? How close to each other do people live? — to glean tract-level data on vacancy rates, building age, housing cost, and population density, and report back to Twitter with their kernels of knowledge.
Others find an excuse for whimsy and speculation and introspection: pointing out men with axes, or dicks (of which there are a surprising number) or ducks or deer or dragons or donkeys or dogs or kale. 26 There’s something of a game to see who can tag which geographic feature or pick out which minutely rendered oddity. There’s humor and snark — especially when the bot blew through the Sunshine State (“Dear Florida, you’re waaaaaaaay stranger than I could have ever know and I follow @FloridaMan,” wrote one fan) — and candor, too.
Photographer Daniel J. Schneider has written an autobiography, one tweet at a time, keyed to 38 posts from Arapahoe County, Colorado, tracts he’s known throughout his life, in paratactic prose:
I’ve lived in New Mexico, New York, Vermont, Maine, Wisconsin, and Massachusetts, none of which @everytract has gotten to yet, and travelled through most of the rest of the states, often on back roads. Lately, my favorite way to read @everytract is to click on the location metadata, which takes you to all the other tweets from that place. It’s a jump into a community that isn’t mine, a digital proxy for going to the local diner or park or bar.
In Gonzales, I saw babies and smiling parents; in Fortuna Hills, a pregnant scorpion and a dog with a tail growing out of its forehead.
In Gonzales, Louisiana, I saw babies and smiling parents; in Fortuna Hills, young couples, a pregnant scorpion and a dog with a tail growing out of its forehead —“a fucking unicorn,” wrote one poster, whose profile picture and header featured camouflage, guns, and dead animals. Alaska brought me big-mountain landscapes, kittens, and a large man who specializes in posting pictures of himself wearing nothing but brightly colored undies. I’ve watched people react to becoming citizens, to voting in the primaries, to COVID and impeachment, to the triumphs and indignities that greet us from the moment we open our eyes in the morning to face the humor and pathos, magnanimity and selfishness, weirdness and terror and beauty of our daily lives.
John Berger wrote of photography that it was an act of remembering, and thus of redemption, and I have come to think that Freeman is redeeming the banal — the common place. “What is remembered has been saved from nothingness. What is forgotten has been abandoned.” In redemption, we begin again, wherever we are. A photograph, for Berger, is a moment, decisively caught, that testifies to “the always slightly surprising range of the possible.” 27
Imagine: pointing a camera earthward from a satellite that is perpetually falling past the planet at 15,000 miles per hour. 28 “It’s an impossible image to take,” Freeman told me, “an impossible perspective. And it’s very hard to explain to someone why that impossible thing is happening.” But it did, it does, every day. Nothing is more common.
Census Tract 9900, Santa Rosa County, Florida. [@everytract]
Census Tract 9704, Harlan County, Kentucky. [@everytract]
Census Tract 994.15, Orange County, California. [@everytract]
Census Tract 149, Clear Creek County, Colorado. [@everytract]
Census Tract 9677, Phillips County, Colorado. [@everytract]
Census Tract 9526, Pike County, Illinois. [@everytract]
Census Tract 4024.03, Baltimore County, Maryland. [@everytract]
Census Tract 5, Concordia Parish, Louisiana. [@everytract]
Census Tract 2435, Cook County, Illinois. [@everytract]
Census Tract 43.28, Pima County, Arizona. [@everytract]
Census Tract 503.11, Broward County, Florida. [@everytract]
Census Tract 21, Allegany County, Maryland. [@everytract]
Census Tract 218, Hawaii County, Hawaii. [@everytract]
Census Tract 32.15, San Joaquin County, California. [@everytract]
Census Tract 9506.03, Walton County, Florida. [@everytract]
Census Tract 4906.05, Baltimore County, Maryland. [@everytract]
Census Tract 9660, Seward County, Kansas. [@everytract]
Census Tract 9510, Lawrence County, Indiana. [@everytract]
Census Tract 601, San Francisco County, California. [@everytract]
Census Tract 4631, Haskell County, Kansas. [@everytract]
@everytract's path around the country, according to a tweet from Neil Freeman, February 20, 2018.
If you would like to comment on this article, or anything else on Places Journal, visit our Facebook page or send us a message on Twitter.