Postcard — September 4, 2014, 7:00 pm

Lunch at the Robot Grill

What Japan’s automats portend for American restaurant chains

This spring, at a time when American fast-food workers were marching to demand pay increases, and local governments were voting to raise the minimum wage, the Chili’s restaurant chain installed more than 45,000 tabletop touchscreen devices at 823 of its franchises nationwide. Customers at these locations can now order drinks and dessert directly through monitors, pay without the assistance of a server, play games, and read the digital edition of USA Today. The company has also installed computerized ovens at 1,200 locations. Applebee’s, meanwhile, has announced plans to follow suit with approximately 100,000 tabletop tablets by the end of 2014, while Panera Bread is replacing many registers with self-serve kiosks and adding technology that will allow customers to sit down, enter their orders and table numbers on a smartphone, and have their food delivered to them.

For consumers, such automation is convenient. For restaurant workers, it raises fears of displacement. Some analysts believe fast-food chains will respond to the push for higher wages by simply replacing servers and cooks with robots — some of which are already arriving: MIT’s Makr Shakr is capable of mixing cocktails ordered through mobile devices, while the Chinese-made NoodleBot cuts fresh noodles at a fraction of the usual cost. And a hamburger maker marketed by Momentum Machines can grind meat, cook patties, slice tomatoes, and assemble and bag approximately 360 burgers per hour. Restaurants around the world are exploring new ways to implement these new technologies. Since 2008, customers at Bagger’s, in Nuremberg, Germany, for example, have been ordering from touch screens, then waiting for the cook to deliver their food by sliding a container down a set of winding metal tracks, in a theatrical touch.

Japanese restaurants embrace a more pragmatic form of mechanization. At the thriving donburi chain Matsuya, customers order and pay for meals at self-serve machines known alternately as shokkenki or kenbaiki (both of which translate to “ticket machine”). So too at the home-style restaurant chain Yayoiken and at the countless independent curry, ramen, takoyaki, and udon restaurants that fill Japan’s bustling train stations and side streets. For these businesses and their customers, shokkenki are a benign convenience, so integrated into daily life that locals barely notice them.

Outside a shokkenki restaurant, Tokyo © Aaron Gilbreath

Outside a shokkenki restaurant, Tokyo © Aaron Gilbreath

I, on the other hand, was rapt. This past winter, I spent three weeks in Japan. Jet-lagged and red-eyed on my first night in Tokyo, I was so hungry that when a cab dropped me off on Kita-dori, I’d just rolled my luggage past my hotel and gone in search for food. I eventually found a building with a menu hanging from a door. The lights around me flashed as though I was emerging from anesthesia. Nothing made sense: not the surrounding maze of streets, nor the language on the menu. I’d been awake for a full day and traveling nearly as long. 

After studying the menu for a while, I peered into the window. Solitary men sat at a U-shaped counter, shoveling meat and rice into their mouths. Nobody talked, but the interior looked inviting, like a 1940s New York City diner in bright orange and yellow. The seeming isolation appealed to me — the less I was forced to speak, the more time I would have to acclimate and absorb my new surroundings.

Through the front window, I watched a man go through the ordering process. The system appeared simple enough: he scanned one of the menus hanging on the entrance doors, entered, approached a machine, inserted coins, and ordered by pushing a button labeled with a photo of his chosen meal. When he sat down, an employee took his ticket, set a cup of tea in front of him, and started cooking his food.

I’d heard of this system years before my visit and had thought it sounded dystopian. But I ended up enjoying my meal so much that I ate at shokkenki restaurants often. I saw them as a distinctive mix of working-class luxury, respectful indifference, and all-hours accessibility, and I appreciated the chance they afforded me to observe a cross-section of regular people: salarymen and salarywomen, blue-haired club kids, construction workers, pods of teenage girls. The restaurants seemed like a great equalizing commons, humane in a way that their mechanized underpinnings seemed to contradict.

I work in a tea shop for a living, which makes it hard for me to reconcile my perceptions of the social aspects of shokkenki restaurants with the antisocial implications of the automation taking place stateside. For one, automation creates the potential for widespread job losses among already low-income workers. And technology in general, while it has made socializing easier in many ways, already dominates the way we spend our time. Americans spend an average of five hours a day consuming digital media, versus about an hour socializing. By seeking to eliminate the interactions between hospitality workers and customers, does automation threaten to isolate us further?

The first automated restaurants in the United States predated the computer age. In 1888, two enterprising Philadelphia businessmen, Joseph Horn and Frank Hardart opened a successful lunchroom and baking company called Horn & Hardart. A decade later, a salesman contacted the pair and offered to sell them a vending machine designed by his employer, a German engineer and businessman named Max Sielaff. The machines were rising in popularity in Europe, but the salesman was struggling to sell American entrepreneurs on the idea that they would benefit urban customers who were immersed in the era’s faster pace of life. He’d heard, though, that the two men in Philadelphia were innovators, and it probably didn’t hurt that Hardart was German-born.

The salesman showed Horn and Hardart drawings of the prototype for a large, ornate, coin-operated vending machine that dispensed drinks and food prepared by staff hidden behind or in a basement. Intrigued, Hardart traveled to Berlin in 1900 and visited one of the “waiterless” or “automatic” restaurants that had begun operating in the city.1 He was impressed enough that he bought one of the machines.2 Then he had it placed on a steamship, which promptly sank. The replacement, whose cost was footed by an insurance company, arrived safely in 1902, and America’s first automat opened in Philadelphia that same year. It proved popular, and soon began to spread.

1 They later spread to Frankfurt, Leipzeig, Hannover, Brussels, Copenhagen, Stockholm and London, bearing names like Kaiser Automat, Cuisine L’Électricité, and Eleklrisch Restaurant Automalisch.
2 Historians disagree about whether Horn and Hardart ultimately bought their first machines from Sielaff or a German company called Quisisana.

Horn & Hardart’s low prices and welcoming atmosphere helped the chain thrive through the Great Depression and World War II. By 1941, the company had machines in 157 restaurants and shops across the country, and at its zenith in the mid-1950s, it was serving more than 750,000 people a day across 180 outlets, making it the world’s largest restaurant chain. The corporation grossed $71 million dollars in 1953.

Rather than seeing the vending machines as threats, American workers seemed to view H&H as a source of jobs. Wages were low and paid holidays scarce, but it was easy to get hired and hard to get fired, and benefits were abundant, from free food during the Depression to free group life insurance, a small pension plan, and the occasional opportunity to tap Joe Horn’s personal fund, from which he gave staff interest-free loans to pay overdue bills and hospital costs. Until the 1960s, institutional racism kept employees of color in back and out of sight, but immigrants were regularly hired, often despite limited English language skills and sometimes within days of arriving in the United States. The company also had a benevolent policy that allowed anyone of any means, including homeless people, to stay inside their restaurants as long as they wanted — an approach a company president summarized as, “Come on in, have a glass of water and a toothpick and stay a while.”

The trends that led to Horn & Hardart’s decline were already in motion in the postwar era, though: migration from cities to suburbs, rising labor and food costs, and competition from frozen dinners and fast-food outlets. As the company began shutting down its restaurants, some locations became McDonald’s, Burger Kings, and Arby’s. The company’s last Automat closed in Manhattan in 1991.

The word “automat” derives from the Greek “automatos,” meaning “self-acting,” but of course Horn & Hardart’s system was never truly automatic. Behind the machines was a hidden hive of stations buzzing with employees who prepped, cooked, stocked, and cleaned. In a sense, fast food inverted the automat model, applying mechanization to the back end — fryers and prepared ingredients, heat lamps and computerized grills, microwaves and heated storage trays — and emphasizing service and human interaction at the front. The recent moves toward mechanization by American chains have suggested to some that businesses want to limit the human interaction — and the labor expense — at both ends.

The new Chili’s and Applebee’s systems do reduce staff interactions, but the companies haven’t replaced people with robots yet. “In no way are we looking at this as a replacement for servers,”  an executive at Chili’s parent company, Brinker International, told Forbes. Panera’s chief executive, Ron Shaich, has assured the press that his company isn’t trying to reduce staff, either.

Inside a shokkenki restaurant, Tokyo © Aaron Gilbreath

Inside a shokkenki restaurant, Tokyo © Aaron Gilbreath

In Japan, where restaurant mechanization has been constant for decades, something in the culture or the economy has ensured that human interaction remains prominent. At Matsuya, shokkenki have freed staff from having to push register keys, make change, chit-chat, and stand idly by while customers decide what to order. But whenever I entered one, people were still on hand to cook, deliver, and clean. The machines seemed to me like a supplement to human service, a way to remove one task from the chain of production and lower costs, rather than a step toward eliminating everyone.

The corporate rhetoric, at least, is that tabletop devices and self-serve kiosks will function the same way in the United States. We already use ATMs instead of bank tellers, place takeout orders by phone, check ourselves out at some grocery stores, and check ourselves in at the airport. We require technicians and programmers to keep the machines running, as well as staff to stand nearby and tell us to place our groceries back down on the scanner before placing them in the bag. The ideal, in this telling, is that technology and automation give us more time for human interactions with our friends and family. The reality, of course, is that they often just give us more time with our other tech.

I felt in Japan as though I saw the ideal at work. That first night at Matsuya, I stood outside the door, considering my options among images of what looked like stewed beef, eggs, and sausages. I pressed a button to enter the restaurant, stepped up to the ticket machine, fed coins into a slot, and scanned the panel of buttons on the shokkenki for my item. When I found the tiny image corresponding to beef gyudon, I selected it and my coins came spilling out the bottom. I turned around, afraid my bumbling was holding up the line, to find a man in a white collared shirt, his chin dark with stubble and his eyes warm. “Sorry,” I said. “I messed up. Gomenasai.

He nodded. “Can I help?” He took the coins from the tray, placed them in the slot, and pressed the gyudon button. This time a small ticket emerged. I’d hit the “return coin” button, thinking that it would confirm my order.

I thanked the man and sat down. We ended up speaking for twenty minutes. His name was Koki, and he worked as a beverage manager at a nearby Hilton. When I told him where I was from, he smiled. “Oh, Oregon. Your first time in Japan?” It was, and despite my disorientation, I was enormously excited to be there. “You like Matsuya?” Until that moment, I hadn’t known its name.

“I do now,” I said.

He asked what else I liked to eat and suggested places for me to visit. Then we shook hands and parted ways. I didn’t see Koki again, but I think often about how he confounded my expectation. I’d believed when I arrived that automation would give me the distance I needed to acculturate, and was reminded instead that other people help us do that best.

Single Page
is an essayist and journalist who lives on the West Coast. He has written for the New York Times, The Paris Review, The Oxford American, Virginia Quarterly Review, Apology, and The Believer, and he wrote the musical appendix to The Oxford Companion to Sweets. His Twitter handle is @AaronGilbreath.

More from Aaron Gilbreath:

Postcard — April 23, 2015, 8:00 am

Driving the San Joaquin Valley

An afternoon with Starbucks customers in the armpit of California.

Postcard February 10, 2014, 2:50 pm

Nothing Is Strange

A trip to Murakami’s jazz club

Get access to 168 years of
Harper’s for only $45.99

United States Canada



August 2018

Combustion Engines

= Subscribers only.
Sign in here.
Subscribe here.

There Will Always Be Fires

= Subscribers only.
Sign in here.
Subscribe here.

The End of Eden

= Subscribers only.
Sign in here.
Subscribe here.

How to Start a Nuclear War

= Subscribers only.
Sign in here.
Subscribe here.

view Table Content


Combustion Engines·

= Subscribers only.
Sign in here.
Subscribe here.

On any given day last summer, the smoke-choked skies over Missoula, Montana, swarmed with an average of twenty-eight helicopters and eighteen fixed-wing craft, a blitz waged against Lolo Peak, Rice Ridge, and ninety-six other wildfires in the Lolo National Forest. On the ground, forty or fifty twenty-person handcrews were deployed, alongside hundreds of fire engines and bulldozers. In the battle against Rice Ridge alone, the Air Force, handcrews, loggers, dozers, parachutists, flacks, forecasters, and cooks amounted to some nine hundred people.

Rice Ridge was what is known as a mega-fire, a recently coined term for blazes that cover more than 100,000 acres. The West has always known forest fires, of course, but for much of the past century, they rarely got any bigger than 10,000 acres. No more. In 1988, a 250,000-acre anomaly, Canyon Creek, burned for months, roaring across a forty-mile stretch of Montana’s Bob Marshall Wilderness in a single night. A few decades on, that anomaly is becoming the norm. Rice Ridge, for its part, swept through 160,000 acres.

At this scale, the firefighting operation is run by an incident management team, a group of about thirty specialists drawn from a mix of state and federal agencies and trained in fields ranging from aviation to weather forecasting and accounting to public information. The management teams are ranked according to experience and ability, from type 3 (the least skilled) to type 1 (the most). The fiercest fires are assigned to type 1s. Teams take the name of their incident commander, the field general, and some of those names become recognizable, even illustrious, in the wildfire-fighting community. One such name is that of Greg Poncin, who is to fire commanders what Wyatt Earp was to federal marshals.

Smoke from the Lolo Peak fire (detail) © Laura Verhaeghe
There Will Always Be Fires·

= Subscribers only.
Sign in here.
Subscribe here.

The pinhal interior, a wooded region of hills and narrow hollows in rural central Portugal, used to be farmland. Well into the latter half of the past century, the fields were worked by peasants from the old stone villages. Portugal was poor and isolated, and the pinhal interior particularly so; when they could, the peasants left. There is electricity and running water now, but most of the people have gone. The fields have been taken over by trees. Each year the forest encroaches farther, and each year the villages grow more lonely. There are remnants of the earlier life, though, and amid the trees the holdouts of the older generations still work a few small fields. The pinhal interior cannot yet be called wilderness, then, and that, in large part, is why it burns.

Thousands of fires burn in the region each summer, almost all of them started not by lightning or some other natural spark but by the remaining Portuguese. (The great majority of the blazes are started unintentionally, though not all.) The pinhal interior—the name means “interior pine forest,” though today there is at least as much eucalyptus as pine—stretches along a sort of climate border between the semiarid Iberian interior and the wet influence of the Atlantic; vegetation grows exceptionally well there, and in the summers fire conditions are ideal. Still, most of the burns are quickly contained, and although they have grown larger in recent years, residents have learned to pay them little mind. The creeping fire that began in the dry duff and twigs of an oak grove on June 17 of last year, in the district of Pe­drógão Grande, therefore occasioned no panic.

A local woman, Dora da Silva Co­sta, drove past the blaze in the midafternoon, by which time it had entered a stand of pines. Firefighters were on hand. “There were no people in the streets,” Costa told me. “It was just another fire.” She continued on her way. It was a Saturday, and she had brought her two young sons to visit their older cousin in Vila Facaia, the village of small farms in which she’d been raised.

Firefighters near Pedrógão Grande (detail) © Pablo Blazquez Dominguez/Getty Images
The End of Eden·

= Subscribers only.
Sign in here.
Subscribe here.

On a blistering morning in July 2017, Ghazi Luaibi rose before dawn and set out in a worn black sedan from his home in Zubair, a town of concrete low-rises in southern Iraq. He drove for a while along sandy roads strewn with plastic bags. On the horizon, he could see gas flares from the oil refineries, pillars of amber flame rising into the sky. As he approached Basra, the largest city in the province, desert scrub gave way to empty apartment blocks and rows of withered palms. Though the sun had barely risen, the temperature was already nearing 100 degrees Fahrenheit. The previous year, Basra had registered one of the highest temperatures ever reliably recorded on earth: about 129 degrees, hot enough to cause birds to drop from the sky.

Ghazi, a sixty-two-year-old with stooped shoulders, an ash-gray beard, and lively brown eyes, would have preferred to stay home and wait out the heat. But he hadn’t had much of a choice. He was the president of the local council of Mandaeans, members of a gnostic religion that appeared in Mesopotamia in the early centuries ad. Today marked the beginning of their new year, and Ghazi, who was born into the Mandaean priestly class, was responsible for making sure everything went smoothly: he needed to find a tent to shield worshippers from the sun and, most importantly, a location near flowing water where they could carry out the ceremony.

Mandaean holidays are celebrated with a mass baptism, a ritual that is deeply rooted in their scripture and theology. Mandaeans follow the teachings of Yahia Yuhana, known to Christians as John the Baptist. Water is central to their religion. They believe that all life originates in the World of Light, a spiritual realm that is the starting point for a great river known as Yardana, or Jordan. Outside the World of Light lie the lifeless, stagnant waters of the World of Darkness. According to one version of the Mandaean creation myth, a demiurge named Ptahil set out to shape a new world from the World of Darkness, which became the material world we inhabit today. Once the world was complete, Ptahil sculpted Adam, the first man, from the same dark waters as the earth, but his soul came from the World of Light. In Mandaean scripture, rivers are manifestations of the World of Light, coursing from the heavenly Jordan to the earth to purify it. To be baptized is to be immersed in this divine realm.

Basra General Hospital (detail) July 2017 © Alex Potter
How to Start a Nuclear War·

= Subscribers only.
Sign in here.
Subscribe here.

Serving as a US Air Force launch control officer for intercontinental missiles in the early Seventies, First Lieutenant Bruce Blair figured out how to start a nuclear war and kill a few hundred million people. His unit, stationed in the vast missile fields at Malmstrom Air Force Base, in Montana, oversaw one of four squadrons of Minuteman II ­ICBMs, each missile topped by a W56 thermonuclear warhead with an explosive force of 1.2 megatons—eighty times that of the bomb that destroyed Hiroshima. In theory, the missiles could be fired only by order of the president of the United States, and required mutual cooperation by the two men on duty in each of the launch control centers, of which there were five for each squadron.

In fact, as Blair recounted to me recently, the system could be bypassed with remarkable ease. Safeguards made it difficult, though not impossible, for a two-man crew (of either captains or lieutenants, some straight out of college) in a single launch control center to fire a missile. But, said Blair, “it took only a small conspiracy”—of two people in two separate control centers—to launch the entire squadron of fifty missiles, “sixty megatons targeted at the Soviet Union, China, and North Korea.” (The scheme would first necessitate the “disabling” of the conspirators’ silo crewmates, unless, of course, they, too, were complicit in the operation.) Working in conjunction, the plotters could “jury-rig the system” to send a “vote” by turning keys in their separate launch centers. The three other launch centers might see what was happening, but they would not be able to override the two votes, and the missiles would begin their firing sequence. Even more alarmingly, Blair discovered that if one of the plotters was posted at the particular launch control center in overall command of the squadron, they could together format and transmit a “valid and authentic launch order” for general nuclear war that would immediately launch the entire US strategic nuclear missile force, including a thousand Minuteman and fifty-four Titan missiles, without the possibility of recall. As he put it, “that would get everyone’s attention, for sure.” A more pacifically inclined conspiracy, on the other hand, could effectively disarm the strategic force by formatting and transmitting messages invalidating the presidential launch codes.

When he quit the Air Force in 1974, Blair was haunted by the power that had been within his grasp, andhe resolved to do something about it. But when he started lobbying his former superiors, he was met with indifference and even active hostility. “I got in a fair scrap with the Air Force over it,” he recalled. As Blair well knew, there was supposed to be a system already in place to prevent that type of unilateral launch. The civilian leadership in the Pentagon took comfort in this, not knowing that the Strategic Air Command, which then controlled the Air Force’s nuclear weapons, had quietly neutralized it.

This reluctance to implement an obviously desirable precaution might seem extraordinary, but it is explicable in light of the dominant theme in the military’s nuclear weapons culture: the strategy known as “launch under attack.” Theoretically, the president has the option of waiting through an attack before deciding how to respond. But in practice, the system of command and control has been organized so as to leave a president facing reports of incoming missiles with little option but to launch. In the words of Lee Butler, who commanded all US nuclear forces at the end of the Cold War, the system the military designed was “structured to drive the president invariably toward a decision to launch under attack” if he or she believes there is “incontrovertible proof that warheads actually are on the way.” Ensuring that all missiles and bombers would be en route before any enemy missiles actually landed meant that most of the targets in the strategic nuclear war plan would be destroyed—thereby justifying the purchase and deployment of the massive force required to execute such a strike.

Among students of nuclear command and control, this practice of precluding all options but the desired one is known as “jamming” the president. Blair’s irksome protests threatened to slow this process. When his pleas drew rejection from inside the system, he turned to Congress. Eventually the Air Force agreed to begin using “unlock codes”—codes transmitted at the time of the launch order by higher authority without which the crews could not fire—on the weapons in 1977. (Even then, the Navy held off safeguarding its submarine-launched nuclear missiles in this way for another twenty years.)

Following this small victory, Blair continued to probe the baroque architecture of nuclear command and control, and its extreme vulnerability to lethal mishap. In the early Eighties, while working with a top-secret clearance for the Office of Technology Assessment, he prepared a detailed report on such shortcomings. The Pentagon promptly classified it as SIOP-ESI—a level higher than top secret. (SIOP stands for Single Integrated Operational Plan, the US plan for conducting a nuclear war. ESI stands for Extremely Sensitive Information.) Hidden away in the Pentagon, the report was withheld from both relevant senior civilian officials and the very congressional committees that had commissioned it in the first place.

From positions in Washington’s national security think tanks, including the Brookings Institution, Blair used his expertise and scholarly approach to gain access to knowledgeable insiders at the highest ranks, even in Moscow. On visits to the Russian capital during the halcyon years between the Cold War’s end and the renewal of tensions in the twenty-first century, he learned that the Soviet Union had actually developed a “dead hand” in ultimate control of their strategic nuclear arsenal. If sensors detected signs of an enemy nuclear attack, the USSR’s entire missile force would immediately launch with a minimum of human intervention—in effect, the doomsday weapon that ends the world in Dr. Strangelove.

Needless to say, this was a tightly held arrangement, known only to a select few in Moscow. Similarly chilling secrets, Blair continued to learn, lurked in the bowels of the US system, often unknown to the civilian leadership that supposedly directed it. In 1998, for example, on a visit to the headquarters of Strategic Command (­STRATCOM), the force controlling all US strategic nuclear weapons, at Offutt Air Force Base, near Omaha, Nebraska, he discovered that the ­­­STRATCOM targeting staff had unilaterally chosen to interpret a presidential order on nuclear targeting in such a way as to reinsert China into the ­SIOP, from which it had been removed in 1982, thereby provisionally consigning a billion Chinese to nuclear immolation. Shortly thereafter, he informed a senior White House official, whose reaction Blair recalled as “surprised” and “befuddled.”

In 2006, Blair founded Global Zero, an organization dedicated to ridding the world of nuclear weapons, with an immediate goal of ending the policy of launch under attack. By that time, the Cold War that had generated the ­SIOP and all those nuclear weapons had long since come to an end. As a result, part of the nuclear war machine had been dismantled—warhead numbers were reduced, bombers taken off alert, weapons withdrawn from Europe. But at its heart, the system continued unchanged, officially ever alert and smooth running, poised to dispatch hundreds of precisely targeted weapons, but only on receipt of an order from the commander in chief.

Bombhead, by Bruce Conner (detail) © Conner Family Trust, San Francisco, and ARS, New York City. Courtesy Kohn Gallery, Los Angeles

Minimum cost of a “pleasure palace” being built for Vladimir Putin:


Israeli researchers claimed to have identified a ruthlessness gene.

Trump and Putin puzzle out cybersecurity in Helsinki, John Kelly didn't like his breakfast in Brussels, and a family of woodchucks ate the wiring in Paul Ryan's car

Subscribe to the Weekly Review newsletter. Don’t worry, we won’t sell your email address!


Happiness Is a Worn Gun


Illustration by Stan Fellows

Illustration by Stan Fellows

“Nowadays, most states let just about anybody who wants a concealed-handgun permit have one; in seventeen states, you don’t even have to be a resident. Nobody knows exactly how many Americans carry guns, because not all states release their numbers, and even if they did, not all permit holders carry all the time. But it’s safe to assume that as many as 6 million Americans are walking around with firearms under their clothes.”

Subscribe Today