Takeout Delivery Drivers: Caught in the System

“Delivering food is like racing against death, pitting yourself against traffic cops and making friends with red lights.” That is how one deliveryman describes his profession. The numbers bear this out. Data from the traffic department of the city of Shanghai show that in the first half of 2017, a deliveryman was injured or killed in Shanghai on average every 2.5 days. That same year, over the course of three months in Shenzhen, twelve deliverymen were injured or killed. In 2018, the Chengdu traffic police reported that over a seven-month period, nearly 10,000 offenses by deliverymen were investigated, 196 accidents were reported, and 155 people were injured or killed; on average, a deliveryman was injured or killed because of a traffic infraction every single day.
What has made food delivery the most dangerous profession? To find out the answer, a team from People undertook six months of investigation. Through interviews with dozens of deliverymen from across the country, with people involved at every link in the business chain, and with sociologists, we gradually uncovered an enormous “black box” system.
The article is long, but after reading it, you will understand why, although we often describe traffic mishap as ‘accidents,’ for these food delivery drivers, there is nothing accidental about it.

“Received.”

Two more minutes had disappeared from the system.

ELEME driver Zhu Dahe remembers that day in October 2019 clearly. When he saw the time allotted by the system for the order, he gripped his motorbike handlebars with sweaty hands. “2 kilometers, delivered within 20 minutes.” He’d been delivering food in Beijing for two years, and before that day, the shortest time he’d been given for that distance had been 32 minutes. But from that day on, two minutes had disappeared.

Around the same time, Meituan drivers experienced similar “disappearing minutes.” A Meituan driver in Chongqing who specialized in long-distance orders noticed that the time on an order was 35 minutes, while similar distances used to be allotted 50 minutes. His roommate was in the same profession, and for him, for anything within three kilometers, the delivery time had been reduced to 30 minutes.

This wasn’t the first time that minutes had disappeared from the system.

Jin Zhuangzhuang was a Meituan dispatch station supervisor for three years, and he remembers that from 2016 to 2019, he received three notices of “increased speed” from the Meituan platform. In 2016, the time limit for three kilometers was one hour; in 2017, that became 45 minutes, and in 2018, the time was reduced by a further seven minutes to 38 minutes. Statistics show that in 2019, across the entire food delivery system in China, the average delivery time on food orders had been reduced by ten minutes over the course of three years.

The system was able to “gobble away” the minutes continually, and for the founders of the system, this was praiseworthy, an accomplishment of the deep learning of which smart machines are capable. At Meituan, this “real-time smart distribution and delivery system” was known as the “big brain.” The ELEME platform named it “the ark.” In November 2016, Meituan founder Wang Xing said in an interview: “Our slogan is, ‘With Meituan delivery, everything’s quick.’ We average a delivery time of under 28 minutes.” He added, “This is the result of excellent technology.”

But in practice, for the deliverymen dealing with the “technological advancement,” it was perhaps a “crazy” and even “life threatening” development. 

The way the system is set up, delivery time is the most important target, and exceeding the time limit on deliveries is not permitted. If a late delivery happens, it leads to a bad review, a reduction in income, and even being fired. On a Baidu forum for takeout deliverymen, one driver write: “Delivering food is like racing against death, pitting yourself against traffic cops and making friends with red lights.”

One driver in Jiangsu renamed his social media handle “superspeed dog” as a constant reminder to himself. A driver who lives in the Songjiang district of Shanghai said that for nearly every order, he ends up driving the wrong way down a street. He figures he can save five minutes on each order that way. Another Shanghai ELEME driver did a rough calculation that showed that if he didn’t break any traffic rules, he would be able to travel only half as far in a day.

“Drivers as individuals have no way of protesting the time the system gives them. We can only rely on speeding to make up the time.” One Meituan driver told People that the “craziest order” he had ever experienced was for a one-kilometer delivery within 20 minutes. Although the distance wasn’t far, he had to go to the restaurant, pick up the food, and deliver it all within that short period of time. That day, he rode so quickly that “my butt bounced up off the seat a couple of times.”

Speeding, running red lights, driving against traffic—according to the Chinese Academy of Social Sciences Research Associate Sun Ping, these deliverymen battle traffic rules because of “counter algorithms.” It is an unavoidable work practice resulting from deliverymen laboring under the control and regulation of the system’s algorithms; and their “counter algorithms” have led directly to a rapid rise in traffic accidents.

Sun Ping began studying the connections between the takeout delivery system algorithms and the work of delivery drivers in 2017. In her conversations with People, she mentioned “continually shortening delivery times,” and “continually increasing numbers of traffic accidents,” expressing that the former was “definitely” the primary reason for the later.

The numbers bear this out. Data from the traffic department of the city of Shanghai show that in the first half of 2017, a deliveryman was injured or killed in Shanghai on average every 2.5 days. That same year, over the course of three months in Shenzhen, twelve deliverymen were injured or killed. In 2018, the Chengdu traffic police reported that over a seven-month period, nearly 10,000 offenses by deliverymen were investigated, 196 accidents were reported, and 155 people were injured or killed; on average, a deliveryman was injured or killed because of a traffic infraction every single day. In September 2018, the Guangzhou traffic police investigated nearly 2000 traffic infractions by deliverymen, with Meituan drivers accounting for half of that, and ELEME drivers coming in second.

Takeout delivery driver has become the most dangerous profession. This sentence has been one of the most popular searches on Weibo more than once.

In an article such as this, specific cases will touch readers more than statistics.

In February 2018, an ELEME driver was speeding in a non-motorized vehicle lane and hit Li Mouqiu, a leading authority in the Shanghai emergency medical community and founder of the emergency medicine departments at Ruijin Hospital and Huashan Hospital. Li Mouqiu was rushed to the hospital, but sadly died a month later. In May 2019, a driver in Jiangxi was speeding while delivering food and hit a pedestrian, leaving him in a vegetative state. One month later, a Chengdu driver sped through a red light and hit Bao Shijie, severing his right leg. That same month, a driver in Xuchang in Henan was driving against traffic, hit someone, flew up into the air and did two full circles before landing, breaking most of the bones in his body.

Zhu Dahe—who was so “shocked” by the shorter delivery time that his hands started sweating on his handlebars—has been in two accidents. He once turned to avoid hitting a bicycle and his speeding motorbike slid into the bicycle lane. The spicy soup he was delivering went flying. As he went down, what hit his brain even before the pain was the thought: “Damn, now I’ll never get there on time.”

In order to avoid being late and getting a bad review, he called the customer and begged him to cancel the order, saying that he would pay for the soup himself. “It was so expensive, more than 80 yuan,” he said, “but it tasted pretty good.” He still thinks about that incident. He had just started the job and had little experience. It would have been better to have given the customer the money for the soup and asked him to order it again, and that way, “At least he would’ve paid the delivery fee. Six and a half yuan. I remember it exactly.”

“You see people slip on their bikes all the time. As long as the food doesn’t spill, it’s not that big of a deal.” Zhu Dahe said that when he was out on a delivery, he frequently saw his colleagues get into accidents. “Mostly you don’t stop,” because “the food won’t wait for it.”

Meituan driver Wei Lai can confirm the truth of that.

Midday in spring this year, Wei Lai and another driver wearing the same type of uniform were both waiting for a red light at an intersection. The other driver started a few seconds too early, rushing though the intersection, and was hit by a speeding car. “The guy and his bike were knocked into the air. He died there at the scene.” Wei Lai said that even when he saw the mangled body of his colleague lying in the middle of the road, he didn’t stop. “I was about to go over the time limit on the order I had.” At the same time, a new order had come in, and a familiar female voice sounded in his ear. “Another order. From X location to X location. Please respond received after the tone.”

 

Rain 

The way the system is set up, after a driver responds ‘received,’ the clock starts ticking.

In 2019, at the Arch Summit annual gathering for innovators, Wang Shengyao, senior algorithm expert for the Meituan delivery technology team, explained the basic workings of the smart system.

From the instant a customer puts in an order, the system begins to choose a driver, based on the drivers’ availability, position, and direction. The orders are usually sent out in groups of three or five at a time. One order involves two separate tasks: picking the food up and delivering it. If a driver has undertaken five orders, which involves ten tasks, the system will use a map of 110,000 possible routes to “solve the problem of how to get 10,000 orders to 10,000 people in seconds,” finding the optimal route.

In reality, however, a bit of rain is enough to destroy any sense of “optimal.”

Drivers feel ambivalent about the rain. On the one hand, they like the rain because it brings in more orders. On the other hand, if the rain is too heavy, the system easily “floods” with orders, and the drivers themselves easily “get into trouble.”

Gengzi, a Meituan driver in Hunan, once encountered a terrifying nighttime downpour. It had been raining heavily all day, and the orders had come in like crazy until the system had been flooded. At their distribution center, all of the drivers had been given ten orders at a time, and their back baskets and bike handles were covered with food orders. Gengzi remembers that his feet could barely rest on the edges of the footrest as he drove, keeping an eye on the food containers stacked between his calves so they wouldn’t get crushed.

The road was very slick and his bike slipped several times, but he would quickly get back up and continue on. He didn’t finish his deliveries until 2 AM. A few days later, he received his pay for the month, and it was much lower than what he usually made. The reason was simple: the day it had rained so hard, many of his orders had been delivered late, and so his pay had been reduced.

It wasn’t only Gengzi’s pay that had been reduced; the dispatch station supervisor also saw his own pay go down.

“I eat depending on the numbers,” Meituan dispatch station supervisor Jin Zhuangzhuang said. For a dispatch station, the most important numbers include: how many total orders, how many late deliveries, how many bad reviews, and how many complaints. Of these, the late deliveries are the most important, because they also are the reason for the bad reviews and complaints.

Generally, a driver’s rate of late deliveries cannot exceed 3%. If he misses that target, the dispatch station’s rating will go down, and the cut of the profit given to the whole station will decrease, including to the supervisor, human resources department, quality control department, and everyone else working there. Even the income of the supply channel and district managers will be affected.

At the end of every year, each station is assessed by the Meituan or ELEME platform. Those stations coming in the bottom 10% for their region face the danger of being eliminated.

With this system of evaluation, “late deliveries” not only bring income losses to the drivers, they also bring emotional pressure.

“You become a thorn in the side of everyone on your team,” Sun Ping said. “Late deliveries are very serious business. Not only is your pay deducted, and the deductions are significant, there’s also the question of your team’s honor. If you drag your team down, the supervisor will give you a talking to. After that, your regional manager will come find you. After that, it’s the area manager, and so on and so on. Everyone will be upset.”

This places tremendous pressure on the drivers. Zhu Dahe, who lost the spicy soup to a fall, told People that he fell into a depression after he’d been a driver for a few months.

He came from a small town and didn’t know the Beijing roads. He was even less familiar with the heavy traffic. Fearful, he followed the traffic rules, and lost money on late deliveries every single day. He felt helpless. “Can’t a deliveryman make more than 10,000 yuan? Doesn’t everyone else manage to make their deliveries? Why can’t I do it too?” He said, “I guess I just wasn’t made to be a deliveryman.”

Slowly, he became used to driving a motorbike and learned the roads. He went from being a newbie to being an expert racer against time. His helpless feeling gradually disappeared. “Compared to making a late delivery, who cares about riding against traffic.” He said that when he and his colleagues drove against traffic, he felt it was “smooth sailing.”

Today, in ordinary circumstances, Zhu Dahe rarely makes late deliveries. But on days when the weather is especially bad, he still can’t escape the curse. The system goes out of control and drags him in too. Faced with too many orders, with the delivery timing out of whack, he still faces penalties for late deliveries. Taking the day off is also out of the question.

In August 2019, Typhoon Lekima hit Shanghai, and an ELEME driver died from an electric shock while on a delivery. A snapshot taken of a discussion on a private wechat group for a dispatch station was posted publicly online. In it, the station supervisor @everyone, saying: “For the next three days no one will be given time off….if anyone isn’t at work over the next three days they will be penalized with double the usual fee. Reply to this message when received.” Under the supervisor’s order is a row of 1s, indicating the drivers have received the message.   

The photo of this exchange was hotly debated online. Some commenters said that during a typhoon, why could the Hema grocery store, Kentucky Fried Chicken, and MacDonald’s all close, but the takeout delivery service platforms couldn’t?

Meituan supervisor Jin Zhuangzhuang could only express helplessness at that question. Each time there was a heavy rain, the drivers would all come to him to ask for time off because of a busted tire, a bad fall, problems at home, or some other excuse. But as the orders poured in, in order to keep up the station’s numbers, he could only enforce the rules. “Unless you’re sick, dying, or about to give birth, you can’t take off on days with bad weather. If you do, you’ll be fined.”

Rainy days were the most tiring for Jin Zhuangzhuang as well. He had to sit in the station in front of his computer, inspecting the position of each of his drivers, how many orders they have, and when they make their deliveries. For the station under his command, Meituan stipulates that each driver can accept a maximum of twelve orders. If they exceed that number, the system will step sending the orders in. But on days with bad weather or on important holidays, there’s no way to meet the high demand. At such times, the system easily collapses: different drivers will receive the same orders, while others receive no orders at all. Some drivers receive orders that take them in opposite directions; nearby orders will be given more time than long-distance orders….

When that happens, Jin Zhuangzhuan had to play a different role: “manual dispatcher.” In that role, he could enter the system and take an order from Driver A and give it to Driver B, attempting to achieve a balance. Although the system sets a maximum at twelve orders, the manual dispatcher has no limit, and as long as there is a person at the controls, drivers may receive a “truly terrifying number” of orders. One driver dealt with 26 orders at once. A dispatch station with 30-odd drivers once took care of 1000 orders within 3 hours. Another driver in a town of half a million people was assigned sixteen orders during rush hour.

One station supervisor for ELEME told People that this kind of manual interference was not intended to save the drivers, but rather to “push each driver to his limits.”

If the drivers reach their limits and it still isn’t enough, Jin Zhuangzhuang would go out and deliver orders himself. The most he’s taken on at once was fifteen orders. During a flood of orders, “first I’d let the drivers stew in it for a while. When they couldn’t handle it anymore, I’d have to put in a request to Meituan to reduce the geographical range of our orders. After 2018, our station wasn’t allowed to make that request and the orders kept coming in. We all had to go out on deliveries.” Jin said that during a flood of orders, for the last few deliveries his whole body would go numb and he’d have to drive by instinct. “I wouldn’t have any human responses left.”

Last year, because of an illness in the family, Jin Zhuangzhuang left the profession. He said he would never go back. Recently a friend decided that he wanted to take on a dispatch station, but Jin convinced him not to do it. “You can’t imagine how intense the constant time pressure is, along with the constant pressure to meet the numbers.” This past summer, southern China experienced heavy rains, and although Jin Zhuangzhuang was happy to be out of the profession, he also worried about how many stations would be flooded with orders, and how many drivers would risk their lives to meet the numbers.

 

Navigation

In her research, Sun Ping has spent time with nearly a hundred takeout drivers over the past four years. Many of them complained about the road directions the system gave them.

In order to allow drivers to concentrate on delivering the food, the smart system tries its best to substitute in for a human brain: it plans out the delivery route and offers navigation. The drivers need not use their own brains; they take on the system’s directions, along with the risks of “taking a wrong turn.”

Sometimes, the navigation system will show a straight line. One driver complained to Sun Ping: “It [the algorithm] calculates the time based on a direct path as the crow flies. But there are no straight paths when we deliver. We always end up taking detours, or wait for red lights….Yesterday, I delivered an order that was supposed to be five kilometers. In the end, I drove seven kilometers. The systems thinks we’re like helicopters, but we’re not.”

Sometimes, the navigation will include directions to drive the wrong way down a one-way lane.

In October 2019, a driver in Guizhou called Xiaodao posted on Zhihu about the Meituan navigation system sending drivers the wrong way down streets. In a conversation with People, he said that he had just spent six months as a driver and had received directions to go the wrong way many times. One time when he was delivering food to a hospital, he should have found somewhere to take a U-turn to be in the proper flow of traffic, but the Meituan navigation told him to simply cross the road and drive against traffic. According to the map it gave him, he’d have to drive against traffic for nearly two kilometers.

“It gets even worse,” Xiaodao said. “In some places, it isn’t easy to drive against traffic. If there’s a pedestrian bridge, the navigator will tell you to cross it, whether or not it allows motorized vehicles. Sometimes it will even try to send you directly over the top of a dividing wall.”

In Beijing, live video streamer Cao Dao has seen the same thing. Wanting the professional experience, she drove for Meituan for under a week. What amazed her was that when she received an order, the system’s navigation would default to a walking route. The walking route does not account for one-way streets, and the time given for the order accorded with the shortest possible route, which included many sections that involved driving the wrong way.

As Xiaodao sees it, whether it’s straight lines or wrong-way directions, the system’s goal is achieved: the system relies on the navigational calculations to determine the distance, time, and delivery fee. If the route is short, the time will also be short, which attracts users to the site and reduces the delivery cost.  

At the end of 2017, Meituan’s technology team brought up the issue of “cost” in an article about optimizing their smart delivery system. The article pointed out that an optimized algorithm helped the platform reduce their transportation losses by 19%. Four drivers could now deliver what in the past five drivers were needed to deliver. “Cost” appeared again in the article’s conclusion: “Efficiency, experience, and costs will become the central targets for our platform.”

In reality, it was Meituan itself that benefitted greatly.

According to the numbers put out by Meituan, in the third quarter of 2019, Meituan takeout orders reached 2.5 billion. Each order brought in 0.04 yuan more than in 2018. This helped Meituan rake in a profit of 400 million yuan in the third quarter of 2019.

But behind this huge profit for the platform, drivers saw their income reduced. Xiaodao told us that each time the system navigation told him to drive against traffic, he faced an impossible choice. Either he refused to go the wrong way and risked a late delivery by taking the longer way around, or he followed the route at the expense of his own safety. No matter what he decided, “my pay always ended up being reduced.”

“Every driver is always having to balance his own safety against his income.” As an “outsider” who was temporarily on the inside, Cao Dao is able to point to the problem: “All of the delivery platforms are trying to turn the biggest profit. In the end, they shift the danger onto the drivers, who have no way of negotiating their own prices.”

In their conversations with People, many drivers expressed the same thing: “They don’t worry about losing drivers. If you don’t do it, someone else will.”

Before becoming a Meituan driver, Afei was a deliveryman for Kentucky Fried Chicken. “Each month, a driver would deliver at most 600 or 700 orders. The restaurants have restrictions and the brand has to give the delivery company 12 or 13 yuan per order, so a deliveryman’s fee is set at 9 yuan and doesn’t change.” He describes that work as being the most “normal,” but his income wasn’t very high. “Each month I could only bring in 5000 yuan at the most.” He decided to leave Kentucky Fried Chicken and become a delivery driver for an online platform.

Meituan and ELEME divide their drivers into two categories: specialist driver and crowdsourced driver.

Specialist drivers are fulltime drivers subordinate to a dispatch station supervisor. They have a base salary and set work hours, receive their orders through the system, and are assessed according to their customer reviews and rate of on-time deliveries. Crowdsourced drivers are part-time drivers, and their access to orders is very limited. One driver, one vehicle, and one app: after they register, they can immediately begin making deliveries. They have no base salary and can freely take on orders or refuse orders given to them by the system. But if they refuse too many orders, the system won’t allow them to take orders on. Crowdsourced drivers are not affected by bad reviews or complaints, but late deliveries mean harsh penalties. One second over the time limit and their pay is reduced by half. Whether a specialist or a crowdsourced driver, not one of the drivers is an actual employee of the delivery platforms.

Afei decided to become a crowdsourced driver for Meituan. Beginning in 2017, he worked nine hours a day, specializing in long-distance deliveries and earning around 10,000 yuan a month. In his best month, he brought in 15,000 yuan. A low threshold to get into the profession and a high income are believed to be important reasons why the platforms “don’t worry about losing drivers.”

But to social scientists, “a driver’s income exceeding 10,000 yuan” was only “a peculiarity” of the early stages of the platforms’ development. After conducting extended research in Wuhan on the labor practices of package deliverymen and takeout deliverymen, Professor Zheng Guanghuai’s research team at the Institute for the Social Sciences at Central China Normal University discovered that after delivery platforms stopped subsidizing income, for more and more drivers, an “income exceeding 10,000 yuan” had become a distant dream.  

In their report, the research team demonstrates that only 2.15% of delivery drivers make over 10,000 yuan per month, while 53.18% of interviewees said that their current income could not meet the requirements of their household expenses.

Afei told People that after working as a deliveryman in Beijing, he went to Chongqing for personal reasons and his income fell. During the pandemic, more people began working as deliverymen, and for a time, he had trouble getting orders. His income fell below 7000 yuan.

According to the Meituan Research Center’s “2019-2020 Job Report for Meituan Drivers During the Pandemic,” after the pandemic hit, the Meituan platform saw their number of registered drivers increase to 336,000 people. The largest group to join their ranks were factory workers, followed by salespeople.

To the question, “On what days do you earn the most?” Afei responds, “Only when it’s really hot or really cold.” Because during those times, “most people don’t want to go outdoors.”

 

Elevators

According to public statements made by the takeout platforms, when the system estimates a delivery time, the amount of time spent waiting for elevators is an important factor to be calculated in.

In an interview with 36Kr, the head of Meituan’s delivery algorithm team He Renqing also brought up elevators: “Meituan’s delivery system pays a lot of attention to the time our drivers spend going up and down elevators. We’ve even researched the speed with which drivers go down to the basement levels and to the top floors of buildings.”

The reality, however, is much more complicated than anything a smart system can calculate. “It’s so painful to wait for an elevator. It really hurts!” says Afei, whose income has never reached 10,000 yuan.

Many drivers report that hospital elevators are the worst. In the four years he spent as a driver, the worst Afei ever encountered was the elevator at Peking University’s Number Three Hospital. It was during the midday lunch rush, and he had seven or eight orders when he arrived at the Number Three Hospital surgery building. “It was horrible,” he said. “I remember clearly. There was a crowd of people in front of the elevators—deliverymen, patients, doctors, visitors—all crammed in together. I waited in line for a long time until I finally pushed on. Everyone was pressed up against each other, so close that I could barely breathe.” That day, after he was finally able to deliver that order, he was late on all six remaining orders.   

After he went to Chongqing, elevators remained the bane of his existence.

There’s the Hongding International building, which is “a nightmare, with 48 levels, all tiny offices packed in tightly together. One level could have 30 or 40 offices. You can image how bad it is.” The building has seven or eight elevators, but at the height of the lunch rush, the crowd waiting for an elevator “is like the crowd waiting to get into a popular scenic attraction.” It can take at least half an hour to make it to the front of the line.

Then there’s the Chongqing World Financial Center. At 74 stories high, the whole building has only one service elevator available to deliverymen. “First, during a meal rush, there just isn’t enough elevator capacity. Second, it’s probably an issue of the shape of the building itself.” Afei puts it this way: “We have to wait right outside the elevator, so when a crowd of people arrives in the elevator, they come pouring out toward us, while we’re pushing to try to get onto the elevator. You have to wait for 20 or 30 or 40 minutes for the elevator to go up, and then you have to wait just as long trying to come back down. You’re only given a certain amount of time to make the delivery, how can you not be late all the time?”

Many office buildings do not allow deliverymen to take the main elevators—deliverymen from Beijing, Shanghai, Shenzhen, Chongqing, Hunan, and elsewhere told People that this is extremely common.

On July 11, 2020, Cao Dao posted a video of her experiences working as a driver. Because of the video, “the Beijing SKP Mall does not allow deliverymen to enter” became one of the most searched phrases, leading to heated online discussion about discrimination on the basis of one’s profession. To Cao Dao, in the systematic discrimination against the delivery business, SKP was only the tip of the iceberg. In the video, which was less than ten minutes long, few people paid attention to the section at the elevator.

“I’ll never forget it,” Cao Dao said. “I was going to get the food. But the building had so many shops, and most of them do takeout. There are other elevators in the building, but the security guards wouldn’t let us use them. They only open up one elevator to deliverymen.” She was a newbie, and barely found the right elevator after spending considerable time searching. Then she waited in line with dozens of other drivers. “Everyone made two lines, with a hole in the middle to let the deliverymen getting off of the elevator through.” That day, she spent more than ten minutes just waiting for the elevator.

Aside from office buildings, some expensive residences are “elevator landmines” to drivers. There, the elevator is activated by a card swipe, but most customers do not want to come down to let deliverymen in. “They’ll make you get into the elevator and then press the elevator button on their own floor, but the elevators don’t necessarily go up.” Afei said that when encountering that kind of customer, many professional drivers will run up twenty-some flights of stairs just to avoid a bad review. But when it’s Afei making a delivery as a crowdsourced driver for Meituan, he isn’t as afraid of the reviews; his preferred solution is this: “If the guy lives on the 14th floor and he wants me to take the stairs, I’ll refuse, but tell him that I’ll go up seven floors if he comes down seven floors. That seems fair to me.”

Waiting for an elevator, Afei has seen many a driver lose their temper, or start to cry or fight with the others. It’s happened many times, because “it’s like you’re just a step away, and all you have to do is cram yourself onto the elevator and you’ll be there.” But in reality, the drivers can only stand there by the elevator doors, “and you can’t do anything about it. You’ve just got to wait.”

In order not to be late, some drivers will prematurely tell the app that they’ve made the delivery while they’re still waiting for the elevator. This is not allowed by the system. “If a customer reports you for prematurely saying you’ve made the delivery, they’ll deduct 500 yuan from your pay,” one Meituan driver from Gansu told People.

“In this instance, ELEME is a bit more humane,” one Guizhou driver told us. According to him, the ELEME system has an “about to deliver” button. “When a driver reaches a building and is waiting at the elevator with only a few minutes to spare, you can press the “about to deliver” button, and it will record the time you arrived at the building. When you leave the building again, you press the ‘delivered’ button.”

Zhengzhou driver Zhang Hu has driven for ELEME and as a crowdsourced driver for Meituan, and comparing his experiences a the two, he feels that Meituan is more ruthless. “Meituan drivers are just a bunch of delivery machines. ELEME doesn’t have that big of a market share, so it seems a little more reasonable.”

Real data supports his impressions.

According to Trustdata, a platform that monitors mobile internet data, research shows that in the first half of 2019, Meituan occupied 64.4% of the national takeout market in China. In terms of actual orders, Meituan drivers on average made 20 more deliveries each day than did ELEME drivers.

“Even if a driver is going as fast as he possibly can, the platform never thinks it’s fast enough.” Zhang Hu can’t help but complain about Meituan, but in the end, he chose to leave ELEME and join Meituan. In Zhengzhou, Meituan provided him with an almost unimaginable number of orders.

That’s the reason Afei finally chose Meituan as well. Although his income went down during the pandemic, he kept his mood up. During that time, many neighborhoods were closed off and office buildings wouldn’t let him in, but at least he didn’t have to face the elevators. As the pandemic was brought under control, more and more neighborhoods and buildings opened up, and once again he had to deal with the elevators.

As a new stage in Afei’s war with the elevators began, Cao Dao was finishing editing her video about being a driver, and she added footage of her waiting for an elevator. In mid-July, when she recalled that scene for People, she said of herself waiting at the elevator, “I was just like a worker ant.”

 

Guarded Gates

In 2019, Li Lei from Zhengzhou jumped from ELEME to Meituan. He went from being a dispatch station supervisor to working in business development. His main responsibility was to find more commercial partners for their dispatch stations. In order to find popular merchants willing to collaborate with the dispatch stations under his purview, he often took to the streets to establish relationships with local merchants. During the meal rush on weekends, he’d often bring a chair and put it by the entrance of a business, not in order to find a new partner, but rather to encourage the restaurants to put together their orders more quickly.

Slow food preparation is even more painful for takeout drivers than elevators.

“Smart” systems are continually using their “intelligence” to optimize, reducing delivery times again and again. But slow food prep has remained a big problem. Meituan senior algorithm specialist Wang Shengyao once said publicly that although they could analyze deliveries that had been made, it was very hard to calculate the time it would take a restaurant to fulfill an order. With that number uncertain, there will always be an unavoidable random variable in the system.   

But the only one shouldering the burden of this random variable while faced with a set “delivery time” is the driver.

According to drivers, there are many reasons a restaurant may be slow. Some popular restaurants get so busy during their peak times that they can barely serve the people eating in the restaurant, but still they are loathe to refuse takeout orders. Some restaurants are small, and the proprietor isn’t very concerned about timing. Sometimes a driver will arrive at a restaurant only to find the boss just heading in the door with the necessary ingredients. Some places, especially noodle restaurants, wait until the driver actually arrives to begin to cook the food, in order to make it taste better when it is delivered.

“That happens especially with three difficult dishes: roasted fish, stew, and barbecue,” one driver told People. “Last time I went to pick up an order of stew, I got to the restaurant before they’d even started it. I waited there for a whole forty minutes while they cooked it.” Another driver once started yelling: “’Hey, man!’ I was panicking and I really wanted to say, ‘Come on, man, just get on with the cooking!” But the chef paid him no attention. “They’re not worried at all. In the end, when everyone gets paid, it’s just my money that gets cut if the order is late.”

“There’s just nothing we can do about slow food prep,” former Meituan dispatch station supervisor Jin Zhuangzhuang said. In the system’s review section, restaurants can leave negative reviews for or report drivers, but drivers have no way of leaving a review for a restaurant. Occasionally, a driver will be blamed for the restaurant’s mistake: “too spicy,” “not salty enough,” “forgot to put in vinegar”.... These complaints about the food will often appear among the drivers’ reviews. Many drivers have appealed to have these reviews removed, but not a single appeal has succeeded.

Drivers can only rely on their own methods to deal with this problem. Jin Zhuangzhuang recounts: “For smaller restaurants that are often slow, drivers will often go and try to make friends. Hand over a few packs of cigarettes, and have a smoke with the boss. Joke around with chef, and maybe he’ll do your order before the others. For big restaurants, they’ll make friends with the woman manning the register or the guy who boxes up the takeout orders. Chat them up, and they’ll use the intercom to tell the chefs to hurry up. It definitely helps.”

But that doesn’t solve the basic problem, and conflict between drivers and merchants often occurs. One takeout deliveryman in Jinan got into a free-for-all with workers at a Xicha tea shop, while in Wuhan, a driver stabbed a restaurant worker during an argument, leading to the worker’s death.

Drivers have seen many conflicts with merchants, some so serious that the police are called. As for how to solve the problem, one driver says, “Increase the delivery time. If there’s enough time for everything, no one’s going to get so tense about it.”

But in reality, delivery times have gotten shorter and shorter, while drivers face calls to increase their speed, and merchants, who are among the players that take up the drivers’ time, have their own opinions about the matter.

In the process of talking with merchants about their partnership, Li Lei realized that what merchants spoke about most was the speed of the drivers. If a driver’s speed doesn’t meet the merchant’s expectations, they will bring it up with Li Lei and ask for a different driver from the dispatch station, or ask to be taken off the station’s list. Often, in a busy business district there will be several dispatch stations for the same platform. A merchant can freely choose which station to partner with. As for what merchants require from a dispatch station, Li Lei says, “It’s simple. First is the delivery capacity, and second is the speed the drivers reach their restaurant.”

In order to attract high-volume businesses, Li Lei will go to the dispatch station to push the drivers to go faster. But if a driver makes a late delivery because of slow food prep, which affects the station’s numbers, all Li Lei can do is go to the merchant talk it over or to encourage the restaurant to move faster. But not everyone can do that: “You can only go hurry a restaurant if you have a really strong personal relationship with them.”

When Li Lei goes to a restaurant, he will keep his eyes glued to the computer screen as takeout orders come in, not letting his attention stray for a second. As soon as the ding of an order happens, his own voice will join in, saying, “Meituan order. We’ve got a Meituan order.” He says, “You have to be on top of it every second.”

 

Peppa Pig and Cola   

            Once, in a conflict with a customer, Meituan driver Xiaolin discovered a hidden “secret” in the system. The customer had received a delivery time that was different from the time he had received.

At that time, he’d started driving as a crowdsourced driver for Meituan. He accepted an order and hurried to restaurant, where he received a confrontational phone call from the customer: “Why aren’t you here yet? You’re super late!” Xiaolin thought the customer was just stirring up trouble, since his cellphone showed that he still had nearly ten minutes to make the delivery. When he made the delivery, he got into an argument with the customer about the timing, so the two pulled out their cellphones to compare. The customer’s “anticipated delivery time” was ten minutes earlier than the driver’s “required delivery time.”

Since discovering this “secret,” Xiaolin has called the Meituan customer service line every month for nearly four years. Each time the representative is different, but they answer in exactly the same way: “You should explain to the customer that this is only an anticipated delivery time.”

That doesn’t fit with what Xiaolin has experienced. Many drivers brought this problem up with People. To them, the system is pandering to customers, and it’s just a way to keep them coming back. But it’s also one of the main sources of conflict between drivers and customers.

As the researcher Lu Taihong pointed out in the book Consumer Behavior: Perspectives on Chinese Consumers, the conveniences offered by the digital age have led to consumers becoming more and more demanding. They pay more and more attention to service and product experience, but loyalty to specific products and brands has fallen. Customers are liable to switch suppliers at will, and because of this “they have an unprecedented, enormous influence on and power over the market.

Faced with this kind of influence, the takeout platforms value numbers of users and orders, and use algorithms to construct a kind of power structure. In that structure, customers are at the very top and wield supreme power.

But customers can also make mistakes. “Sometimes customers…they’re unbelievable.” On this topic, Gansu driver Wang Bing has a lot to say: “Lots of people don’t even know where they live. They live at number 804, but they’ll put down 801. They clearly live by the south gate of a complex, but they’ll put down north gate. Then there are the customers who order food and then forget about it. I’ll call them and no one will answer. Then the next day, the customer will call me and say, where’s my food.... Some people will put in an order and not even pay attention to the address they put in, and I’ll look at the order and think, this isn’t right, this address is in the next province!” But customers don’t have to pay for their own mistakes; if an order is late, it’s the driver who is punished.

Sun Ping, a social scientist who has studied the plight of takeout deliverymen for a long time, also mentions the issue of the “customer’s supreme power” in a research article. In the takeout delivery process, the customer can learn almost anything about the driver: name, cell number, percentage of on-time deliveries, how many positive ratings they’ve gotten, how long they take to pick up food, what route they take, and how long it will take them to arrive. Before the food arrives, the customer even has the option of canceling the order.

 “They can see everything, the whole process, but we don’t know anything about who they are. And if a problem comes up, we don’t have the option of canceling the order like they do,” one driver complained to Sun Ping. He added one of his own experiences with canceled orders:

“I had two orders, one 1.5 kilometers away with 45 minutes left, and one 3 kilometers away with 20 minutes left. I went to pick up the food for the further order first. The 1.5-kilometer customer got mad at me, because he saw on the GPS that I passed by his home without delivering his food. He was really angry, canceled the order, and put in a complaint against me with the platform.”

In the course of People’s investigation, several drivers described similar experiences. That day, when the customer finally received his food, he said to the driver, “Well, isn’t my order the only one you have?”

With faster and faster delivery times, along with a one-sided review process and a system that loves them, customers have become more and more impatient.

Shanghai-based Jingjing admits that he’s been “spoiled.” His work keeps him busy, he doesn’t know how to cook, and he relies almost entirely on takeout for his meals. He often orders from a nearby sandwich shop, and he remembers when it used to take about 45 minutes from the time he ordered until his first bite of Caesar salad. To pass the time, he would often watch a 45-minute TV show. Lately, however, his waiting time has been reduced to 26 minutes. Not long ago, a delivery took more than 30 minutes, and he couldn’t bear the wait, and called half a dozen times to ask about the order.

In 2017, the French market research company Ipsos carried out a survey of consumer “impatience” in twelve major cities in China. They showed that advancements in mobile technology made consumers more impatient in every one of the cities. This phenomenon was most striking in economically developed areas and among younger consumers. Among those groups, “Beijing consumers were the most impatient.”

Drivers have to come up with their own ways to appease these increasingly impatient customers.  

Wang Bing has a lot to say on the topic. In a handful of orders with similar delivery times, he will chose the most expensive one to deliver first. Those customers who spend the most money are the most likely to lose their temper. “You can explain all day to them, but they won’t listen. They’ll suddenly explode and tell you to take the food back. I can’t afford to lose a hundred yuan order every day.”

Drivers also try their best to satisfy the customers’ requests that come along with the order: for example, picking up a pack of cigarettes or some water, or even “to bring a razor to an internet café.” For while, under the influence of the website Douyin, customers would ask Wang Bing to draw a picture of Peppa the Pig and bring it with their takeout order. If he refused, they’d give him a bad review. Wang Bing was furious, but couldn’t say no. “I’d buy some paper and draw Peppa, and underneath I’d write, Are you some kind of idiot?”

“The delivery business is a form of social theater with the customer at the center,” Sun Ping wrote in a report. She calls drivers’ attempts to please customers and garner five-star reviews “emotional labor.” She believes that this aspect of their work is often ignored, but in fact it is even more depleting than the physical labor involved.

In her conversations with People, Sun Ping mentioned one driver who left a deep impression on her. “In three days, his motorbike was stolen twice, and his battery was stolen three times. He cried when he told me about it. He said, ‘The platform requires us to say please enjoy your meal! But nobody cares that I’m from the countryside, and I’ve grown food myself. It’s humiliating to sound like a waiter, and beg for five-star reviews. I’m a man, how am I going to make myself say it?’”

When the news website Jiemian conducted interviews about the “SKP incident,” Shanghai Jiatong University public economics and social policy associate professor Sheng Yang expressed that although deliverymen might have a monthly income that exceeded 10,000 yuan, they still occupied an unfairly low social status: “In order to make that amount of money, they are forced to sacrifice their time and health, and work harder than others, both in terms of manual labor and emotional labor.”

Wang Bing has yet another tactic to placate customers. In the summer, many customers will order a cup of cola along with their meal. But it rains a lot in the summer, and with the number of times his motorbike slips, the cola is usually done for. If he goes somewhere else for a refill, not only will he have to spend his own money, but he’ll end up delivering the order late. To avoid upsetting customers, he keeps a bottle of cola in his back baskets, and if a drink spills, he’ll find somewhere private and refill the cup with cola from the bottle, and wipe the cup clean. The strategy works for him.

At the same time, nervous customers have posted on several legal advice websites. One posted: “I kept telling the deliveryman to hurry up and he got into an accident. Am I legally responsible for anything?” Below, a lawyer responded: “You have no legal responsibility.”

 

Games

Recently, Meituan and ELEME came out with their financial reports for the second quarter of 2020. During that quarter, ELEME managed to turn a profit on every single order, while Meituan made a net profit of 2.2 billion yuan, a 95.5% increase of which takeout constituted the largest source.

On August 24, 2020, Meituan stock prices also shot up, and the market value exceeded $200 billion US dollars to become the fifth largest company on the Hong Kong Stock Exchange.

In our investigation over the course of six months, People had contact with nearly 30 takeout delivery drivers, and they frequently mentioned one word: mao.[1]

A Hunan Meituan driver said, “If your on-time delivery rate drops below 98%, you lose one mao per order. If it goes below 97%, you lose two mao per order. Isn’t that just a way to force drivers to go faster? After all, to us one mao less on each order is a lot of money.”

One Shanghai ELEME driver told us, “The lowest order ELEME will accept is for 4.5 yuan. The farther away, the higher the threshold is. Sometime one mao makes all the difference in the way you feel about an order. There’s a big difference between a 4.9 yuan order and a 5 yuan order.

In order to keep that “one mao,” drivers are not only willing to go faster, but also further.

This is what the system hopes to see. The system also conceals another secret: one involving a “game” of rankings.

Whether Meituan or ELEME, the system creates a point-based ranking system. The more orders you deliver, the more on-time deliveries you make, and the more good reviews you’re given, the more points you get. The more points you get, the higher your ranking, and the more bonus money you receive. The system packages this review system like a game. Different ranks of drivers have different titles. On Meituan, for example, the titles from lowest to highest are: ordinary, bronze, silver, gold, diamond, and king.

One crowdsourced driver for Meituan in a city in southeastern China describes the details of the system: In a given week, if you successfully deliver 140 orders with an on-time rate of 97%, you become a “silver driver,” and each week you can receive 140 yuan in extra bonuses. If you deliver 200 orders with a 97% on-time rate, you become a “gold driver” and can receive 220 yuan in bonuses. On ELEME, the number of orders and the compensation are directly linked. For below 500 orders a month, each order earns 5 yuan for the driver; from 500 to 800, each order earns 5.5 yuan; for 800 to 1000, each order earns 6 yuan, and so on. But according to the rules of the game, a driver’s credits return to zero at the beginning of each week or each month.

In her research article, “Orders and Labor: An Economic View of the Algorithms and Labor Practices of Chinese Takeout Platforms,” Sun Ping writes that aside for the punishments for late deliveries, the system also uses a gamification of assessments to draw many drivers into an unending spiral. “They want to make us to work all night and then work all the next day,” one driver told her. But they have no way out: “Last month I was a black gold knight. If I want to keep my status, I need 832 points. I have a lot of work to do.”

“The higher the status, the more pressure drivers face to maintain that status.” According to Sun Ping, this kind of gamification not only has the potential to elicit addictive behavior, but it also combines the drivers’ self worth with capital management, while the game-like exterior provides “a normalized, internalized, seemingly reasonable cover for exploitative algorithms.”

According to a Meituan public announcement, “Report on the Driver Profession for the First Half of 2020,” right now, there are a total of 2.952 million Meituan drivers. ELEME’s official website Fengchao reports a total of 3 million drivers. With more than 6 million people whose work has been completely systematized, Central China Normal University social sciences researcher Zheng Guanghuai refers to the concept of “downloaded labor.”

In the research report “Wuhan Package Deliverymen and Takeout Personnel: Platform Workers and ‘Downloaded Labor,’” Zheng Guanghuai’s team offers a penetrating explanation of this concept.

Drivers use a “downloaded” app to enter the profession. Superficially, this app is just a tool to assist them in their work, but in reality, the drivers “download” a finely tuned method of control over their labor. With this method, “workers’ subjectivity is completely remolded and superseded.” They seem to have a job that involves more freedom, but at the same time, “they are subject to more control on a much deeper level.”

“The platforms use downloaded labor to create ‘platform workers,’” Zheng Guanghua’s team reports. Particular to this form of labor is its attractiveness to workers, weak contracts, observation and control mechanisms, and few means of resistance against it.

It is the drivers’ own cellphones that enables the system to create this pool of “downloaded labor.” Yet, even though it is a vital work tool, takeout platforms have made many announcements about their great efforts to help their drivers set aside their cellphones.

“We worry that drivers will have accidents on the road as they’re accepting orders.” In an April 2018 interview with 36Kr, Meituan delivery algorithm team leader He Renqing emphasized, “Meituan’s most intractable problem is how to prevent drivers from looking at their cellphones while they’re on their motorbikes.”

Meituan spent seven months developing bluetooth earphones with smart speech technology and two-way communication. According to He Renqing, the earphones are wind and water resistant, have noise-cancelling capabilities, and are made with smart technology. All the drivers have to do is put them on, and they can do everything they need to do with voice commands, enabling them to put down their cellphones while they drive.

In reality, however, among all of the Meituan drivers who spoke with People, not one had received or used this kind of smart bluetooth earphones, and none of them had been able to give up using their cellphones.

Although she experienced only a few days of the life of a driver, Cao Dao has lingering fears about how much control the cellphone had over her. “For example, while you’re using the navigation, the system will keep reminding you, Meituan crowdsource has a new order, please check the information. And it will keep talking over the voice of the navigation. Then you’re about to be late and the customer calls to ask where you are, so you’re listening to the navigation and trying to accept an order and explaining to the customer why you’re late….” Cao Dao said that it all made her feel like absolutely every second matters, like someone was chasing after her. “You just have to keep going faster and faster.”

 

Electric Motorbikes

“You can’t lose any time on the road, since they give you the shortest delivery time possible,” one ELEME driver told People. Another Meituan driver told us that once an order was in his hands, all the time that was left was spent on the road, so “unless a cop is right on your ass telling you you can’t speed, whenever you have a bunch of orders, all of us practically take off flying.” Then he added, “Even if you fly there isn’t enough time.”

In those moments, all the driver has to rely on is his motorbike.

Before they begin work, a driver has to figure out his own transportation. Usually, a dispatch station has a long-term partnership with a third-party company, which offers motorbikes for lease to drivers. In order to lower their startup costs, most drivers will choose to lease a motorbike for a few hundred yuan. But the majority of these motorbikes have seen a lot of action: some don’t have mirrors anymore; while on others, the pedals and front of the bike are held on with six or seven layers of heavy-duty tape. One driver said that eventually he had become his own “motorbike repairman.”

If a driver does not want to lease a bike, some stations will encourage their drivers to buy a bike on an installment plan.

One Chengdu Meituan driver was required by his station to buy an off-brand motorbike for more than 1000 yuan more than its market price. Another driver said that he’d spent several thousand yuan to buy a motorbike from his station, but he hadn’t driven it more two days before the battery gave out.

Compared to those colleagues who wasted their hard-earned money, Meituan driver Wang Fugui considers himself lucky. On the first day of his job, he and the battery in his motorbike flew off at the same time, and he struck his head against a guardrail. The bike was leased from his station for 200 yuan a month, “and it was just a bunch of parts held together with spit.” It had no lights, and the brake pads were so worn away that sometimes when he stepped on the brakes, they didn’t respond at all. Every so often when he hit the gas, the bike would roll backwards.

But none of that was an issue for him. The day after his bad fall, he spent ten yuan of his own money on brake pads. When he drove at night, he would hold a small flashlight in his teeth or tape a flashlight to the front of his bike to substitute for headlights. After all, the motorbike did have its good points. “It’s super fast. It can hit 65 kilometers an hour,” Wang Fugui said.

According to statistics put out in 2018 by the Public Security Bureau, in China between 2013 and 2017, there were 50,620 traffic accidents involving electric motorbikes, leading to 8,431 deaths and a loss of 111 million yuan in property damage. In April 2019, the country formally put in place a new stricter national standard for motorbikes. The regulations require that motorbikes not exceed 25 kilometers per hour. A motorbike that conforms to the national standards sells for at least 1000 yuan.  

But among the more than 30 drivers that People spoke to, none of the drivers from either Meituan or ELEME drove a motorbike that met the new national standards. Their motorbikes generally could reach 40 kilometers per hour, much faster than the limit. On online forums like Qishouqun and Tieba, there are many discussions of how to retrofit new motorbikes so as to get around their speed-capping technology.

After working as a driver for a year, Wang Fugui’s junker of a motorbike broke down more and more often, and he sometimes had to take a cab to deliver his orders. Fortunately, he lives in a small town in the northwest, and compared to what he would lose delivering late orders on his bike, taking a cab saved him money. For 50 yuan in cab fare, he could easily deliver ten orders.

Finally he decided that he should grit his teeth and buy a new motorbike in order to travel faster. The old bike he had leased was broken down into spare parts that are now used in an unknown number of other leased bikes.

Driving his old motorbike and then his new one, Wang Fugui made his way up in the rankings to number five and finally to number three in his region. But after a while, he ended up quitting. He couldn’t handle a new Meituan requirement that he attract new users: “Meituan wants to expand their business, so they made us go out and find new users. You have to get two new users a day to install the app and make a new profile. I did it for a few days, but then I just couldn’t take it anymore and I quit.”

 

Smile!

When the phrase “takeout delivery driver has become the most dangerous profession” went viral, the system also responded.

In the early days of these platforms, both Meituan and ELEME provided their drivers with safety training. But nearly all of this training occurs right when a driver begins working for the platform, when both specialist and crowdsourced drivers must pass a simple safety test in order to start receiving orders.

Station supervisors will often warn specialist drivers to be careful. One Meituan station supervisor told People that each time he does a safety training, he shows a video that he made himself of a montage of motorbike accidents. After his 300-plus drivers have watched it, he tells them gravely, “I know you’re in a hurry, and sometimes you have to drive against traffic. But you have to take care and watch the road.” Another ELEME dispatch station supervisor echoed this sentiment: “No matter how many times you tell them, what drivers are focused on is not time, and sometimes it’s not even their own safety—what they’re focused on is fear they’ll be late on an order.”

Faced with an ever increasing number of traffic accidents involving deliverymen and in order to increase their drivers’ awareness of safety issues, takeout platforms have come up with new strategies, such as asking policemen to come give lectures, organizing groups of drivers to go to the traffic police station to be tested on safety procedures, and so on. Meituan even designed a yellow headband in the shape of kangaroo ears. The ears have safety slogans written on them, the most common being: “Busy with delivery, but never forgetting safety!” On the back is: “Everything’s fast with Meituan!” But in reality, these two slogans are in conflict. Most drivers didn’t want to wear the ears, because, as one driver told People, “They’re so annoying. As soon as you start going fast, the ears get knocked off by the wind.” For safety, the system also added a new feature: after a driver’s cellphone connects to the system, an informational video about safety will randomly appear.

“Often when I’m out making deliveries, the system will suddenly bar me from accepting orders. I have to stop my bike right away and watch the video before the system will go back to normal,” says the Hunan-based Meituan driver Adou. Once when he was making deliveries during rush hour, he had to suddenly stop by the side of the road to watch a safety video, and he was struck by a speeding bicycle, spraining his ankle and forcing him to take off work for a while.

Sometime around June 2017, Meituan began its “Smile!” initiative. This is a systematic spot check of conduct, which can happen at any moment. During a check, drivers must immediately stop their motorbike, and take a photo from their chest to head, demonstrating that their face is clearly visible, and showing their helmet, uniform, and badge. All this must be done within five minutes. If the photo isn’t uploaded in time, or the content is not up to spec, the system may decide that the driver has failed the test, and he will be faced with a fine between 300 and 1000 yuan. Drivers may also have their account frozen for three days, or even permanently shut down.

As soon as the “Smile!” initiative came on line, it became a kind of metaphysical challenge for Meituan drivers.

Each driver has had the test appear at different times: climbing stairs, waiting for an elevator, waiting for the food, at peak delivery times….

For Adou, the most unforgettable “Smile!” happened at peak delivery time during a rainstorm. That day, he was wearing a rain slicker and he couldn’t see the road very well, but he had to pull his bike over, take off his slicker to show his badge and uniform, and take a photo of himself. Another driver from the same station had his cellphone in his pocket and didn’t hear the “Smile!” announcement. He ended up being fined 400 yuan.

On a rainy day this past February in Nanchang, Jiangxi, a driver who has cerebral palsy didn’t have enough time to take the photograph, and his account was subsequently shut down. Fortunately the event was caught on a video that was uploaded to Douyin and attracted a lot of attention. After receiving a flood of online feedback, Meituan quickly unfroze his account.

Not every driver receives that kind of special treatment.

Every day on drivers’ forums, the same question is posted: My photo clearly met the requirements, but it failed to pass, and when I appealed I learned that the system will not allow my account to be unfrozen. “Our voices never reach top management,” one driver complained.

At the same time, some photos that obviously do not meet the requirements do pass muster. One driver from Shenzhen revealed that after his own account was frozen, he used his wife’s account and continued to make deliveries. But his photo passed the “Smile!” check. Other drivers will prepare a photo in advance, and pass the check that way.

With the pandemic, wearing a mask became one of the elements of the “Smile!” spot check. One Hubei driver said that his mask got soaked from rain, and he didn’t have time to get another one. The system failed him and prevented him from taking orders. Another driver from Guangdong took a photo of himself with his hand covering his mouth and passed the spot check.

Last winter, in Hailar, Inner Mongolia, a Meituan driver was spot checked as he was making deliveries. In the minus-30-degree Celsius weather, he pulled over, took off all of his winter gear, showed his Meituan uniform and helmet, and uploaded a photo, all within the five minutes allowed. The drivers interviewed by People called the “Smile!” initiative “terrible,” “inhumane,” and “a total waste of time.”

ELEME has a similar system of spot checks called “Blue storm.” The difference is that the “Blue storm” checks allow fifteen minutes for a response, and the punishments are comparatively light, between 5 and 30 yuan. When People carried out its investigation in 2019, none of the ELEME drivers complained about their spot checks.

But all good things come to an end, and according to leaks from ELEME drivers, in order to compete with Meituan, this year ELEME’s “Blue storm” spot checks will also come with a five-minute window.

 

“Five-star Reviews”   

 As traffic violations and accidents involving drivers became more common, the traffic police went from being outsiders with respect to the system to being a concerned party.

Xiong Chongjun is a Shenzhen traffic policeman, and he has spent almost ten years directing the traffic department’s public outreach efforts. Many of his traffic enforcement videos have gotten popular on the web and he’s become such a recognizable figure that he’s acquired the moniker “Shenzhen’s Officer Xiong.” Last summer, he compelled two Meituan drivers who were driving against traffic to write self-criticisms and read them aloud. The incident went viral under the tag: “Your deliveryman might be out writing a self-criticism.” One netizen commented: “Officer Xiong is too kind, the punishment is too light.”  

In fact, in the past two years, traffic police departments across China have developed policies to deal with takeout delivery drivers.

In the Pudong district of Shanghai, each takeout driver is required by the traffic police to wear a vest that shows a digital image of their unique serial number. At the same time, each driver must have a “takeout driver good behavior score card.” Each card is loaded with 36 points. Traffic policemen and sensors both enforce the rules: if a driver does not wear his vest, 12 points are deduced; driving without properly registering a motorbike, 12 points; running a red light, 6 points; driving the wrong way, 3 points. When the 36 points have been exhausted, the driver will be faced with a closed account or being fired by the company. Pudong is the first place in the country to require drivers to wear a digital vest.

One after another, Gengtai city in Hebei, Guangdong, and Shenzhen all followed Shanghai’s example and began to use a system of points for good behavior. Qingdao enacted a system of blacklisting drivers who broke the rules. In Jiangsu, drivers who break traffic rules are barred from one day of work for each traffic infraction. In Nanjing, drivers who have a second infraction must do a day of training.

With all of the intense time pressures weighing on these drivers, however, none of the police efforts have been much use.

In December 2019 and May 2020, People went to Pudong’s Lujiazui neighborhood to investigate the digital vests drivers are required to wear. Of the drivers who were spotted on Century Avenue within a given daytime hour when there are more traffic police on duty, approximately 70% were wearing their digital vests. However, even some drivers who are wearing their vests still choose to break the rules.

This is the result of a “careful calculation” made by the drivers. In the daytime with more police around, it’s likely that they’ll be caught if they aren’t wearing their vests, and fined 12 points. But if they wear their vests and are caught breaking a traffic rule, “running a red light or driving against traffic still involve fewer points.” At night, the percentage of drivers wearing their vests drops dramatically. The reason is simple: “The cops are off duty.”

As enforcers of the law, many policemen including Officer Xiong feel conflicted. The people they see most often breaking traffic regulations are takeout drivers, but at the same time, officers understand their dilemma.

Officer Xiong told People that he has often been on the scene of accidents involving takeout drivers, and sees the overturned motorbikes, damaged vehicles, injured pedestrians, injured deliverymen. In his observation, the first reaction of the deliverymen is to jump up, check whether the food has spilled or not, and then to call their customers to explain. “They’re never concerned about themselves.”

This has given him deep sympathy for the difficulties the drivers face. Officer Xiong says that he often talks with the drivers, and their thoughts are very simple. They don’t want to make late deliveries, they don’t want to get bad reviews, and they don’t consider their own safety carefully. “Personal safety is never their first priority. Their priority is getting the food into the hands of their customers on time.”

As a frontline traffic policeman, Officer Xiong believes that the whole situation is being created by the fierce competition between takeout platforms. At the same time, it also shows that in many cities there are not enough lanes for non-motorized vehicles. “The competition between platforms leads to shorter delivery times and more anxious drivers. They have to choose between breaking the law and making a late delivery.”

For this reason, when some police officers come across a driver breaking the rules, they make some allowances while still enforcing the law. On the day that Officer Xiong made the drivers write self-criticisms, he told them to find a place in the shade to write. And often, police will even help the drivers deliver their food.

Many such cases have been reported in the news.

On March 25, 2020, a takeout driver in Tongxiang, Zhejiang was caught driving against traffic. His punishment was to stand at an intersection and urge others to follow the traffic regulations. He told the police that he had just accepted an order and hadn’t had time to pick up the food, and if he made a late delivery he would be fined. Finally, the traffic policeman told one of his assistants to take the driver’s motorbike and go pick up the food. On the way, the engine died three times. The officer finally reached the customer’s building, but when he got off the bike, the food toppled over onto the ground.

Fortunately such situations are uncommon, and usually the traffic police manage to deliver the food without a problem.

On April 16 in Nanchang, Jiangxi, a Meituan driver rushing to deliver food committed three traffic offenses in a row, and was stopped and punished by the traffic police. In early June, in Wenzhou, Zhejiang, an ELEME driver was detained for driving an illegal motorbike; on June 29, a traffic officer in Dongguan, Guangdong was out on patrol and spotted a Meituan driver without a license plate on his motorbike and immediately detained the driver. In each case, these drivers were unable to complete their deliveries, so the officers or their assistants made the deliveries for them.  

After delivering the food, nearly all of the traffic officers did the same thing: they told the customers, “Please enjoy your meal, and please leave a five-star review.”

 

The Final Screen

            Among all the drivers People contacted in the course of this investigation, Shi Chen is a unique case. Because “your life is worth more than a late delivery fee,” he insists that in more than a year as a deliveryman, he has never run a red light and never driven against traffic. Every day, he wears a tidy uniform and helmet.

Despite all of that, he has still been in accidents. One evening in July 2019, he was hit by a car while making deliveries. The impact broke his right ankle, and when the traffic police arrived, they determined that fault lay with the driver of the car. After he was taken to the hospital, the owner of the car paid for his expenses, including surgery and the rest of his medical care costs.  

As a specialist driver, Shi Chen’s station deducts 106 yuan from his monthly salary for insurance, including accident insurance. In ordinary circumstances, Shi Chen would receive compensation from the insurance company as well. But when he was released from the hospital and contacted the station, he realized that his driver’s account had been deleted.

The station claims that the reason for this is that since Shi Chen was in the hospital after surgery for a long time and was unable to make deliveries, he did not make his work target and so was deleted by the system. When a driver’s account is deleted, the record of his insurance payments is also deleted, and Shi Chen would have to go to the insurance company directly to settle his claim. In order to retrieve the record of his payments, he tried to communicate with Meituan via his station, but realized that he had been kicked out of the station’s online group.

Shi Chen is certainly not the only driver to have encountered this kind of situation. In the system, insurance is the sole, and the ultimate, safeguard that drivers have; however, People’s research uncovered that in most cases, when a driver has a traffic accident, they have difficulty getting compensation. 

The takeout platforms are designed so that the insurance fees for specialist drivers are taken out of their salaries each month by their individual dispatch station, and the station is also responsible for setting the specific amount deducted. Crowdsourced drivers pay for their insurance by the day at the rate of three yuan per day. They are covered from the time they accept their first order of the day until midnight that same day; if at midnight, the driver is still making deliveries, the insurance can only be extended for one and a half hours.

According to the social scientist Zheng Guanghuai, this kind of work insurance system ingeniously shifts responsibility away from the takeout platforms.

In an interview he gave to the culture section of Jiemian on Labor Day last year, Zheng Guanghuai described the takeout platforms as “managers who wash their hands of the situation”: “the platforms outsource to other companies, thereby cutting off any direct relationship [with their workers]. Workers buy their own insurance for protection if they are injured, and if a worker is in an accident, the platform pushes the responsibility off on the insurance company.” This kind of ‘shifting of responsibility’ leads to “an ambiguous labor relationship that makes it even more difficult for workers to protect their rights.”

Sun Ping has also investigated this issue. If an accident just involves bumps and bruises, most of the drivers she spoke to choose to deal with it themselves. “A lot of them told me that the process of getting compensation was too long and complex. It’s so onerous that they’d rather take care of everything on their own instead of engaging in a complicated process.”

But if an accident involves something more serious, what happened to Shi Chen will play out yet again.

 One crowdsourced driver for ELEME told People that he’d had an accident while delivering food. He collided with a pedestrian and ended up in the hospital. The insurance company delayed for a year without paying him any compensation; finally he had to get an online loan in order to pay his medical bills.

When a driver from Suqian began working for Meituan, the station supervisor required him to fill out a “Proof of Voluntary Refusal of Insurance Contract.” He felt uncertain about it, so the manager laid it out for him: Driving is the most dangerous profession, and each day could be his last, so none of them had any guarantees of anything. This is not unusual. Former Meituan distribution station supervisor Jin Zhuangzhuang says that insurance for crowdsourced drivers is paid directly through an app and is obligatory. But specialist drivers’ insurance is paid through each station, and “a lot of stations think it’s a hassle, so they don’t give their drivers the option of getting insurance.”

There are others who don’t have the option of any guarantees: pedestrians who are hit by takeout drivers.

Last April, Lin Wei was on his way home when a Meituan driver hit him, breaking his left leg. It was the driver’s first day on the job, and the station supervisor said that there hadn’t been time to sign him up for insurance, and moreover the accident had nothing to do with the station: “We tell our drivers to go out and deliver food, not go out and hit somebody.”

After much negotiation, the station offered the following solution: they would try to convince the driver to split the injured party’s medical bills and living expenses.

Eventually, the issue was resolved by “personal connections.” A manager at Lin Wei’s company knew a Meituan executive, who put pressure on his subordinates until the station finally agreed to pay for Lin’s medical expenses.

A netizen put the following comment below a post on a social media platform written by a Meituan driver about drivers’ rights: “Deliverymen help Meituan send out tons of orders and drive up its market value. But Meituan, a company that has been relying on takeout to grow its business, doesn’t provide any kind of formal work agreement for its deliverymen.”

A year after his accident, Shi Chen’s driver’s account is still shut down, and he hasn’t received any compensation from the insurance company. He told People, “I decided to leave the profession, and never go back.” But those drivers who are still out there hurdling down the roads can only silently pray to themselves. As the Meituan driver Wei Lai, who once witnessed a fatal crash at an intersection, wrote in his online blog: “I hope all the drivers safely return home.”

 

An Endless Game

When Cao Dao uploaded her video about her experiences as a takeout driver, she was driving across China making a new film. As she headed through an unpopulated part of Tibet, she remembered her few days as a takeout driver to People, and recalled the stifling feeling.

As someone with brief personal experience with the system, Cao Dao had a recommendation: All of the directors and algorithm engineers in the takeout platforms should be forced to spend a month as a driver. “That way, they’ll understand how much pressure the system puts on people.”

In a report describing how the Meituan system reduced delivery times to 28 minutes, a driver raised a similar idea: “Why don’t you all come down to the frontline and make delivers for a few days? See if you can do it in 28 minutes without running red lights, driving the wrong way, and going at crazy speeds.”

To some extent, this suggestion accords with the work of Nick Seaver, a sociologist who studies technoculture.

Seaver came up with the idea of “algorithms as culture.” In his view, “algorithms are not only formed by logical processes, but also are formed by institutions, people, intersecting environments, and rough readymade assumptions drawn from ordinary culture and life.” He believes that algorithms are “created by people’s collective practices,” and he recommends that scholars should research algorithms from an anthropological perspective.

As a scholar, Sun Ping completely agrees with Seaver’s analysis. But in reality, algorithms are still mostly created on the bases of numerical logic.

“It is essential to improve the training and moral values of computer programmers. But the current situation is China is that most of our programmers think completely in terms of science and technology, and very few of them think in terms of social science. So they lack the concepts necessary to consider questions of fairness and moral values.”

In her research, Sun Ping has spoken with algorithm programmers, and she discovered that these programmers have their own kind of logic and do think about all kinds of unexpected circumstances. However, these programmers only carry out the plans of the designers: “The designers of the rules are the takeout platforms, and the programmers are only implementing the platforms’ decisions.”

In the course of this investigation, People made many attempts to contact the platforms’ algorithm teams, but all of them refused to discuss the system due to “company regulations.” One Meituan algorithm engineer told us, “This is all top secret at the company.”

Sun Ping says that the “one-sided right to speech” is now the biggest problem for these algorithms. But for the system as a whole, the most intractable problem is how to make drivers go faster and faster while denying responsibility for the consequences, as well as for the drivers themselves.

This constitutes an even larger and more invisible game: “Each time a driver makes a delivery, all the data produced will be transferred onto the platform’s cloud, and this makes up the majority of their databases.” Sun Ping says that the system requires the drivers to go faster and faster, and faced with punishment for late deliveries, the drivers try their best to meet the requirements laid down by the system. “The deliverymen work fast and faster, and as such add to the system’s ‘short-term data.’ The algorithms are based on the data, so when an algorithm goes through the data and realizes that everyone’s getting faster, it increases the required speed.”

In Sun Ping’s view, the collection of data produced during deliveries involves questions of proprietary rights; yet drivers are still out there running against the clock. According to the latest announcement from Meituan, in the first half of 2020, in 2800 cities around China, drivers “disregarded the pandemic and worked around the clock to bring food, groceries, medicine, and other essentials to more than 400 million users.”

After Meituan broke the news that their market value had risen above 200 billion US dollars, amid the astonishment, some people brought up Wang Xing’s obsession with speed, and Finite and Infinite Games, a book Wang once mentioned as “having a huge effect on me.” In the book, New York University professor of religious history James Carse divides all of the games on earth into two categories: “finite games” and “infinite games.” The former refers to games that end with a definitive winner, while the latter refers to games that go on forever.

The system is still grinding on, and the game continues. But the drivers seem to have no idea that they are players in this “infinite game.” They’re just out there running and running, hoping for a chance at a better life.

 

(At the request of the interviewees, the drivers in this article have been given pseudonyms.)    

[1] Translators note: A mao is one-tenth of a yuan.