In China, Uber and Competitor Didi Are Still Plagued by Driver Fraud

Is China's ride-hailing system doomed due to corruption?

|
May 31 2016, 9:00am

During his days as a driver for Didi, China's largest ride-sharing service, Mr. Zhang would often lose money sitting on the highway. He remembers one day driving the length of the city, from north to south, without getting a single booking.

You know how horrible Beijing traffic is from 3 to 8 PM, he tells me. For hours, "my car was just stuck in the middle [of the highway], empty."

That day, Zhang decided to take matters into his own hands. If business wasn't coming to him, he would make his own. He took out his phone and started surfing online for an accomplice. In particular, he was looking for someone to help him perform click fraud.

A new form of fraud

Click fraud usually refers to the practice of hiring low-paid laborers to click on profitable internet links. Click fraud might take the form of companies buying fake Facebook "likes," for instance, generated by overseas click farms that pay workers $15 per 1,000 thumbs-up.

For Uber and Didi, click fraud means something else.

The general idea behind click fraud is simple: Drivers partner up with customers who book fake rides. Then the pair split the driver's payment, which is boosted by the large bonuses Uber and Didi give to drivers.

It's a system buoyed by the various incentives both companies have set up to establish business in China's still relatively-young ride-sharing market—namely, generous driver bonuses and free ride vouchers given to new sign-ups.

China's "sick" drivers

Zhang, who no longer works for Didi, told me that the industry uses the euphemisms "nurse," "patient," and "injection" to refer to the components of click fraud. The patient is the driver in need of business, the nurse is the fake customer, and the injection is the fake booking that brings in payment for the driver.

Drivers can request injections to boost profits during a shift. A 'sick' driver might request an injection from a nurse, for instance, on his way to pick up an actual customer. By coordinating a ghost ride en route, he can make money during transit and maximize his earnings.

Another lucrative time for injections is during surge hours—"particularly during morning or afternoon rush hours, which are the high-bonus periods," said Zhang.

Drivers can also trick Uber's platform into believing that they are serving real customers, when they are in fact sitting at home

Nurses often hoard real vouchers for free rides, which Uber and Didi give out en masse to build their user bases. When a driver asks for an injection, a nurse can redeem a coupon for a pick-up near the driver's location. This prompts the system to match the booker with the driver.

Mr. Zhang told me that during his click fraud days, he might pay a nurse 20 yuan ($3 USD) for redeeming a 50 yuan ($8 USD) coupon. His nurse would book a route that worked for him, he'd run the car empty for 17 or 18 kilometers, and then he'd split his profit with the nurse once the ride was over—all without ever meeting in person.

This scheme can also be done without the help of coupons, but each party profits less.

Uber the underdog

Uber has been more vulnerable to click fraud than Didi, largely because of its generous driver bonuses in China, which can reach up to three times the fare of a ride.

Didi did not provide me an exact number of how much click fraud the company deals with, but spokesperson Sun Liang acknowledged over the phone that click fraud is a "minor issue" for Didi. In regards to the problem, the company is allocating more resources towards big data and deep learning to improve its anti-fraud detection tools, said Liang.

Uber declined to comment.

Overall, while Uber's high driver subsidies may work to attract new drivers, they also prime the pump for fraud. People looking to cheat the system have already purchased hundreds of thousands of Uber driver accounts through e-commerce marketplaces such as Taobao, China's largest consumer-to-consumer retail network.

This black market lets people bypass Uber's vetting process for drivers. It also allows a single driver to perform click fraud with multiple accounts, using hacked software that can simulate a fake ride with "real-time" location coordinates.

Using this strategy, drivers can also trick Uber's platform into believing that they are serving real customers, when they are in fact sitting at home. In an extreme case, according to Chinese state media, someone can own no car but still make up to 10,000 yuan ($1,527) per month as an Uber driver.

A search on Taobao for Uber click-farming returns some positive results.

The risks that come with Uber's driver rewards create a vicious cycle for a company that has struggled to gain a foothold in China since day one. Didi, which launched an app in 2012, claims to own a 87 percent share of China's private car-hailing market; Uber claims to have 30 to 35 percent. The reality is likely somewhere in between, but Didi is certainly ahead.

Uber is in a somewhat contradictory position here. While it is certainly not ideal to get ripped off by dishonest drivers, the underdog benefits from being able to tout an inflated number of rides per day. After all, high numbers mean popularity, which, in the virtual-capital market equates to investor attention.

A steep daily ride count could help Uber bring in more solid corporate revenue at a time when the company desperately needs it to compete. Recent reports suggest that Uber is burning $1 billion a year to reach only one million rides per day, compared to Didi's 11 million rides per days. Meanwhile, earlier this month, Apple announced a $1 billion investment in Didi.

New algorithm, new scam?

I asked a trainee working in Didi's marketing department to talk more specifically about the company's strategies for dealing with click fraud. He told me that Didi's operating platform has gradually shifted algorithms.

Whereas the company previously connected drivers and customers based on how geographically close they were, Didi is now migrating to an algorithm that mostly depends on the rating history of the driver. This means top-rated drivers will likely be exposed to more potential riders in a given area than low-rated drivers.

A ratings-oriented algorithm makes click fraud more difficult, because it means that having a nurse book a pick-up right near a patient's location doesn't necessarily guarantee that they will be matched.

However, a system based on ratings is still susceptible to its own scamming. Devious customers may use positive ratings as leverage, for instance, to demand discounts from drivers.

The ride-hailing industry in China is situated within a larger technology landscape that has displaced traditional jobs and made it increasingly difficult for people to make a living.

One Chinese Uber driver I spoke to over Wechat, a Chinese messaging app, told me that he had a customer start bargaining with him for a lower fare and coupons in exchange for a full-star review the minute the passenger came into the car.

A rating system could be a real nightmare, Zhang agreed, "especially when you meet some customers that have horrible attitudes." He recalled once politely asking a customer to move some luggage that was blocking his sight. "Out of nowhere, he just got really angry and threatened to give me a one-star review," said Zhang.

Struggling to see the future

My conversation with Mr. Zhang ended on a depressing note. After grappling to make ends meet for a year and a half, he just put in notice to quit. He couldn't see a future driving for Didi anymore. Many Didi drivers are quitting their jobs, he told me, because they are not making enough money.

"It's hard being a driver nowadays," Zhang said. "Didi does not have [a] generous driver bonus policy, and for each booking service we provide, we only get 70 percent out of the total fare."

These low wages feed into the problem itself. The ride-hailing industry in China is situated within a larger technology landscape that has displaced traditional jobs and made it increasingly difficult for people, such as drivers, to make a living. Rather than merely blaming cheating drivers, we might consider how a poor working environment in China victimizes this demographic in the first place.

Overall, the Chinese ride-hailing system seems trapped in a whack-a-mole situation, where fighting one type of corruption has the potential to give rise to another. Click fraud, hacked phones, ratings bargaining—perhaps, in the end, it's not helpful to take each scam as it comes. Perhaps the better question to ask is whether, so long as drivers can't make enough money, ride-hailing in China is doomed to be manipulated, one way or another.

Steph Yin contributed writing and editing.

Uber Earth is Motherboard's exploration of the ways Uber has already changed the world and how it stands to do so in the future. Follow along here.