Tuesday, November 1, 2016

Getting Taken For A Ride

The problem with the "gig economy" where we all magically get to choose our own customers and work schedules and then use the power of technology to leverage that is that 1) bad choices are still made by people and 2) the tech has no ability to fix those bad choices.  Take ride-sharing apps, for example.

The National Bureau of Economic Research, a respected non-profit and non-partisan research organization, has released the findings of a two-year study that tracked discrimination of riders using Uber, Lyft, and Flywheel in Seattle and Boston. The study was done by researchers at MIT, Stanford and the University of Washington.

The study involved nearly 1,500 rides across the two cities, with work beginning in Seattle late last year to this March. Undergrads from the University of Washington were given identical phones with the three ride-sharing apps pre-loaded, instructed to take a handful of prescribed routes, and then noting when the ride was requested, when it was accepted by the driver, when they were picked up, and finally when they got to their destination.

In the Seattle experiment, trip requests from black riders took between 16 to 28 percent longer to be accepted by both UberX and Lyft, and breaking UberX out showed a wait time of 29 to 35 percent longer than their white counterparts
Those figures are based on UberX usage, primarily because of the different ways a new ride is displayed to the driver through the Uber or Lyft app. 
For Uber, drivers don’t see the name of the person they’re picking up until they accept the fare, at which point they can cancel. But for Lyft, which displays the rider’s name and picture (if they included it) before they accept the fare, means trying to quantize discriminatory practices through Lyft is largely impossible—a model Uber could conceivably adopt. 
For the Boston experiment, the team turned to a previous study by NBER that sent identical resumes to potential employers, with one set having what’s described as an “African America-sounding” first name and another with a “white-sounding” first name. The results showed that callbacks for the former happened about one-third less.
The researchers for the ride-sharing study took a similar tack, using names from the 2003 survey to set up two different Uber and Lyft accounts for each rider—one with an “African American-sounding” name and another with a “white-sounding” name. The team also made another change from its Seattle methodology; rather than picking an equal number of black and white and male and female riders, they recruited students with a range of ethnicities “whose appearance allowed them to plausibly travel as a passenger of either race,” the study states.

For the men involved in the study, those that used the profile and appeared to be black riders had a cancellation rate more than twice as high as a profile of a rider who appeared to be white (11.2 percent vs. 4.5 percent). 
Women didn’t fare much better, with a cancellation rate of 8.4 percent when using the African American-sounding name and 5.4 percent when using the white-sounding first name. 
Worse, in low population density areas where getting a ride can be challenging, users with African American-sounding names were cancelled at a rate of 15.7 percent—triple that of white males.

And frankly, until Uber gets the bejesus sued out of it, they're not going to do a damn thing.  Yes, I know I pick on Uber a lot and that systemic racism in America was a problem long before Silicon Valley ever got the idea to let people turn their cars into cabs for cash, but the problem with Uber and a lot of other social tech is that the "freedom to choose your customers" and the "selectivity aspect" of who you want to deal with in a business relationship starts to look a whole lot like a pattern of systemic racism and sexism when put out in the real world.

Worse, it's looking more and more like that setup is one of the main selling points as working as a contractor for the gig economy.  It's not a bug, kids,

It's a feature.  And they don't want to "fix" it.

No comments:

Related Posts with Thumbnails