This article was taken from the June 2015 issue of WIRED magazine. Be the first to read WIRED's articles in print before they're posted online, and get your hands on loads of additional content by subscribing online.
Here is a confession: occasionally, I drop a wet towel on a hotel bathroom floor and leave it there. But when I stay in a place I have rented on Airbnb, I behave quite differently. The last day of my reservation is set aside for cleaning because I know that not only am I going to rate the host, they will rate how "good" I am as a guest. The tables have turned. Its customers can easily rate companies, but companies can also grade customers.
This is a new phenomenon: person-to-person-driven companies such as Airbnb, TaskRabbit and Uber are starting to gather swathes of data on whether you are late or punctual, rude or a darling, dirty or clean. In other words, we're being judged. This two-way street is rife with opportunities and risks. There are very few rules on how people are graded or standards around transparency. The bigger question, however, is who owns this data. For instance, on Uber, users don't know their grade. I became aware of the impact of a lowered score after I lent my smartphone to a close family member for a week. When my phone was returned, my requests for lifts were frequently rejected; I had been weeded out of the system.
On a recent Reddit blog an Uber driver asked: "What makes you give a passenger a bad rating?" Responses include: "When the passenger slams the door"; "Stupidity"; "Attitude"; "Vomit"; "Smelling of cologne"; and "Doing coke in the back of my car". It's the range and subjectivity (coke and vomit aside) of responses that is informative yet worrying. A driver called uberdrivebos drew up a 20-point plus or minus checklist to try to remove arbitrary ratings. You lose a point if you make the driver wait for more than three minutes, for example, but gain one if you apologise.
In January 2015, Georgios Zervas, Davide Proserpio and John Byers of Boston University wrote "A First Look at Online Reputation on Airbnb, Where Every Stay is Above Average." Part of the study compared more than 2,000 properties cross-listed on both Airbnb, with a two-way rating system, and TripAdvisor, which only allows guests to rate properties. Did they rate the same? The study revealed that the number of cross-listed properties rated 4.5 stars or above is 14 per cent higher on Airbnb than on TripAdvisor.
The team concluded that the two-way system incentivised generous reporting of good experiences and underreporting of bad ones. A reviewing bias happens when we are wary of being too critical in case it affects our future ability to transact. But perhaps there is a different interpretation: hosts and their guests do in fact behave differently on Airbnb and TripAdvisor. Maybe guests pick up their towels and maybe the hosts put in more effort, making the overall experience truly better for both sides.
When we better understand the science of two-way peer-review systems, it can inject much needed accountability and respect into the way people treat one another. But both sides need standards: we have the right to know our score, why it was given, how we improve our rating and how this information will be used.
This article was originally published by WIRED UK
