Uber on the Kalamazoo shooter: “Overall his rating was good.”

Jason Dalton.
Jason Dalton.
Image: Reuters / handout
We may earn a commission from links on this page.

Uber held a conference call Monday afternoon to discuss a horrific shooting carried out by one of its drivers in Kalamazoo, Michigan, over the weekend. Jason Brian Dalton killed six people and critically injured two, apparently targeting people at random over the course of seven hours.

On the call, Uber revealed some new details about Dalton’s record as a driver. He cleared his background check and was approved to be a driver on Jan. 25. Since then, he had completed slightly more than 100 rides.

“Overall, his rating was good,” said Joe Sullivan, Uber’s chief security officer. “Before the events it was a 4.73.” He added that Dalton “generally speaking had received very favorable feedback from riders … there were no red flags if you will.”

But on Uber, a rating of 4.73 isn’t good. In fact, it’s a number that Uber has previously defined as “below average, needs to improve,” and close to the threshold at which drivers risk deactivation.

It’s no secret that Uber’s rating system is broken. While the scale ostensibly runs from one to five stars, there’s a tacit understanding that unless something horrible happens during your Uber ride, you will leave five stars. Anecdotally, passengers who have questionable trips leave their driver with four stars at worst. Uber’s system is essentially bifurcated: There’s five stars, and then there’s everything else.

Inflated ratings might be disingenuous, but it takes an event like the weekend’s to see why they can also be outright dangerous. While Dalton’s case is an extreme one, you have to wonder how many other subpar Uber drivers aren’t being flagged in the ratings system precisely because of that unspoken code: unless all hell breaks loose, leave five stars. That’s a problem when Uber is leaning on its ratings as a safety mechanism.

“The rating system and our reporting mechanisms in the app are important signals for us, and we objectively use that information to remove a driver if the person is repeatedly performing badly,” Sullivan said on the conference call.

Dalton might not have been “repeatedly performing badly.” But if his rating was 4.73, then by Uber’s own standards he was not “good.”