The recalled Jeep shifter that may have been involved in Star Trek actor Anton Yelchin’s death is a straightforward example of how things get harder to use when you take controls out of hardware and put them into software. It’s a UI problem, and an entirely avoidable one.
First things first: if you have a 2014 or 2015 Jeep Grand Cherokee or a 2012-2014 Dodge Charger or Chrysler 300, you should call your dealer right now and set up an appointment for the recall. According to my local dealer, the update takes 3.5 hours, and it patches the car’s software to engage the emergency brake if the driver’s door is opened when the car is in neutral. That’s it. It’s a software update that was finally accelerated in the past few weeks after a death, even though problems with this shifter were so widespread that it made local TV news stories last year, the National Highway Transportation Administration issued a harsh report in February, and FCA recalled the shifter in April. The timeline is damning, but at least the update is here now. Go get it.
Onto the underlying issue: like most modern automatic shifters, the Jeep’s shift lever itself doesn’t mechanically control the transmission, even though it looks and moves like a traditional shift lever. Instead, it’s fundamentally a software switch that controls the transmission electronically. The problem is that the “Monostable” design doesn’t provide any meaningful feedback about what gear you’re in — it returns to the center position after each shift. To completely confirm if you’re in drive or park or reverse, you have to look at either the LEDs on the shifter (often covered by your palm) or the digital display in the instrument cluster. This has confused thousands of people, led to over a hundred injuries, and now potentially a death. And it’s all because of a design that prioritizes screens over switches.
Here’s what the NHTSA said about the shifter in February:
… [T]he Monostable shifter is not intuitive and provides poor tactile and visual feedback to the driver, increasing the potential for unintended gear selection.
ODI’s analysis… identified 306 incidents of vehicle rollaway following intended shifts to Park in the 2014-2015 Grand Cherokee. These resulted in 117 alleged crashes. Twenty-eight of the crashes reportedly caused injuries, including 3 with a fractured pelvis and 4 others requiring some degree of hospitalization (a ruptured bladder, fractured kneecap, broken ribs, damaged to right leg). Other injuries include reports of a broken nose, facial lacerations requiring stitches, sprained knees, severe bruising, and trauma to legs.
We made the quick video above in my car to show off how the shifter works — there are little notches as you push the shifter in either direction to indicate what gear you’re in, but they’re fairly subtle and easy to miss if you’re not paying attention. It’s actually really easy to land in reverse when you’re trying to put the thing in park; if you read Jeep forums the users there have settled on “just hold it all the way forward for an extra second” as their unofficial strategy to make sure it’s in park.
electronic gearshifts are a design playground across the car industry
This reminds me more of kids making up moves in their favorite video games than people discussing how to safely make sure their 4,500-pound SUV doesn’t roll away when they’re not looking, but that’s what happens when you replace familiar controls with systems that are more like joysticks than gearshifts without thinking the implications through.
It doesn’t have to be this way: electronic shifters can be great, and they’re a design playground across the car industry. If you don’t have to physically link the gear selector to the transmission, you don’t have to have a shifter at all: you can have a dial, like on Chrysler’s own 200. You can have push buttons, like on the Acura TLX. You can build a weird spaceship pod, like automatic BMWs have. I love the weird BMW spaceship shifter! It’s everything I’ve ever wanted from the future of car design.
All of those designs are radically different, and they’re all better than the Jeep shifter, because they unambiguously link the results of user input to mechanical results. Turn the knob all the way to the left or push the P button, and you’re in park. On the Jeep, a forward press of the shifter can land you in reverse or neutral or park, and it all feels like the same physical action unless you’re paying slightly more attention than you’re used to. That’s bad.
there is tremendous value in physical controls that always do what users understand
Here’s the thing though: we are surrounded every day by more and more things that demand we check screens to figure out how they’re working instead of examining physical states. There are tons of reasons why this is generally good — being able to remap switches to different software controls adds flexibility and power to all kinds of devices, and the consequences of user error usually aren’t deadly. No one ever filed a safety incident report because they tried to mute their iPad and locked the screen rotation instead.
But there is also tremendous value in physical controls that always do what users understand them to do just by looking at them, and provide tactile feedback about their state. The Jeep debacle is what happens when interface designers assume what happens on a screen is as meaningful as what happens in real life. Turns out that’s a huge mistake.